{ "release_name": "dctabgan-ids-benchmark-datasets", "release_created_utc": "2026-07-05T13:11:41+00:00", "description": "Public-safe subset of the full 30-task cleaned, preserved-ratio, benchmark-ready IDS surface: 17 downloadable CSVs are redistributed and 13 tasks are metadata-only omissions for local reproduction.", "active_benchmark_definition": "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "inherited_dataset_surface_definition": "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json", "benchmark_collection": "benchmark_dataset_collection_v11_preserved_ratio_single_label_nodos_30d_clean_dct_gan_dose_matched_train500_test500", "benchmark_version": "benchmark_preserved_ratio_single_label_train500_test500_2026-05-20_scope30_nodos_clean_dct_gan_dose_matched_operational", "inherited_benchmark_version": "benchmark_preserved_ratio_single_label_train500_test500_2026-05-11_scope30_nodos_20method_linearsvm", "dataset_count": 30, "source_corpus_counts": { "5G-NIDD": 3, "CIC-IDS-2017": 4, "CIC-UNSW-NB15": 5, "CICIoMT2024Small mirror": 2, "CSE-CIC-IDS2018": 4, "Edge-IIoTset": 6, "HIKARI-2021": 2, "RT-IoT2022": 4 }, "label_semantics": { "0": "majority benign/normal class", "1": "minority attack class" }, "split_policy": "Classwise temporal train/test split with test_size=0.5 and val_size=0.0, as specified by the benchmark definition. Downloadable CSVs are the full cleaned selected rows; omitted tasks retain the same count/split metadata and exact benchmark source paths for local reproduction. Benchmark runners reconstruct train/test splits from the definition and row order.", "datasets": [ { "ordinal": 1, "task_key": "friday_bot", "title": "CIC-IDS-2017 Friday Bot vs BENIGN", "source_corpus": "CIC-IDS-2017", "source_slice": "retained_non_dos", "positive_class": "Bot", "negative_class": "BENIGN", "processed_csv": "data/cic-ids-2017/friday_bot/friday_bot_vs_benign_preservedratio_train500minority_48000benign_test500minority_48000benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/friday_bot_vs_benign_preservedratio_train500minority_48000benign_test500minority_48000benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "friday_bot_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 48000, "minority": 500 }, "test_counts": { "majority": 48000, "minority": 500 }, "total_counts": { "majority": 96000, "minority": 1000, "rows": 97000 }, "actual_csv_label_counts": { "0": 96000, "1": 1000 }, "ratio_floor": 96, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Friday-WorkingHours-Morning.pcap_ISCX.csv.", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/friday_bot_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cic-ids-2017/friday_bot/friday_bot_vs_benign_preservedratio_train500minority_48000benign_test500minority_48000benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: CIC policy pages appear to allow redistribution with citation, but this release does not relicense the upstream corpus and derivative-subset rights are not expressed as a standard open-data license.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/friday_bot_vs_benign_preservedratio_train500minority_48000benign_test500minority_48000benign_chrono.csv", "local_data_path_if_rebuilt": "data/cic-ids-2017/friday_bot/friday_bot_vs_benign_preservedratio_train500minority_48000benign_test500minority_48000benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/friday_bot_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 2, "task_key": "thursday_web_attack_bruteforce", "title": "CIC-IDS-2017 Thursday Web Attack - Brute Force vs BENIGN", "source_corpus": "CIC-IDS-2017", "source_slice": "retained_non_dos", "positive_class": "Web Attack - Brute Force", "negative_class": "BENIGN", "processed_csv": "data/cic-ids-2017/thursday_web_attack_bruteforce/thursday_web_attack_bruteforce_vs_benign_preservedratio_train500minority_55500benign_test500minority_55500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/thursday_web_attack_bruteforce_vs_benign_preservedratio_train500minority_55500benign_test500minority_55500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "thursday_web_attack_bruteforce_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 55500, "minority": 500 }, "test_counts": { "majority": 55500, "minority": 500 }, "total_counts": { "majority": 111000, "minority": 1000, "rows": 112000 }, "actual_csv_label_counts": { "0": 111000, "1": 1000 }, "ratio_floor": 111, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Thursday-WorkingHours-Morning-WebAttacks.pcap_ISCX.csv.", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/thursday_web_attack_bruteforce_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cic-ids-2017/thursday_web_attack_bruteforce/thursday_web_attack_bruteforce_vs_benign_preservedratio_train500minority_55500benign_test500minority_55500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: CIC policy pages appear to allow redistribution with citation, but this release does not relicense the upstream corpus and derivative-subset rights are not expressed as a standard open-data license.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/thursday_web_attack_bruteforce_vs_benign_preservedratio_train500minority_55500benign_test500minority_55500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cic-ids-2017/thursday_web_attack_bruteforce/thursday_web_attack_bruteforce_vs_benign_preservedratio_train500minority_55500benign_test500minority_55500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/thursday_web_attack_bruteforce_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 3, "task_key": "tuesday_ftp_patator", "title": "CIC-IDS-2017 Tuesday FTP-Patator vs BENIGN", "source_corpus": "CIC-IDS-2017", "source_slice": "retained_non_dos", "positive_class": "FTP-Patator", "negative_class": "BENIGN", "processed_csv": "data/cic-ids-2017/tuesday_ftp_patator/tuesday_ftp_patator_vs_benign_preservedratio_train500minority_27000benign_test500minority_27000benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/tuesday_ftp_patator_vs_benign_preservedratio_train500minority_27000benign_test500minority_27000benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "tuesday_ftp_patator_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 27000, "minority": 500 }, "test_counts": { "majority": 27000, "minority": 500 }, "total_counts": { "majority": 54000, "minority": 1000, "rows": 55000 }, "actual_csv_label_counts": { "0": 54000, "1": 1000 }, "ratio_floor": 54, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Tuesday-WorkingHours.pcap_ISCX.csv.", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/tuesday_ftp_patator_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cic-ids-2017/tuesday_ftp_patator/tuesday_ftp_patator_vs_benign_preservedratio_train500minority_27000benign_test500minority_27000benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: CIC policy pages appear to allow redistribution with citation, but this release does not relicense the upstream corpus and derivative-subset rights are not expressed as a standard open-data license.