Events Calendar

Title: The Power of Stagewise Learning: From Support Vector Machine to Generative Adversarial Nets

Presented by: Tianbao Yang, Asst. Professor

Computer Science, University of Iowa

Abstract: In the past decade, we have seen dramatic evolution of machine learning from convex methods to non-convex methods and from shallow models to deep models.  Lying at the heart of machine learning, mathematical optimization has played an important and indispensable role in solving many different learning problems.  However, there is still a big gap between the practice used in deep learning community and the existing theory.  In this talk, I will focus on a learning paradigm called stagewise learning that is different from conventional learning methods based on stochastic gradient descent with a continuously decreasing step size.  I will show that the proposed stagewise gradient methods for solving many machine learning problems, including support vector machine, AUC optimization, deep neural networks and degenerative adversarial networks.