We discuss the new supervised, semi-supervised and unsupervised methods. Main subjects include:
Loss Function and Optimization
Bias-variance trade-off, model selection criteria
Bayesian linear models
Lasso regression, Least angle regression
Nonparametric regression and smoothing splines
Discrimination techniques (logistic regression, linear and quadratic discriminants, nonparametric logistic regression, mixture of discriminants)
Separating hyperplanes and support vector machines
Additive trees, MARS, and CART
EM algorithm and applications in semi-supervised learning
Bagging and boosting
Neural netwokrs and deep learning
Bayesian methods and variational inference

To get an idea what this course is about, draw in the square and a k-nearest-neighbour classification algorithm recognizes the digit you drew.
Check a nice convolutional neural networks which is much more precise compred to nearest-neighbours HERE.
Check another visualization on deep neural networks by Google HERE.