conftrace_

2013 COLT COLT 2013

Learning a set of directions

Abstract

Assume our data consists of unit vectors (directions) and we are to find a small orthogonal set of the “the most important directions” summarizing the data. We develop online algorithms for this type of problem. The techniques used are similar to Principal Component Analysis which finds the most important small rank subspace of the data.The new problem is significantly more complex since the online algorithm maintains uncertainty over the most relevant subspace as well as directional information.

🐣 Hot Topic Early Bird - online algorithm

Authors

Wouter M. Koolen, Jiazhong Nie, Manfred Warmuth

Topics

Machine Learning > Core Methods > Dimensionality Reduction Machine Learning > Learning Types > Online Learning Machine Learning > Core Methods > Feature Learning

Keywords

dimensionality reduction principal component analysis subspace learning orthogonal decomposition online algorithm orthogonal set directional information

More from COLT 2013

All COLT 2013 papers →

Optimal Probability Estimation with Applications to Prediction and Classification

Polynomial Time Optimal Query Algorithms for Finding Graphs with Arbitrary Real Weights

Differentially Private Feature Selection via Stability Arguments, and the Robustness of the Lasso

A Tensor Spectral Approach to Learning Mixed Membership Community Models

Adaptive Crowdsourcing Algorithms for the Bandit Survey Problem

Boosting with the Logistic Loss is Consistent