Agnostic active learning
| Autor: | Alina Beygelzimer, Maria-Florina Balcan, John Langford |
|---|---|
| Rok vydání: | 2009 |
| Předmět: |
Computer Science::Machine Learning
Unit sphere Active learning Wake-sleep algorithm Computer Networks and Communications Competitive learning Sample complexity Semi-supervised learning Theoretical Computer Science Instance-based learning Empirical risk minimization Mathematics Learning classifier system Linear separators business.industry Applied Mathematics Supervised learning Pattern recognition Exponential function Computational Theory and Mathematics Homogeneous Agnostic setting Unsupervised learning Artificial intelligence business Algorithm Classifier (UML) |
| Zdroj: | ICML |
| ISSN: | 0022-0000 |
| Popis: | We state and analyze the first active learning algorithm that finds an @e-optimal hypothesis in any hypothesis class, when the underlying distribution has arbitrary forms of noise. The algorithm, A^2 (for Agnostic Active), relies only upon the assumption that it has access to a stream of unlabeled examples drawn i.i.d. from a fixed distribution. We show that A^2 achieves an exponential improvement (i.e., requires only O([email protected]) samples to find an @e-optimal classifier) over the usual sample complexity of supervised learning, for several settings considered before in the realizable case. These include learning threshold classifiers and learning homogeneous linear separators with respect to an input distribution which is uniform over the unit sphere. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect