Linear Classification: Signed Distance and Perceptron

This lecture covers the concept of signed distance in N dimensions for linear classification, the problem statement, and the non-centered perceptron algorithm. It also discusses test time evaluation, Python and JAVA implementations, the problem with the perceptron, logistic regression, sigmoid function, cross entropy, and probabilistic interpretation. The transition from binary to multi-class classification, linear discriminant, decision regions, and multi-class linear classification are explained. The lecture concludes with multi-class logistic regression, cross entropy, activation, and the results of multi-class classification.