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/tuesday_ftp_patator_vs_benign_preservedratio_train500minority_27000benign_test500minority_27000benign_chrono.csv", "local_data_path_if_rebuilt": "data/cic-ids-2017/tuesday_ftp_patator/tuesday_ftp_patator_vs_benign_preservedratio_train500minority_27000benign_test500minority_27000benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/tuesday_ftp_patator_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 4, "task_key": "tuesday_ssh_patator", "title": "CIC-IDS-2017 Tuesday SSH-Patator vs BENIGN", "source_corpus": "CIC-IDS-2017", "source_slice": "retained_non_dos", "positive_class": "SSH-Patator", "negative_class": "BENIGN", "processed_csv": "data/cic-ids-2017/tuesday_ssh_patator/tuesday_ssh_patator_vs_benign_preservedratio_train500minority_36500benign_test500minority_36500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/tuesday_ssh_patator_vs_benign_preservedratio_train500minority_36500benign_test500minority_36500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "tuesday_ssh_patator_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 36500, "minority": 500 }, "test_counts": { "majority": 36500, "minority": 500 }, "total_counts": { "majority": 73000, "minority": 1000, "rows": 74000 }, "actual_csv_label_counts": { "0": 73000, "1": 1000 }, "ratio_floor": 73, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Tuesday-WorkingHours.pcap_ISCX.csv.", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/tuesday_ssh_patator_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cic-ids-2017/tuesday_ssh_patator/tuesday_ssh_patator_vs_benign_preservedratio_train500minority_36500benign_test500minority_36500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: CIC policy pages appear to allow redistribution with citation, but this release does not relicense the upstream corpus and derivative-subset rights are not expressed as a standard open-data license.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/tuesday_ssh_patator_vs_benign_preservedratio_train500minority_36500benign_test500minority_36500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cic-ids-2017/tuesday_ssh_patator/tuesday_ssh_patator_vs_benign_preservedratio_train500minority_36500benign_test500minority_36500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/tuesday_ssh_patator_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 5, "task_key": "cse_cic_ids2018_bot", "title": "CSE-CIC-IDS2018 Bot vs BENIGN", "source_corpus": "CSE-CIC-IDS2018", "source_slice": "retained_non_dos", "positive_class": "N, o, r, m, a, l, i, z, e, d, , l, a, b, e, l, , e, q, u, a, l, s, , ', b, o, t, ', .", "negative_class": "BENIGN", "processed_csv": "data/cse-cic-ids2018/cse_cic_ids2018_bot/cse_cic_ids2018_bot_vs_benign_preservedratio_train500minority_23500benign_test500minority_23500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_bot_vs_benign_preservedratio_train500minority_23500benign_test500minority_23500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "cse_cic_ids2018_bot_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 23500, "minority": 500 }, "test_counts": { "majority": 23500, "minority": 500 }, "total_counts": { "majority": 47000, "minority": 1000, "rows": 48000 }, "actual_csv_label_counts": { "0": 47000, "1": 1000 }, "ratio_floor": 47, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: All day-organized processed-flow CSVs in file order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_bot_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cse-cic-ids2018/cse_cic_ids2018_bot/cse_cic_ids2018_bot_vs_benign_preservedratio_train500minority_23500benign_test500minority_23500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: UNB/CIC redistribution language appears broad, but this release does not relicense the upstream corpus.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_bot_vs_benign_preservedratio_train500minority_23500benign_test500minority_23500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cse-cic-ids2018/cse_cic_ids2018_bot/cse_cic_ids2018_bot_vs_benign_preservedratio_train500minority_23500benign_test500minority_23500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_bot_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 6, "task_key": "cse_cic_ids2018_ftp_bruteforce", "title": "CSE-CIC-IDS2018 FTP-BruteForce vs BENIGN", "source_corpus": "CSE-CIC-IDS2018", "source_slice": "retained_non_dos", "positive_class": "N, o, r, m, a, l, i, z, e, d, , l, a, b, e, l, , e, q, u, a, l, s, , ', f, t, p, , b, r, u, t, e, f, o, r, c, e, ', .", "negative_class": "BENIGN", "processed_csv": "data/cse-cic-ids2018/cse_cic_ids2018_ftp_bruteforce/cse_cic_ids2018_ftp_bruteforce_vs_benign_preservedratio_train500minority_34500benign_test500minority_34500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_ftp_bruteforce_vs_benign_preservedratio_train500minority_34500benign_test500minority_34500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "cse_cic_ids2018_ftp_bruteforce_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 34500, "minority": 500 }, "test_counts": { "majority": 34500, "minority": 500 }, "total_counts": { "majority": 69000, "minority": 1000, "rows": 70000 }, "actual_csv_label_counts": { "0": 69000, "1": 1000 }, "ratio_floor": 69, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: All day-organized processed-flow CSVs in file order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_ftp_bruteforce_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cse-cic-ids2018/cse_cic_ids2018_ftp_bruteforce/cse_cic_ids2018_ftp_bruteforce_vs_benign_preservedratio_train500minority_34500benign_test500minority_34500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: UNB/CIC redistribution language appears broad, but this release does not relicense the upstream corpus.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_ftp_bruteforce_vs_benign_preservedratio_train500minority_34500benign_test500minority_34500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cse-cic-ids2018/cse_cic_ids2018_ftp_bruteforce/cse_cic_ids2018_ftp_bruteforce_vs_benign_preservedratio_train500minority_34500benign_test500minority_34500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_ftp_bruteforce_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 7, "task_key": "cse_cic_ids2018_infilteration", "title": "CSE-CIC-IDS2018 Infilteration vs BENIGN", "source_corpus": "CSE-CIC-IDS2018", "source_slice": "retained_non_dos", "positive_class": "N, o, r, m, a, l, i, z, e, d, , l, a, b, e, l, , e, q, u, a, l, s, , ', i, n, f, i, l, t, e, r, a, t, i, o, n, ', .", "negative_class": "BENIGN", "processed_csv": "data/cse-cic-ids2018/cse_cic_ids2018_infilteration/cse_cic_ids2018_infilteration_vs_benign_preservedratio_train500minority_41500benign_test500minority_41500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_infilteration_vs_benign_preservedratio_train500minority_41500benign_test500minority_41500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "cse_cic_ids2018_infilteration_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 41500, "minority": 500 }, "test_counts": { "majority": 41500, "minority": 500 }, "total_counts": { "majority": 83000, "minority": 1000, "rows": 84000 }, "actual_csv_label_counts": { "0": 83000, "1": 1000 }, "ratio_floor": 83, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: All day-organized processed-flow CSVs in file order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_infilteration_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cse-cic-ids2018/cse_cic_ids2018_infilteration/cse_cic_ids2018_infilteration_vs_benign_preservedratio_train500minority_41500benign_test500minority_41500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: UNB/CIC redistribution language appears broad, but this release does not relicense the upstream corpus.