input Conv1D ReLU Conv1D ReLU Pooling Dropout
Flatten Dense ReLU Dropout Dense Softmax
output
Architecture (from plan): Input (7, 6) - Conv1D(16, kernel=3, same) + ReLU - Conv1D(32, kernel=3, same) + ReLU - MaxPool1D(2) - Dropout(0.3) - Flatten - Dense(32) + ReLU - Dropout(0.4) - Dense(4) + Softmax
Design rationale: - 16/32 filters (not 64/128) ; keeps model under 10KB for ESP32-S3 - kernel_size=3 ; receptive field covers 3 seconds at 1 Hz - Single MaxPool(2) > (7,32) > (3,32), can't pool more at this length - Dropout 0.3/0.4 > aggressive regularization for tiny dataset - ~5,108 total parameters > ~5-10 KB quantized INT8