CNN Image Classification Receptive field: 感受野 kernel size: 3×3 stride:步长 Parameter Sharing: Filter Convolutional Layer The Filters are used to detect the patterns. one Filter corresponding to one Feature Map 64 Filters corrsponding to 64 channels Understanding Neuron的通过Sharing Parameters感受不同的Receptive Field,这就是Filter Pooling: Max Pooling Convolution后,对Feature Map进行Max Pooling: CNN structure