NIPS 2011 Domain Adaptation Workshop: Overfitting and Small Sample Statistics

Channel:
United States Google TechTalks
Subscribers:
348,000
Published on ● Video Link: https://www.youtube.com/watch?v=7AcmBvBb5zU


Duration: 15:47
457 views
1

Domain Adaptation Workshop: Theory and Application at NIPS 2011
Invited Talk: Overfitting and Small Sample Statistics by Ruslan Salakhutdinov

Abstract:
We study the prevalent problem when a test distribution differs from the training distribution. We consider a setting where our training set consists of a small number of sample domains, but where we have many samples in each domain. Our goal is to generalize to a new domain. For example, we may want to learn a similarity function using only certain classes of objects, but we desire that this similarity function be applicable to object classes not present in our training sample (e.g. we might seek to learn that "dogs are similar to dogs" even though images of dogs were absent from our
training set). Our theoretical analysis shows that we can select many more features than domains while avoiding overfitting by utilizing data-dependent variance properties. We present a greedy feature selection algorithm based on using T-statistics. Our experiments validate this theory showing that our T-statistic based greedy feature selection is more robust at avoiding overfitting than the classical greedy procedure.



Other Videos By Google TechTalks


2012-02-07NIPS 2011 Music and Machine Learning Workshop: Learning melodic analysis rules
2012-02-07NIPS 2011 Music and Machine Learning Workshop: Symbolic Music Information Retrieval with music21
2012-02-07NIPS 2011 Music and Machine Learning Workshop: Brain-Computer Interfaces for Music Recommendation
2012-02-07NIPS 2011 Music and Machine Learning Workshop: Automating Music Search and Recommendation
2012-02-03NIPS 2011 Learning Semantics Workshop: Towards More Human-like Machine Learning of Word Meanings
2012-02-03NIPS 2011 Learning Semantics Workshop: Learning Semantics of Movement
2012-02-03NIPS 2011 Learning Semantics Workshop: From Machine Learning to Machine Reasoning
2012-02-03NIPS 2011 Learning Semantics Workshop: How to Recognize Everything
2012-02-03NIPS 2011 Learning Semantics Workshop: Learning Dependency-Based Compositional Semantics
2012-02-02NIPS 2011 Domain Adaptation Workshop: Cool world: domain adaptation of virtual and real worlds....
2012-02-02NIPS 2011 Domain Adaptation Workshop: Overfitting and Small Sample Statistics
2012-02-02NIPS 2011 Domain Adaptation Workshop: Domain Adaptation with Multiple Latent Domains
2012-02-02NIPS 2011 Domain Adaptation Workshop: On the utility of unlabeled samples in Domain Adaptation
2012-02-02NIPS 2011 Domain Adaptation Workshop: History Dependent Domain Adaptation
2012-02-02NIPS 2011 Domain Adaptation Workshop: Adaptation without Retraining
2012-02-02NIPS 2011 Domain Adaptation Workshop: Training Structured Prediction Models
2012-02-02NIPS 2011 Domain Adaptation Workshop: Discrepancy and Adaptation
2012-02-02Will Extreme Poverty Ever End?
2012-02-01The Expedition For the Future
2011-12-22The Thorium Molten-Salt Reactor: Why Didn't This Happen (and why is now the right time?)
2011-12-16SPDY Essentials


Tags:
new
domain
adaption
008