Table 2 Performance of learnable features for EEG classification.
From: Understanding action concepts from videos and brain activity through subjects’ consensus
Recurrent networks | Convolutional networks | ||
|---|---|---|---|
Vanilla LSTM | 16.78% | Vanilla CNN | 21.80% |
Two-branched LSTM | 27.04% | EEG images + ResNet-50 | 33.56% |
Two-branched LSTM + attention | 28.01% | DE + MLP | 39.78% |
