Cell type deconvolution on OOD experiment dataset
20Total Configurations (Higher TCS)Dismir (CNN+LSTM)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Dismir (CNN+LSTM)Labeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/o pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 20 | 17 | |
| Dismir (CNN+LSTM)Labeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/ pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 20 | 16 | |
| CancerDetectorPrior strategy=Training counts prior, Classifier category=Labeling-independent classifiers2026.07 | 20 | 10 | |
| Dismir (CNN+LSTM)Labeling scheme=Canonical Soft Labels, Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 16 | 9 | |
| Lookup Classifier (1-NN)Labeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/o pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 16 | 9 | |
| Lookup Classifier (1-NN)Labeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/ pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 15 | 7 | |
| MethylBERTLabeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/ pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 14 | 3 | |
| Dismir (CNN+LSTM)Labeling scheme=Hard Labels w/ bckg., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 12 | 6 | |
| MethylBERTLabeling scheme=Canonical Soft Labels, Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 11 | 1 | |
| CancerDetectorPrior strategy=Uniform prior, Classifier category=Labeling-independent classifiers2026.07 | 11 | 4 | |
| Dismir (CNN+LSTM)Labeling scheme=Hard Labels w/ bckg., Feature selection strategy=DiagBckg, Classifier category=Classifiers trained on labeled data2026.07 | 8 | 1 | |
| MethylBERTLabeling scheme=DD Soft Labels (Data-Driven), Pooling strategy=w/o pool., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 6 | 0 | |
| MethylBERTLabeling scheme=Hard Labels w/ bckg., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 4 | 0 | |
| Dismir (CNN+LSTM)Labeling scheme=Canonical Soft Labels, Feature selection strategy=DiagBckg, Classifier category=Classifiers trained on labeled data2026.07 | 2 | 0 | |
| MethylBERTLabeling scheme=Hard Labels w/ bckg., Feature selection strategy=DiagBckg, Classifier category=Classifiers trained on labeled data2026.07 | 0 | 0 | |
| MethylBERTLabeling scheme=Canonical Soft Labels, Feature selection strategy=DiagBckg, Classifier category=Classifiers trained on labeled data2026.07 | 0 | 0 | |
| Lookup Classifier (1-NN)Labeling scheme=Hard Labels w/ bckg., Feature selection strategy=Top156, Classifier category=Classifiers trained on labeled data2026.07 | 0 | 0 |