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_infilteration_vs_benign_preservedratio_train500minority_41500benign_test500minority_41500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cse-cic-ids2018/cse_cic_ids2018_infilteration/cse_cic_ids2018_infilteration_vs_benign_preservedratio_train500minority_41500benign_test500minority_41500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_infilteration_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 8, "task_key": "cse_cic_ids2018_ssh_bruteforce", "title": "CSE-CIC-IDS2018 SSH-Bruteforce vs BENIGN", "source_corpus": "CSE-CIC-IDS2018", "source_slice": "retained_non_dos", "positive_class": "N, o, r, m, a, l, i, z, e, d, , l, a, b, e, l, , e, q, u, a, l, s, , ', s, s, h, , b, r, u, t, e, f, o, r, c, e, ', .", "negative_class": "BENIGN", "processed_csv": "data/cse-cic-ids2018/cse_cic_ids2018_ssh_bruteforce/cse_cic_ids2018_ssh_bruteforce_vs_benign_preservedratio_train500minority_35500benign_test500minority_35500benign_chrono.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_ssh_bruteforce_vs_benign_preservedratio_train500minority_35500benign_test500minority_35500benign_chrono.csv", "benchmark_subset": "train500_test500_single_label_25d_chrono", "benchmark_code_prefix": "cse_cic_ids2018_ssh_bruteforce_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 35500, "minority": 500 }, "test_counts": { "majority": 35500, "minority": 500 }, "total_counts": { "majority": 71000, "minority": 1000, "rows": 72000 }, "actual_csv_label_counts": { "0": 71000, "1": 1000 }, "ratio_floor": 71, "feature_count": 78, "column_count": 80, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: All day-organized processed-flow CSVs in file order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_ssh_bruteforce_vs_benign.json", "task_manifest_files": [], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/cse-cic-ids2018/cse_cic_ids2018_ssh_bruteforce/cse_cic_ids2018_ssh_bruteforce_vs_benign_preservedratio_train500minority_35500benign_test500minority_35500benign_chrono.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with conservative CIC provenance: UNB/CIC redistribution language appears broad, but this release does not relicense the upstream corpus.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_chrono/cse_cic_ids2018_ssh_bruteforce_vs_benign_preservedratio_train500minority_35500benign_test500minority_35500benign_chrono.csv", "local_data_path_if_rebuilt": "data/cse-cic-ids2018/cse_cic_ids2018_ssh_bruteforce/cse_cic_ids2018_ssh_bruteforce_vs_benign_preservedratio_train500minority_35500benign_test500minority_35500benign_chrono.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cse_cic_ids2018_ssh_bruteforce_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 9, "task_key": "cic_unsw_nb15_exploits", "title": "CIC-UNSW-NB15 Exploits vs BENIGN", "source_corpus": "CIC-UNSW-NB15", "source_slice": "retained_non_dos", "positive_class": "Exploits", "negative_class": "BENIGN", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_exploits_vs_benign_preservedratio_train500minority_5500benign_test500minority_5500benign_fixed.csv", "benchmark_subset": "train500_test500_single_label_25d_fixed", "benchmark_code_prefix": "cic_unsw_nb15_exploits_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 5500, "minority": 500 }, "test_counts": { "majority": 5500, "minority": 500 }, "total_counts": { "majority": 11000, "minority": 1000, "rows": 12000 }, "actual_csv_label_counts": { "0": 11000, "1": 1000 }, "ratio_floor": 11, "feature_count": 76, "column_count": 78, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Data.csv joined row-wise with Label.csv in original row order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_exploits_vs_benign.json", "task_manifest_files": [], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "omission_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "redistribution_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_exploits_vs_benign_preservedratio_train500minority_5500benign_test500minority_5500benign_fixed.csv", "local_data_path_if_rebuilt": "data/cic-unsw-nb15/cic_unsw_nb15_exploits/cic_unsw_nb15_exploits_vs_benign_preservedratio_train500minority_5500benign_test500minority_5500benign_fixed.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_exploits_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 10, "task_key": "cic_unsw_nb15_fuzzers", "title": "CIC-UNSW-NB15 Fuzzers vs BENIGN", "source_corpus": "CIC-UNSW-NB15", "source_slice": "retained_non_dos", "positive_class": "Fuzzers", "negative_class": "BENIGN", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_fuzzers_vs_benign_preservedratio_train500minority_6000benign_test500minority_6000benign_fixed.csv", "benchmark_subset": "train500_test500_single_label_25d_fixed", "benchmark_code_prefix": "cic_unsw_nb15_fuzzers_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 6000, "minority": 500 }, "test_counts": { "majority": 6000, "minority": 500 }, "total_counts": { "majority": 12000, "minority": 1000, "rows": 13000 }, "actual_csv_label_counts": { "0": 12000, "1": 1000 }, "ratio_floor": 12, "feature_count": 76, "column_count": 78, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Data.csv joined row-wise with Label.csv in original row order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_fuzzers_vs_benign.json", "task_manifest_files": [], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "omission_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "redistribution_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_fuzzers_vs_benign_preservedratio_train500minority_6000benign_test500minority_6000benign_fixed.csv", "local_data_path_if_rebuilt": "data/cic-unsw-nb15/cic_unsw_nb15_fuzzers/cic_unsw_nb15_fuzzers_vs_benign_preservedratio_train500minority_6000benign_test500minority_6000benign_fixed.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_fuzzers_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 11, "task_key": "cic_unsw_nb15_generic", "title": "CIC-UNSW-NB15 Generic vs BENIGN", "source_corpus": "CIC-UNSW-NB15", "source_slice": "retained_non_dos", "positive_class": "Generic", "negative_class": "BENIGN", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_generic_vs_benign_preservedratio_train500minority_38500benign_test500minority_38500benign_fixed.csv", "benchmark_subset": "train500_test500_single_label_25d_fixed", "benchmark_code_prefix": "cic_unsw_nb15_generic_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 38500, "minority": 500 }, "test_counts": { "majority": 38500, "minority": 500 }, "total_counts": { "majority": 77000, "minority": 1000, "rows": 78000 }, "actual_csv_label_counts": { "0": 77000, "1": 1000 }, "ratio_floor": 77, "feature_count": 76, "column_count": 78, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Data.csv joined row-wise with Label.csv in original row order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_generic_vs_benign.json", "task_manifest_files": [], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "omission_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "redistribution_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_generic_vs_benign_preservedratio_train500minority_38500benign_test500minority_38500benign_fixed.csv", "local_data_path_if_rebuilt": "data/cic-unsw-nb15/cic_unsw_nb15_generic/cic_unsw_nb15_generic_vs_benign_preservedratio_train500minority_38500benign_test500minority_38500benign_fixed.