This chapter introduces artificial neural networks (ANNs) and convolutional neural networks (CNNs) from a signal processing viewpoint. We connect CNN layers to cascaded filter banks with nonlinearities and biases, show how standard feedforward networks implement weighted sums followed by activation functions, and derive compact backpropagation updates via the chain rule. Practical Python snippets (Keras/PyTorch) illustrate training, weight inspection, and real-time blockwise inference by converting convolution into dense (linear) layers. Throughout, we highlight parallels to multirate systems, matched filtering, and polyphase thinking.