Multi-class classification on MedSec-25
97.87PrecisionTM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TMParameters=Binarizer: KBinsDiscretizer, n_bins=40, encode=onehot-dense, strategy=quantile, number_of_clauses=2000, T=30, s=15, weighted_clauses=False, Epochs=45, Inference time=66.24 ± 1.11 µs2026.05 | 97.87 | 97.83 | 97.83 | |
| STMParameters=Binarizer: KBinsDiscretizer, n_bins=25, encode=onehot-dense, strategy=quantile, number_of_clauses=1500, T=30, s=20, weighted_clauses=False, literal_sampling=1, max_included_literals=16, absorbing=0, Epochs=50, Inference time=16.89 ± 0.29 µs2026.05 | 97.38 | 97.32 | 97.33 | |
| RFParameters=compute_class_weight method with class_weight=balanced, Inference time=3.73 ± 0.10 µs2026.05 | 95.04 | 95.21 | 95.11 | |
| DTParameters=compute_class_weight method with class_weight=balanced, Inference time=0.08 ± 0.04 µs2026.05 | 94.67 | 94.35 | 94.51 | |
| KNNParameters=n_neighbors=5, Inference time=186.44 ± 3.19 µs2026.05 | 93.66 | 93.1 | 93.36 | |
| XGBoostParameters=objective=multi:softprob, eval_metric=mlogloss, tree_method=hist, learning_rate=0.1, max_depth=8, n_estimators=200, Inference time=1.77 ± 0.08 µs2026.05 | 91.75 | 97.97 | 94.47 | |
| LGBMParameters=objective=multiclass, learning_rate=0.1, n_estimators=200, compute_class_weight method with class_weight=balanced, Inference time=6.00 ± 0.38 µs2026.05 | 91.74 | 97.95 | 94.43 | |
| NNParameters=input layer=79 neurons, first hidden layer=64 neurons, second hidden layer=32 neurons, output layer=1 neurons, hidden layer activation function=relu, output layer activation function=softmax, optimizer=adam, loss=sparse_categorical_crossentropy, metrics=accuracy, epochs=10, batch_size=32, verbose=0, compute_class_weight method with class_weight=balanced, Inference time=6.76 ± 0.75 µs2026.05 | 82.45 | 95.19 | 87.29 | |
| LRParameters=solver=lbfgs, max_iter=500, multi_class=multinomial, compute_class_weight method with class_weight=balanced, Inference time=0.05 ± 0.00 µs2026.05 | 58.58 | 77.1 | 61.19 | |
| NBParameters=default settings, Inference time=0.65 ± 0.03 µs2026.05 | 49.91 | 56.81 | 44.76 |