Log inImperial users Other users No account?Information onFinding a talk Adding a talk Syndicating talks Who we are Everything else |
Confidence-Based Active LearningAdd to your list(s) Download to your calendar using vCal
If you have a question about this talk, please contact Patrick Kelly. We study agnostic active learning, where the goal is to learn a classifier in a pre-specified hypothesis class interactively with as few label queries as possible, while making no assumptions on the true function generating the labels. The main algorithms for this problem are disagreement-based active learning, which has a high label requirement, and margin-based active learning, which is better, but only applies to fairly restricted settings. Thus a challenge is to find an algorithm which achieves better label complexity, is consistent in an agnostic setting, and applies to general classification problems. In this paper, we provide such an algorithm. Our solution is based on two novel contributions—a reduction from consistent active learning to confidence-rated prediction with guaranteed error, and a novel confidence-rated predictor. This talk is based on joint work with Chicheng Zhang. About the presenter: Kamalika Chaudhuri received a Bachelor of Technology degree in Computer Science and Engineering in 2002 from Indian Institute of Technology, Kanpur, and a PhD in Computer Science from University of California at Berkeley in 2007. She held a postdoctoral researcher position at the Information Theory and Applications Center at UC San Diego from 2007-2009, and a postdoctoral researcher position in the CSE department at UC San Diego from 2009-2010. In July 2010, she joined the CSE department at UC San Diego as an assistant professor. Dr Chaudhuri’s research interests are in machine-learning, a subfield that lies at the intersection of statistics and computer science. Much of her work is on privacy-preserving machine learning and unsupervised learning, but she is broadly interested in a number of topics in learning theory, such as confidence-rated prediction, online learning, and active learning. This talk is part of the Featured talks series. This talk is included in these lists:
Note that ex-directory lists are not shown. |
Other listsTalks@ee.imperial Type the title of a new list here COMMSP SeminarOther talksStability and power sharing in microgrids Multimodal Learning for Robot Behavior Generation utilizing Neuro-dynamical Model |