CNN Architecture - LeNet-5 Example
Due to Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner. 1998
- Inputs: 32 x 32 pixel digits images
- Hidden Layer 1: Uses 5x5 kernel (with bias this means 26 weights), outputs 6, 24 x 24 feature maps.
- Hidden Layer 2: Uses 2x2 pixel maxpool layer, outputs 6, 14 x 14 feature maps.
- Hidden Layer 3: Uses 5x5 pixel kernel, outputs 16, 10 x 10 feature maps.
- Hidden Layer 4: Uses 2x2 pixel maxpool layer, outputs 16, 5 x 5 feature maps.
- Hidden Layer 6: Uses 5x5 pixel kernel, outputs 120, 1 x 1 feature maps.
- Hidden Layer 7: Fully connected to 120 input, 84 outputs.
- Output: 10 radial basis functions each getting 84 inputs, and whose outputs correspond to the different digits.