This summer school aims to facilitate learning and interaction in various topics concerning non-convex optimization in machine leanring. There are several applications in machine learning where non-convex optimization problems naturally arise. These include feature selection, kernel learning,structure learning in graphical models, inference, inductive logic programming, semi-supervised learning, tranductive learning, active learning, hyper-parameter learning, summarization etc. The focus of the summer school is on approaches for non-convex optimization while illustrating through some of these motivating applications. Popular approaches for non-convex optimization are: approximation algorithms, convex relaxations, randomized algorithms,submodular optimization, dynamic programming, interger and mixed-integer programming. The duration of the summer school will be 10 days. The first eight days will comprise of lecture series interspersed with tutorials/lab-sessions. A competition based on the topics covered will be held on the final two days. The competition will facilitate students to design and implement algorithms for a select set of machine learning problems.