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_generic_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 12, "task_key": "cic_unsw_nb15_reconnaissance", "title": "CIC-UNSW-NB15 Reconnaissance vs BENIGN", "source_corpus": "CIC-UNSW-NB15", "source_slice": "retained_non_dos", "positive_class": "Reconnaissance", "negative_class": "BENIGN", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_reconnaissance_vs_benign_preservedratio_train500minority_10500benign_test500minority_10500benign_fixed.csv", "benchmark_subset": "train500_test500_single_label_25d_fixed", "benchmark_code_prefix": "cic_unsw_nb15_reconnaissance_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 10500, "minority": 500 }, "test_counts": { "majority": 10500, "minority": 500 }, "total_counts": { "majority": 21000, "minority": 1000, "rows": 22000 }, "actual_csv_label_counts": { "0": 21000, "1": 1000 }, "ratio_floor": 21, "feature_count": 76, "column_count": 78, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Data.csv joined row-wise with Label.csv in original row order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_reconnaissance_vs_benign.json", "task_manifest_files": [], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "omission_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "redistribution_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_reconnaissance_vs_benign_preservedratio_train500minority_10500benign_test500minority_10500benign_fixed.csv", "local_data_path_if_rebuilt": "data/cic-unsw-nb15/cic_unsw_nb15_reconnaissance/cic_unsw_nb15_reconnaissance_vs_benign_preservedratio_train500minority_10500benign_test500minority_10500benign_fixed.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_reconnaissance_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 13, "task_key": "cic_unsw_nb15_shellcode", "title": "CIC-UNSW-NB15 Shellcode vs BENIGN", "source_corpus": "CIC-UNSW-NB15", "source_slice": "retained_non_dos", "positive_class": "Shellcode", "negative_class": "BENIGN", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_shellcode_vs_benign_preservedratio_train500minority_85000benign_test500minority_85000benign_fixed.csv", "benchmark_subset": "train500_test500_single_label_25d_fixed", "benchmark_code_prefix": "cic_unsw_nb15_shellcode_single25_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 85000, "minority": 500 }, "test_counts": { "majority": 85000, "minority": 500 }, "total_counts": { "majority": 170000, "minority": 1000, "rows": 171000 }, "actual_csv_label_counts": { "0": 170000, "1": 1000 }, "ratio_floor": 170, "feature_count": 76, "column_count": 78, "train_test_source_row_index_overlap": null, "dropped_columns": [], "provenance_note": "Retained non-DoS task from the preserved-ratio single-label 25D benchmark surface. Source description: Data.csv joined row-wise with Label.csv in original row order..", "risk_note": "", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_shellcode_vs_benign.json", "task_manifest_files": [], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "omission_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "redistribution_caveat": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_single_label_25d/tuning_fixed/cic_unsw_nb15_shellcode_vs_benign_preservedratio_train500minority_85000benign_test500minority_85000benign_fixed.csv", "local_data_path_if_rebuilt": "data/cic-unsw-nb15/cic_unsw_nb15_shellcode/cic_unsw_nb15_shellcode_vs_benign_preservedratio_train500minority_85000benign_test500minority_85000benign_fixed.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_preserved_ratio_train500_test500_single_label_benchmark.py", "source_manifest": "metadata/source_manifests/retained_single_label_25d/cic_unsw_nb15_shellcode_vs_benign.json", "task_manifest_files": [] } }, { "ordinal": 14, "task_key": "hikari_bruteforce_vs_benign", "title": "HIKARI-2021 Bruteforce vs BENIGN", "source_corpus": "HIKARI-2021", "source_slice": "modern7", "positive_class": "Bruteforce", "negative_class": "benign_control", "processed_csv": "data/hikari-2021/hikari_bruteforce_vs_benign/hikari_bruteforce_vs_benign_preservedratio_train500minority_43500majority_test500minority_43500majority_modern7.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/hikari_bruteforce_vs_benign/hikari_bruteforce_vs_benign_preservedratio_train500minority_43500majority_test500minority_43500majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "hikari_bruteforce_vs_benign_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 43500, "minority": 500 }, "test_counts": { "majority": 43500, "minority": 500 }, "total_counts": { "majority": 87000, "minority": 1000, "rows": 88000 }, "actual_csv_label_counts": { "0": 87000, "1": 1000 }, "ratio_floor": 87, "feature_count": 81, "column_count": 83, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/hikari_bruteforce_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/hikari_bruteforce_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/hikari-2021/hikari_bruteforce_vs_benign/hikari_bruteforce_vs_benign_preservedratio_train500minority_43500majority_test500minority_43500majority_modern7.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented CC BY 4.0 source route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/hikari_bruteforce_vs_benign/hikari_bruteforce_vs_benign_preservedratio_train500minority_43500majority_test500minority_43500majority_modern7.csv", "local_data_path_if_rebuilt": "data/hikari-2021/hikari_bruteforce_vs_benign/hikari_bruteforce_vs_benign_preservedratio_train500minority_43500majority_test500minority_43500majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/hikari_bruteforce_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/hikari_bruteforce_vs_benign/split_metadata.json" ] } }, { "ordinal": 15, "task_key": "hikari_probing_vs_benign", "title": "HIKARI-2021 Probing vs BENIGN", "source_corpus": "HIKARI-2021", "source_slice": "modern7", "positive_class": "Probing", "negative_class": "benign_control", "processed_csv": "data/hikari-2021/hikari_probing_vs_benign/hikari_probing_vs_benign_preservedratio_train500minority_11000majority_test500minority_11000majority_modern7.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/hikari_probing_vs_benign/hikari_probing_vs_benign_preservedratio_train500minority_11000majority_test500minority_11000majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "hikari_probing_vs_benign_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 11000, "minority": 500 }, "test_counts": { "majority": 11000, "minority": 500 }, "total_counts": { "majority": 22000, "minority": 1000, "rows": 23000 }, "actual_csv_label_counts": { "0": 22000, "1": 1000 }, "ratio_floor": 22, "feature_count": 81, "column_count": 83, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/hikari_probing_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/hikari_probing_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/hikari-2021/hikari_probing_vs_benign/hikari_probing_vs_benign_preservedratio_train500minority_11000majority_test500minority_11000majority_modern7.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented CC BY 4.0 source route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/hikari_probing_vs_benign/hikari_probing_vs_benign_preservedratio_train500minority_11000majority_test500minority_11000majority_modern7.csv", "local_data_path_if_rebuilt": "data/hikari-2021/hikari_probing_vs_benign/hikari_probing_vs_benign_preservedratio_train500minority_11000majority_test500minority_11000majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/hikari_probing_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/hikari_probing_vs_benign/split_metadata.json" ] } }, { "ordinal": 16, "task_key": "ciciomt2024_arp_spoofing_vs_benign", "title": "CICIoMT2024Small mirror ARP Spoofing vs BENIGN", "source_corpus": "CICIoMT2024Small mirror", "source_slice": "modern7", "positive_class": "ARP Spoofing", "negative_class": "Benign", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/ciciomt2024_arp_spoofing_vs_benign/ciciomt2024_arp_spoofing_vs_benign_preservedratio_train500minority_12500majority_test500minority_12500majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "ciciomt2024_arp_spoofing_vs_benign_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 12500, "minority": 500 }, "test_counts": { "majority": 12500, "minority": 500 }, "total_counts": { "majority": 