1 layer是模型的本质，是计算的基本单元。Layers convolve filters, pool, take inner products, apply nonlinearities like rectified-linear and sigmoid and other elementwise transformations, normalize, load data, and compute losses like softmax and hinge. See the layer catalogue for all operations.
2 每层定义三种计算： setup, forward, and backward. there will be two Forward and Backward functions implemented, one for CPU and one for GPU.
3 Developing custom layers requires minimal effort by the compositionality of the network and modularity of the code. Define the setup, forward, and backward for the layer and it is ready for inclusion in a net.