Table 2.
Performances on Guess What? data set. Results are reported with standard deviation over 5 different runs for each model.
Model | Accuracy, mean (SD) | Precision, mean (SD) | Recall, mean (SD) | F1 score, mean (SD) | AUROCa, mean (SD) |
Random forest | 0.697 (0.013) | 0.687 (0.010) | 0.744 (0.247) | 0.694 (0.013) | 0.740 (0.09) |
Convolutional neural network | 0.793 (0.013) | 0.804 (0.014) | 0.793 (0.014) | 0.790 (0.014) | 0.822 (0.010) |
Wav2vec 2.0 | 0.769 (0.005) | 0.782 (0.021) | 0.746 (0.031) | 0.768 (0.006) | 0.815 (0.077) |
aAUROC: area under the receiver operating characteristic curve.