25000, "minority": 1000, "rows": 26000 }, "actual_csv_label_counts": { "0": 25000, "1": 1000 }, "ratio_floor": 25, "feature_count": 79, "column_count": 81, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "Materialized from public Hugging Face CICIoMT2024Small mirror; official UNB endpoint remains registration-gated.", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/split_metadata.json", "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/access_status.md" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "omission_caveat": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "redistribution_caveat": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/ciciomt2024_arp_spoofing_vs_benign/ciciomt2024_arp_spoofing_vs_benign_preservedratio_train500minority_12500majority_test500minority_12500majority_modern7.csv", "local_data_path_if_rebuilt": "data/ciciomt2024small-mirror/ciciomt2024_arp_spoofing_vs_benign/ciciomt2024_arp_spoofing_vs_benign_preservedratio_train500minority_12500majority_test500minority_12500majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/split_metadata.json", "metadata/source_manifests/modern_ids/ciciomt2024_arp_spoofing_vs_benign/access_status.md" ] } }, { "ordinal": 17, "task_key": "ciciomt2024_mqtt_malformed_data_vs_benign", "title": "CICIoMT2024Small mirror MQTT Malformed Data vs BENIGN", "source_corpus": "CICIoMT2024Small mirror", "source_slice": "modern7", "positive_class": "MQTT Malformed Data", "negative_class": "Benign", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/ciciomt2024_mqtt_malformed_data_vs_benign/ciciomt2024_mqtt_malformed_data_vs_benign_preservedratio_train500minority_8500majority_test500minority_8500majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "ciciomt2024_mqtt_malformed_data_vs_benign_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 8500, "minority": 500 }, "test_counts": { "majority": 8500, "minority": 500 }, "total_counts": { "majority": 17000, "minority": 1000, "rows": 18000 }, "actual_csv_label_counts": { "0": 17000, "1": 1000 }, "ratio_floor": 17, "feature_count": 79, "column_count": 81, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "Materialized from public Hugging Face CICIoMT2024Small mirror; official UNB endpoint remains registration-gated.", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/split_metadata.json", "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/access_status.md" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "omission_caveat": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "redistribution_caveat": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/ciciomt2024_mqtt_malformed_data_vs_benign/ciciomt2024_mqtt_malformed_data_vs_benign_preservedratio_train500minority_8500majority_test500minority_8500majority_modern7.csv", "local_data_path_if_rebuilt": "data/ciciomt2024small-mirror/ciciomt2024_mqtt_malformed_data_vs_benign/ciciomt2024_mqtt_malformed_data_vs_benign_preservedratio_train500minority_8500majority_test500minority_8500majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/split_metadata.json", "metadata/source_manifests/modern_ids/ciciomt2024_mqtt_malformed_data_vs_benign/access_status.md" ] } }, { "ordinal": 18, "task_key": "edge_iiotset_sql_injection_vs_normal", "title": "Edge-IIoTset SQL injection vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "modern7", "positive_class": "SQL_injection_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_sql_injection_vs_normal/edge_iiotset_sql_injection_vs_normal_preservedratio_train500minority_15500majority_test500minority_15500majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "edge_iiotset_sql_injection_vs_normal_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 15500, "minority": 500 }, "test_counts": { "majority": 15500, "minority": 500 }, "total_counts": { "majority": 31000, "minority": 1000, "rows": 32000 }, "actual_csv_label_counts": { "0": 31000, "1": 1000 }, "ratio_floor": 31, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_sql_injection_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_sql_injection_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_sql_injection_vs_normal/edge_iiotset_sql_injection_vs_normal_preservedratio_train500minority_15500majority_test500minority_15500majority_modern7.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_sql_injection_vs_normal/edge_iiotset_sql_injection_vs_normal_preservedratio_train500minority_15500majority_test500minority_15500majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_sql_injection_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_sql_injection_vs_normal/split_metadata.json" ] } }, { "ordinal": 19, "task_key": "edge_iiotset_password_vs_normal", "title": "Edge-IIoTset Password vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "modern7", "positive_class": "Password_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_password_vs_normal/edge_iiotset_password_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "edge_iiotset_password_vs_normal_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 16000, "minority": 500 }, "test_counts": { "majority": 16000, "minority": 500 }, "total_counts": { "majority": 32000, "minority": 1000, "rows": 33000 }, "actual_csv_label_counts": { "0": 32000, "1": 1000 }, "ratio_floor": 32, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_password_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_password_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_password_vs_normal/edge_iiotset_password_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_modern7.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_password_vs_normal/edge_iiotset_password_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_password_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_password_vs_normal/split_metadata.json" ] } }, { "ordinal": 20, "task_key": "edge_iiotset_uploading_vs_normal", "title": "Edge-IIoTset Uploading vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "modern7", "positive_class": "Uploading_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_uploading_vs_normal/edge_iiotset_uploading_vs_normal_preservedratio_train500minority_21000majority_test500minority_21000majority_modern7.csv", "benchmark_subset": "train500_test500_modern_ids_7d", "benchmark_code_prefix": "edge_iiotset_uploading_vs_normal_modern7_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 21000, "minority": 500 }, "test_counts": { "majority": 21000, "minority": 500 }, "total_counts": { "majority": 42000, "minority": 1000, "rows": 43000 }, "actual_csv_label_counts": { "0": 42000, "1": 1000 }, "ratio_floor": 42, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_uploading_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_uploading_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_7d/edge_iiotset_uploading_vs_normal/edge_iiotset_uploading_vs_normal_preservedratio_train500minority_21000majority_test500minority_21000majority_modern7.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_uploading_vs_normal/edge_iiotset_uploading_vs_normal_preservedratio_train500minority_21000majority_test500minority_21000majority_modern7.