This book provides an introduction to basic notions and relevant techniques used to obtain quantitative results in statistical mechanics of learning. A major portion of the book deals with the perceptron, the basic building block for neural networks. Early chapters discuss aspects of perceptron learning, learning rules, the storage problem, and discontinuous, unsupervised, and on-line learning. Later chapters discuss multilayer networks. A final chapter looks at related problems from statistical mechanics of complex systems, such as support vector machines, computationally hard problems, error-correcting codes, and game theory. Appendices present basic computational techniques. Examples and exercises are included.