Poster
Gen-Oja: Simple & Efficient Algorithm for Streaming Generalized Eigenvector Computation
Kush Bhatia · Aldo Pacchiano · Nicolas Flammarion · Peter Bartlett · Michael Jordan

Thu Dec 6th 05:00 -- 07:00 PM @ Room 210 #20
in Thu Poster Session B »

In this paper, we study the problems of principle Generalized Eigenvector computation and Canonical Correlation Analysis in the stochastic setting. We propose a simple and efficient algorithm for these problems. We prove the global convergence of our algorithm, borrowing ideas from the theory of fast-mixing Markov chains and two-Time-Scale Stochastic Approximation, showing that it achieves the optimal rate of convergence. In the process, we develop tools for understanding stochastic processes with Markovian noise which might be of independent interest.

Keywords: Online Learning · Stochastic Methods · Non-Convex Optimization · Learning Theory

Author Information

Kush Bhatia (UC Berkeley)
Aldo Pacchiano (UC Berkeley)
Nicolas Flammarion (UC Berkeley)
Peter Bartlett (UC Berkeley)
Michael Jordan (UC Berkeley)

More from the Same Authors