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/modern7_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_uploading_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_uploading_vs_normal/split_metadata.json" ] } }, { "ordinal": 21, "task_key": "5g_nidd_tcp_connect_scan_vs_benign", "title": "5G-NIDD TCPConnectScan vs BENIGN", "source_corpus": "5G-NIDD", "source_slice": "additional10", "positive_class": "TCPConnectScan", "negative_class": "Benign", "processed_csv": "data/5g-nidd/5g_nidd_tcp_connect_scan_vs_benign/5g_nidd_tcp_connect_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_tcp_connect_scan_vs_benign/5g_nidd_tcp_connect_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "5g_nidd_tcp_connect_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 11500, "minority": 500 }, "test_counts": { "majority": 11500, "minority": 500 }, "total_counts": { "majority": 23000, "minority": 1000, "rows": 24000 }, "actual_csv_label_counts": { "0": 23000, "1": 1000 }, "ratio_floor": 23, "feature_count": 44, "column_count": 46, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "Offset", "SrcTCPBase", "DstTCPBase" ], "provenance_note": "Official IEEE DataPort route remains gated; local artifact was sourced from the open Fairdata route.", "risk_note": "5G Offset and TCP sequence-base artifacts filtered at benchmark materialization; remaining low-card categorical handling should be monitored.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_tcp_connect_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_tcp_connect_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/5g-nidd/5g_nidd_tcp_connect_scan_vs_benign/5g_nidd_tcp_connect_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented open Fairdata CC BY 4.0 route; the official IEEE DataPort route remains gated.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_tcp_connect_scan_vs_benign/5g_nidd_tcp_connect_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "local_data_path_if_rebuilt": "data/5g-nidd/5g_nidd_tcp_connect_scan_vs_benign/5g_nidd_tcp_connect_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_tcp_connect_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_tcp_connect_scan_vs_benign/split_metadata.json" ] } }, { "ordinal": 22, "task_key": "5g_nidd_syn_scan_vs_benign", "title": "5G-NIDD SYNScan vs BENIGN", "source_corpus": "5G-NIDD", "source_slice": "additional10", "positive_class": "SYNScan", "negative_class": "Benign", "processed_csv": "data/5g-nidd/5g_nidd_syn_scan_vs_benign/5g_nidd_syn_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_syn_scan_vs_benign/5g_nidd_syn_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "5g_nidd_syn_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 11500, "minority": 500 }, "test_counts": { "majority": 11500, "minority": 500 }, "total_counts": { "majority": 23000, "minority": 1000, "rows": 24000 }, "actual_csv_label_counts": { "0": 23000, "1": 1000 }, "ratio_floor": 23, "feature_count": 44, "column_count": 46, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "Offset", "SrcTCPBase", "DstTCPBase" ], "provenance_note": "Official IEEE DataPort route remains gated; local artifact was sourced from the open Fairdata route.", "risk_note": "5G Offset and TCP sequence-base artifacts filtered at benchmark materialization; remaining low-card categorical handling should be monitored.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_syn_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_syn_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/5g-nidd/5g_nidd_syn_scan_vs_benign/5g_nidd_syn_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented open Fairdata CC BY 4.0 route; the official IEEE DataPort route remains gated.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_syn_scan_vs_benign/5g_nidd_syn_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "local_data_path_if_rebuilt": "data/5g-nidd/5g_nidd_syn_scan_vs_benign/5g_nidd_syn_scan_vs_benign_preservedratio_train500minority_11500majority_test500minority_11500majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_syn_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_syn_scan_vs_benign/split_metadata.json" ] } }, { "ordinal": 23, "task_key": "5g_nidd_udp_scan_vs_benign", "title": "5G-NIDD UDPScan vs BENIGN", "source_corpus": "5G-NIDD", "source_slice": "additional10", "positive_class": "UDPScan", "negative_class": "Benign", "processed_csv": "data/5g-nidd/5g_nidd_udp_scan_vs_benign/5g_nidd_udp_scan_vs_benign_preservedratio_train500minority_15000majority_test500minority_15000majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_udp_scan_vs_benign/5g_nidd_udp_scan_vs_benign_preservedratio_train500minority_15000majority_test500minority_15000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "5g_nidd_udp_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 15000, "minority": 500 }, "test_counts": { "majority": 15000, "minority": 500 }, "total_counts": { "majority": 30000, "minority": 1000, "rows": 31000 }, "actual_csv_label_counts": { "0": 30000, "1": 1000 }, "ratio_floor": 30, "feature_count": 44, "column_count": 46, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "Offset", "SrcTCPBase", "DstTCPBase" ], "provenance_note": "Official IEEE DataPort route remains gated; local artifact was sourced from the open Fairdata route.", "risk_note": "5G Offset and TCP sequence-base artifacts filtered at benchmark materialization; remaining low-card categorical handling should be monitored.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_udp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_udp_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/5g-nidd/5g_nidd_udp_scan_vs_benign/5g_nidd_udp_scan_vs_benign_preservedratio_train500minority_15000majority_test500minority_15000majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented open Fairdata CC BY 4.0 route; the official IEEE DataPort route remains gated.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/5g_nidd_udp_scan_vs_benign/5g_nidd_udp_scan_vs_benign_preservedratio_train500minority_15000majority_test500minority_15000majority_additional10.csv", "local_data_path_if_rebuilt": "data/5g-nidd/5g_nidd_udp_scan_vs_benign/5g_nidd_udp_scan_vs_benign_preservedratio_train500minority_15000majority_test500minority_15000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/5g_nidd_udp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/5g_nidd_udp_scan_vs_benign/split_metadata.json" ] } }, { "ordinal": 24, "task_key": "edge_iiotset_vulnerability_scanner_vs_normal", "title": "Edge-IIoTset Vulnerability scanner vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "additional10", "positive_class": "Vulnerability_scanner_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_vulnerability_scanner_vs_normal/edge_iiotset_vulnerability_scanner_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "edge_iiotset_vulnerability_scanner_vs_normal_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 16000, "minority": 500 }, "test_counts": { "majority": 16000, "minority": 500 }, "total_counts": { "majority": 32000, "minority": 1000, "rows": 33000 }, "actual_csv_label_counts": { "0": 32000, "1": 1000 }, "ratio_floor": 32, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "High-cardinality Edge timestamp/IP/payload/text/sequence artifacts filtered at benchmark materialization; remaining risk is moderate runtime and retained destination-port cardinality.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_vulnerability_scanner_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_vulnerability_scanner_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_vulnerability_scanner_vs_normal/edge_iiotset_vulnerability_scanner_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_additional10.