Table 13 Results of isomorphism calculations for the subsets of dataset 2
From: DECIMER 1.0: deep learning for chemical image recognition using transformers
Metrics | Subset 5 | Subset 6 |
|---|---|---|
Train data size | 15,360,000 | 33,304,320 |
Test data size | 1,536,000 | 3,700,480 |
Predictions with Tanimoto 1.0 | 1,155,483 | 3,325,656 |
Isomorphic predictions | 96.42% | 98.50% |
Non isomorphic predictions | 3.58% | 1.50% |