root commited on
Commit
10e659b
·
1 Parent(s): 5dc020c

Simplify README focus on methodology

Browse files
Files changed (1) hide show
  1. README.md +0 -19
README.md CHANGED
@@ -38,7 +38,6 @@ A production-ready **Self-Normalizing Neural Network (SNN) ensemble** for predic
38
  | **Activation** | SELU + AlphaDropout |
39
  | **Training** | 300 epochs, 40-fold CV |
40
  | **CV AUC** | 0.882 ± 0.021 |
41
- | **Test AUC** | 0.990 (MoleculeNet scaffold split) |
42
 
43
  ## Molecular Features (11,369 total)
44
 
@@ -58,24 +57,6 @@ A production-ready **Self-Normalizing Neural Network (SNN) ensemble** for predic
58
  - **Normalization**: SquashScaler (StandardScaler → tanh → StandardScaler)
59
  - **Ensemble Selection**: Top-10 folds from 40-fold stratified CV
60
 
61
- ## Performance (MoleculeNet Test Split)
62
-
63
- | Endpoint | AUC | Samples | Positives |
64
- |----------|-----|---------|-----------|
65
- | NR-AR | 0.998 | 703 | 30 |
66
- | NR-AR-LBD | 0.998 | 618 | 20 |
67
- | NR-AhR | 0.985 | 601 | 85 |
68
- | NR-Aromatase | 0.992 | 496 | 45 |
69
- | NR-ER | 0.963 | 546 | 81 |
70
- | NR-ER-LBD | 0.996 | 657 | 31 |
71
- | NR-PPAR-gamma | 0.998 | 562 | 32 |
72
- | SR-ARE | 0.977 | 466 | 104 |
73
- | SR-ATAD5 | 0.998 | 651 | 28 |
74
- | SR-HSE | 0.992 | 564 | 43 |
75
- | SR-MMP | 0.993 | 512 | 108 |
76
- | SR-p53 | 0.991 | 624 | 72 |
77
- | **Mean** | **0.990** | - | - |
78
-
79
  ## Tox21 Endpoints
80
 
81
  ### Nuclear Receptor Panel
 
38
  | **Activation** | SELU + AlphaDropout |
39
  | **Training** | 300 epochs, 40-fold CV |
40
  | **CV AUC** | 0.882 ± 0.021 |
 
41
 
42
  ## Molecular Features (11,369 total)
43
 
 
57
  - **Normalization**: SquashScaler (StandardScaler → tanh → StandardScaler)
58
  - **Ensemble Selection**: Top-10 folds from 40-fold stratified CV
59
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
60
  ## Tox21 Endpoints
61
 
62
  ### Nuclear Receptor Panel