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_vulnerability_scanner_vs_normal/edge_iiotset_vulnerability_scanner_vs_normal_preservedratio_train500minority_16000majority_test500minority_16000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_vulnerability_scanner_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_vulnerability_scanner_vs_normal/split_metadata.json" ] } }, { "ordinal": 25, "task_key": "edge_iiotset_backdoor_vs_normal", "title": "Edge-IIoTset Backdoor vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "additional10", "positive_class": "Backdoor_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_backdoor_vs_normal/edge_iiotset_backdoor_vs_normal_preservedratio_train500minority_32000majority_test500minority_32000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "edge_iiotset_backdoor_vs_normal_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 32000, "minority": 500 }, "test_counts": { "majority": 32000, "minority": 500 }, "total_counts": { "majority": 64000, "minority": 1000, "rows": 65000 }, "actual_csv_label_counts": { "0": 64000, "1": 1000 }, "ratio_floor": 64, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "High-cardinality Edge timestamp/IP/payload/text/sequence artifacts filtered at benchmark materialization; remaining risk is moderate runtime and retained destination-port cardinality.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_backdoor_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_backdoor_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_backdoor_vs_normal/edge_iiotset_backdoor_vs_normal_preservedratio_train500minority_32000majority_test500minority_32000majority_additional10.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_backdoor_vs_normal/edge_iiotset_backdoor_vs_normal_preservedratio_train500minority_32000majority_test500minority_32000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_backdoor_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_backdoor_vs_normal/split_metadata.json" ] } }, { "ordinal": 26, "task_key": "edge_iiotset_port_scanning_vs_normal", "title": "Edge-IIoTset Port Scanning vs NORMAL", "source_corpus": "Edge-IIoTset", "source_slice": "additional10", "positive_class": "Port_Scanning_attack", "negative_class": "Normal", "processed_csv": null, "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_port_scanning_vs_normal/edge_iiotset_port_scanning_vs_normal_preservedratio_train500minority_35500majority_test500minority_35500majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "edge_iiotset_port_scanning_vs_normal_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 35500, "minority": 500 }, "test_counts": { "majority": 35500, "minority": 500 }, "total_counts": { "majority": 71000, "minority": 1000, "rows": 72000 }, "actual_csv_label_counts": { "0": 71000, "1": 1000 }, "ratio_floor": 71, "feature_count": 41, "column_count": 43, "train_test_source_row_index_overlap": 0, "dropped_columns": [ "frame.time", "ip.src_host", "ip.dst_host", "arp.dst.proto_ipv4", "arp.src.proto_ipv4", "http.file_data", "http.request.uri.query", "http.referer", "http.request.full_uri", "tcp.ack", "tcp.ack_raw", "tcp.checksum", "tcp.options", "tcp.payload", "tcp.seq", "tcp.srcport", "udp.stream", "mqtt.msg", "mqtt.topic", "mbtcp.trans_id" ], "provenance_note": "Materialized from public Kaggle mirror; official IEEE DataPort remains login/subscription-gated.", "risk_note": "High-cardinality Edge timestamp/IP/payload/text/sequence artifacts filtered at benchmark materialization; remaining risk is moderate runtime and retained destination-port cardinality.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_port_scanning_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_port_scanning_vs_normal/split_metadata.json" ], "downloadable": false, "public_bundle_status": "omitted_not_redistributed", "public_csv": null, "omission_reason": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "omission_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "redistribution_caveat": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/edge_iiotset_port_scanning_vs_normal/edge_iiotset_port_scanning_vs_normal_preservedratio_train500minority_35500majority_test500minority_35500majority_additional10.csv", "local_data_path_if_rebuilt": "data/edge-iiotset/edge_iiotset_port_scanning_vs_normal/edge_iiotset_port_scanning_vs_normal_preservedratio_train500minority_35500majority_test500minority_35500majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/edge_iiotset_port_scanning_vs_normal/candidate_manifest.json", "metadata/source_manifests/modern_ids/edge_iiotset_port_scanning_vs_normal/split_metadata.json" ] } }, { "ordinal": 27, "task_key": "rt_iot2022_nmap_udp_scan_vs_benign", "title": "RT-IoT2022 NMAP UDP SCAN vs BENIGN", "source_corpus": "RT-IoT2022", "source_slice": "additional10", "positive_class": "NMAP_UDP_SCAN", "negative_class": "benign_control", "processed_csv": "data/rt-iot2022/rt_iot2022_nmap_udp_scan_vs_benign/rt_iot2022_nmap_udp_scan_vs_benign_preservedratio_train500minority_2000majority_test500minority_2000majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_udp_scan_vs_benign/rt_iot2022_nmap_udp_scan_vs_benign_preservedratio_train500minority_2000majority_test500minority_2000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "rt_iot2022_nmap_udp_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 2000, "minority": 500 }, "test_counts": { "majority": 2000, "minority": 500 }, "total_counts": { "majority": 4000, "minority": 1000, "rows": 5000 }, "actual_csv_label_counts": { "0": 4000, "1": 1000 }, "ratio_floor": 4, "feature_count": 83, "column_count": 85, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "No extra materialization drops; remaining risk is small train size relative to feature count and retained port-like fields.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_udp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_udp_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/rt-iot2022/rt_iot2022_nmap_udp_scan_vs_benign/rt_iot2022_nmap_udp_scan_vs_benign_preservedratio_train500minority_2000majority_test500minority_2000majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented UCI CC BY 4.0 route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_udp_scan_vs_benign/rt_iot2022_nmap_udp_scan_vs_benign_preservedratio_train500minority_2000majority_test500minority_2000majority_additional10.csv", "local_data_path_if_rebuilt": "data/rt-iot2022/rt_iot2022_nmap_udp_scan_vs_benign/rt_iot2022_nmap_udp_scan_vs_benign_preservedratio_train500minority_2000majority_test500minority_2000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_udp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_udp_scan_vs_benign/split_metadata.json" ] } }, { "ordinal": 28, "task_key": "rt_iot2022_nmap_xmas_tree_scan_vs_benign", "title": "RT-IoT2022 NMAP XMAS TREE SCAN vs BENIGN", "source_corpus": "RT-IoT2022", "source_slice": "additional10", "positive_class": "NMAP_XMAS_TREE_SCAN", "negative_class": "benign_control", "processed_csv": "data/rt-iot2022/rt_iot2022_nmap_xmas_tree_scan_vs_benign/rt_iot2022_nmap_xmas_tree_scan_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_xmas_tree_scan_vs_benign/rt_iot2022_nmap_xmas_tree_scan_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "rt_iot2022_nmap_xmas_tree_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 3000, "minority": 500 }, "test_counts": { "majority": 3000, "minority": 500 }, "total_counts": { "majority": 6000, "minority": 1000, "rows": 7000 }, "actual_csv_label_counts": { "0": 6000, "1": 1000 }, "ratio_floor": 6, "feature_count": 83, "column_count": 85, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "No extra materialization drops; remaining risk is small train size relative to feature count and retained port-like fields.