GA-177
/

File size: 1,924 Bytes
b5528bb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
{
  "schema": "secebl_rev20_l2_public_summary_v1",
  "l1_release": "SecEBL-Rev20 public L1 release",
  "l2_release": "SecEBL-Rev20 public L2 artifact",
  "model_type": "logistic_regression_session_scorer",
  "scoring_scope": "experimental fitted L2 session scorer over cached L1 semantic features",
  "runtime_feature_exclusions": [
    "raw_command_text",
    "user_name",
    "host_name",
    "session_id"
  ],
  "training_mixture": {
    "sessions": 5747,
    "positive_sessions": 426,
    "negative_sessions": 5321,
    "synthetic_pressure_positive_sessions": 60,
    "reviewed_real_pressure_positive_sessions": 1,
    "random_real_pressure_background_negative_sessions": 5000,
    "reviewed_hard_negative_pressure_sessions": 23
  },
  "validation_oof": {
    "folds": 5,
    "accuracy": 0.9939098660170523,
    "attack_precision": 0.9643705463182898,
    "attack_recall": 0.9530516431924883,
    "normal_recall": 0.9971809810186055,
    "tp": 406,
    "fn": 20,
    "fp": 15,
    "tn": 5306
  },
  "withheld_session_benchmark_fit_check": {
    "sessions": 663,
    "rows_seen": 12594,
    "attack_sessions": 365,
    "normal_sessions": 298,
    "accuracy": 1.0,
    "attack_precision": 1.0,
    "attack_recall": 1.0,
    "normal_recall": 1.0,
    "tp": 365,
    "fn": 0,
    "fp": 0,
    "tn": 298
  },
  "pressure_stream_fit_check": {
    "rows_seen": 6286568,
    "sessions": 102117,
    "alert_sessions": 61,
    "reviewed_real_alert_sessions": 1,
    "synthetic_alert_sessions": 60,
    "raw_rows_redistributed": false,
    "real_session_identifiers_redistributed": false
  },
  "thresholds": {
    "score_threshold": 0.5,
    "model_probability_threshold": 0.9,
    "score_transform": "threshold_margin",
    "score_transform_scale": 2.0
  },
  "caveat": "This is an experimental fitted L2 artifact for reproducible SecEBL-Rev20 session experiments, not an independent claim of general production IDS accuracy."
}