简介: |
讲演摘要
Large-scale classification is an important part of machine learning in the big-data era. Large hierarchies and graphs of categories present significant challenges and opportunities for structured learning. We present novel methods and scalable algorithms for Hierarchical Bayesian Logistic Regression (HBLR) and Regularized Regularization (RR) over both hierarchical and graphical dependency structures. Using these methods we obtained the best results on the largest data sets in benchmark
evaluations (the PASCAL Challenges for large scale classification), and successfully solved the joint optimization problems with 600+ thousands of classifiers and one trillion of model parameters in 37 hours.
讲演者简介
Yiming Yang is a professor with a joint appointment in the Language Technologies Institute and the Machine Learning Department. She received her Ph.D. in Computer Science from Kyoto University (Japan), and has been a faculty member at Carnegie Mellon University since 1996.Her research has centered on statistical learning methods/algorithms and application to very-large-scale text categorization, web-mining for concept graph discovery, graphical models for topic detection and tracking, semi-supervised clustering, multi-task learning, novelty-based information retrieval, large-scale optimization for online advertising, social network analysis for personalized email prioritization, etc. Her publications have received over 22,000 citations as of Feb 2014 in Google Scholar. She brings strengths in statistical modeling techniques and scalable algorithms.
|
