Baby Steps towards Convolution Nets - IV

In an earlier post we saw what a crosscorr2d() operator is. In this we will see how it is applied.

A convolution layer performs crosscorr2d() operation on the input and kernels. It then adds a scalar bias to produce the output. A convolution layer accepts the kernel and a scalar bias as inputs.

class conv2D : public Block {
    conv2D(kernel_size, ...)
    {
        Shape s{1};
        my_weight = get_params("weight", kernel_size);
        my_bias = get_params("bias", s);
    }

    void forward(x)
    {
        return crosscorr2d(x, my_weight.data(), my_bias.data());
    }
};

The forward computation function calls crosscorr2d function and adds the bias.