NeuralNetwork framework mainly for AVR/ESP microcontrollers, but can be used on other architectures.
- Fully Connected (DenseLayer)
- Convolution (ConvLayer1D and ConvLayer2D)
- Max Pooling (MaxPoolingLayer1D and MaxPoolingLayer1D)
- Zero Padding (ZeroPaddingLayer1D and ZeroPaddingLayer2D)
- Reshaping Layer (ReshapeLayer)
- Flattening Layer (FlattenLayer)
- Drop Out Layer (DropOutLayer, DropOutLayer1D and DropOutLayer2D)
- Concatenation layer (ConcatLayer)
- Supported trainers:
- Gradient Descent Back Propagation for Dense and Convolution layers.