conftrace_

2011 NeurIPS NeurIPS 2011

Active Learning with a Drifting Distribution

Abstract

We study the problem of active learning in a stream-based setting, allowing the distribution of the examples to change over time. We prove upper bounds on the number of prediction mistakes and number of label requests for established disagreement-based active learning algorithms, both in the realizable case and under Tsybakov noise. We further prove minimax lower bounds for this problem.

📈 Trend Setter - Distribution Shift

🐣 Hot Topic Early Bird - distribution shift

Authors

Liu Yang

Topics

Machine Learning > Learning Types > Active Learning Machine Learning > Optimization & Theory > Learning Theory Machine Learning > Optimization & Theory > Statistical Learning Artificial Intelligence > Learning Paradigms > Active Learning Machine Learning > Learning Types > Distribution Shift

Keywords

active learning label complexity distribution shift tsybakov noise distribution drift stream-based learning minimax bounds disagreement-based learning mistake bound minimax lower bound label request prediction mistake

More from NeurIPS 2011

All NeurIPS 2011 papers →

Beyond Spectral Clustering - Tight Relaxations of Balanced Graph Cuts

Why The Brain Separates Face Recognition From Object Recognition

Analytical Results for the Error in Filtering of Gaussian Processes

Evaluating the inverse decision-making approach to preference learning

Policy Gradient Coagent Networks

Periodic Finite State Controllers for Efficient POMDP and DEC-POMDP Planning