Convolutional Neural Networks, also known as ConvNets, use this convolution property in a neural network to compute better features, which can then be used to classify images or detect objects. As shown in the previous section, convolution consists of kernels which compute an output by sliding and taking a dot product with the input image. In a simple neural network, the neurons of a layer are connected to all the neurons of the next layer, but CNNs consist of convolution layers which have the property of the receptive field. Only a small portion of a previous layer's neurons are connected to the neurons of the current layer. As a result, small region features are computed through every layer as shown in the following figure:

As we have seen in a simple neural network, the neuron takes an input from one or more of previous neurons' output...