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_xmas_tree_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_xmas_tree_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/rt-iot2022/rt_iot2022_nmap_xmas_tree_scan_vs_benign/rt_iot2022_nmap_xmas_tree_scan_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented UCI CC BY 4.0 route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_xmas_tree_scan_vs_benign/rt_iot2022_nmap_xmas_tree_scan_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "local_data_path_if_rebuilt": "data/rt-iot2022/rt_iot2022_nmap_xmas_tree_scan_vs_benign/rt_iot2022_nmap_xmas_tree_scan_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_xmas_tree_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_xmas_tree_scan_vs_benign/split_metadata.json" ] } }, { "ordinal": 29, "task_key": "rt_iot2022_nmap_os_detection_vs_benign", "title": "RT-IoT2022 NMAP OS DETECTION vs BENIGN", "source_corpus": "RT-IoT2022", "source_slice": "additional10", "positive_class": "NMAP_OS_DETECTION", "negative_class": "benign_control", "processed_csv": "data/rt-iot2022/rt_iot2022_nmap_os_detection_vs_benign/rt_iot2022_nmap_os_detection_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_os_detection_vs_benign/rt_iot2022_nmap_os_detection_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "rt_iot2022_nmap_os_detection_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 3000, "minority": 500 }, "test_counts": { "majority": 3000, "minority": 500 }, "total_counts": { "majority": 6000, "minority": 1000, "rows": 7000 }, "actual_csv_label_counts": { "0": 6000, "1": 1000 }, "ratio_floor": 6, "feature_count": 83, "column_count": 85, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "No extra materialization drops; remaining risk is small train size relative to feature count and retained port-like fields.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_os_detection_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_os_detection_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/rt-iot2022/rt_iot2022_nmap_os_detection_vs_benign/rt_iot2022_nmap_os_detection_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented UCI CC BY 4.0 route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_os_detection_vs_benign/rt_iot2022_nmap_os_detection_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "local_data_path_if_rebuilt": "data/rt-iot2022/rt_iot2022_nmap_os_detection_vs_benign/rt_iot2022_nmap_os_detection_vs_benign_preservedratio_train500minority_3000majority_test500minority_3000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_os_detection_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_os_detection_vs_benign/split_metadata.json" ] } }, { "ordinal": 30, "task_key": "rt_iot2022_nmap_tcp_scan_vs_benign", "title": "RT-IoT2022 NMAP TCP scan vs BENIGN", "source_corpus": "RT-IoT2022", "source_slice": "additional10", "positive_class": "NMAP_TCP_scan", "negative_class": "benign_control", "processed_csv": "data/rt-iot2022/rt_iot2022_nmap_tcp_scan_vs_benign/rt_iot2022_nmap_tcp_scan_vs_benign_preservedratio_train500minority_6000majority_test500minority_6000majority_additional10.csv", "source_benchmark_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_tcp_scan_vs_benign/rt_iot2022_nmap_tcp_scan_vs_benign_preservedratio_train500minority_6000majority_test500minority_6000majority_additional10.csv", "benchmark_subset": "train500_test500_modern_ids_additional10", "benchmark_code_prefix": "rt_iot2022_nmap_tcp_scan_vs_benign_additional10_train500_test500", "split_mode": "classwise_temporal", "test_size": 0.5, "val_size": 0.0, "label_column": "label", "majority_label": 0, "minority_label": 1, "train_counts": { "majority": 6000, "minority": 500 }, "test_counts": { "majority": 6000, "minority": 500 }, "total_counts": { "majority": 12000, "minority": 1000, "rows": 13000 }, "actual_csv_label_counts": { "0": 12000, "1": 1000 }, "ratio_floor": 12, "feature_count": 83, "column_count": 85, "train_test_source_row_index_overlap": 0, "dropped_columns": [], "provenance_note": "", "risk_note": "No extra materialization drops; remaining risk is attack support just above the 1000-row floor plus retained port-like fields.", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_tcp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_tcp_scan_vs_benign/split_metadata.json" ], "downloadable": true, "public_bundle_status": "downloadable_csv_included", "public_csv": "data/rt-iot2022/rt_iot2022_nmap_tcp_scan_vs_benign/rt_iot2022_nmap_tcp_scan_vs_benign_preservedratio_train500minority_6000majority_test500minority_6000majority_additional10.csv", "omission_reason": null, "omission_caveat": null, "redistribution_caveat": "Included with attribution caveat under the documented UCI CC BY 4.0 route; upstream terms still govern use.", "local_reproduction": { "benchmark_repo_processed_csv": "data/preserved_ratio_experiments/processed/train500_test500_modern_ids_additional10/rt_iot2022_nmap_tcp_scan_vs_benign/rt_iot2022_nmap_tcp_scan_vs_benign_preservedratio_train500minority_6000majority_test500minority_6000majority_additional10.csv", "local_data_path_if_rebuilt": "data/rt-iot2022/rt_iot2022_nmap_tcp_scan_vs_benign/rt_iot2022_nmap_tcp_scan_vs_benign_preservedratio_train500minority_6000majority_test500minority_6000majority_additional10.csv", "benchmark_repo_definition_paths": [ "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "research/docs/dctabgan_iteration_map/planning_docs/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "mirrored_definition_paths": [ "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_CLEAN_DCT_AND_GAN_DOSE_MATCHED_OPERATIONAL_BENCHMARK_DEFINITION_2026-05-20.json", "metadata/source_manifests/benchmark_definitions/PRESERVED_RATIO_TRAIN500_TEST500_SINGLE_LABEL_NODOS_30D_BENCHMARK_DEFINITION_2026-05-11.json" ], "rebuild_script": "python3 repo/scripts/prepare_additional_modern_ids_preserved_ratio_train500_test500_benchmark.py", "source_manifest": "metadata/source_manifests/materialization/additional10_materialization_manifest.json", "task_manifest_files": [ "metadata/source_manifests/modern_ids/rt_iot2022_nmap_tcp_scan_vs_benign/candidate_manifest.json", "metadata/source_manifests/modern_ids/rt_iot2022_nmap_tcp_scan_vs_benign/split_metadata.json" ] } } ], "downloadable_dataset_count": 17, "omitted_dataset_count": 13, "downloadable_source_corpus_counts": { "CIC-IDS-2017": 4, "CSE-CIC-IDS2018": 4, "HIKARI-2021": 2, "5G-NIDD": 3, "RT-IoT2022": 4 }, "omitted_source_corpus_counts": { "Edge-IIoTset": 6, "CIC-UNSW-NB15": 5, "CICIoMT2024Small mirror": 2 }, "public_release_policy": { "full_benchmark_task_count": 30, "downloadable_csv_count": 17, "omitted_not_redistributed_count": 13, "downloadable_corpora": { "CIC-IDS-2017": 4, "CSE-CIC-IDS2018": 4, "HIKARI-2021": 2, "5G-NIDD": 3, "RT-IoT2022": 4 }, "omitted_corpora": { "Edge-IIoTset": 6, "CIC-UNSW-NB15": 5, "CICIoMT2024Small mirror": 2 }, "omission_reasons_by_corpus": { "Edge-IIoTset": "Omitted from downloadable CSV bundle for a conservative public release posture: the official IEEE DataPort route is gated and the public mirror terms carry noncommercial/share-alike obligations that should not be overclaimed for a general public bundle.", "CIC-UNSW-NB15": "Omitted from downloadable CSV bundle because academic use and citation requirements are clear, but redistribution of repackaged or derived benchmark subsets was not explicitly verified from the documented source route.", "CICIoMT2024Small mirror": "Omitted from downloadable CSV bundle because the benchmark used a public small-artifact mirror rather than an official open Small release; mirror-specific license terms were not clearly exposed and labels were assigned from filenames." }, "license_position": "This repository does not relicense upstream IDS corpora. Downloadable CSVs are included only where the documented release posture is comparatively stronger; omitted tasks remain documented for local reproduction from the benchmark repo and user-controlled source access." } }