A Novel Time-Incremental End-to-End Shared Neural Network with Attention-Based Feature Fusion for Multiclass Motor Imagery Recognition

<div>The structure and principle of bidirectional long short-term memory network with time increments. “<i>S</i>” represents sigmoid operator; “tanh” signifies hyperbolic tangent operator. “<i>C</i><sub><i>t</i></sub>” represents the state of BiLSTM cell at <i>t</i> moment.</div>

Computational Intelligence and Neuroscience

fig5

Figure 5

Figure 5: A Novel Time-Incremental End-to-End Shared Neural Network with Attention-Based Feature Fusion for Multiclass Motor Imagery Recognition 