Study of neural network architectures and algorithms from the perspective of physics, using both analytical and numerical tools. Topics will include classification and regression with fully-connected networks; physics-inspired optimizers and activation functions; generative models such as normalizing flows and diffusion models; residual networks and transformers.