by Dale Schuurmans, Russell Greiner
In Proceedings IJCAI-95
http://www.cis.upenn.edu/~daes/papers/ijcai95.ps.gz
Add To MetaCart
Abstract:
We present new strategies for "probably approximately correct " (pac) learning that use fewer training examples than previous approaches. The idea is to observe training examples one-at-a-time and decide "on-line " when to return a hypothesis, rather than collect a large fixed-size training sample. This yields sequential learning procedures that pac-learn by observing a small random number of examples. We provide theoretical bounds on the expected training sample size of our procedure--- but establish its efficiency primarily by a series of experiments which show sequential learning actually uses many times fewer training examples in practice. These results demonstrate that paclearning can be far more efficiently achieved in practice than previously thought. 1
Citations
|
2961
|
Pattern Classification and Scene Analysis
– Duba, Hart
- 1973
|
|
1364
|
A theory of the learnable
– Valiant
- 1984
|
|
654
|
On the uniform convergence of relative frequencies of events to their probabilities. Theory Probab
– Vapnik, Červonekis
- 1971
|
|
317
|
Computer Systems that learn
– Weiss, Kulikowski
- 1991
|
|
293
|
What size net gives valid generalization
– Baum, Haussler
- 1989
|
|
251
|
Heuristic classification
– Clancey
- 1985
|
|
173
|
Sequential Analysis
– Wald
- 1947
|
|
130
|
A conservation law for generalization performance
– Schaffer
- 1994
|
|
58
|
From on-line to batch learning
– LITTLESTONE, N
- 1989
|
|
38
|
Learnability by fixed distributions
– Benedek, Itai
- 1988
|
|
33
|
Probably Approximately Correct Learning
– Haussler
- 1990
|
|
29
|
et al. Classification and Regression Trees
– Breiman, H
- 1984
|
|
14
|
Denker et al., Backpropagation applied to handwritten zip code recognition
– Cun
- 1989
|
|
12
|
Investigating the distributional assumptions of the pac learning model
– Bartlett, Williamson
- 1991
|
|
10
|
Implementing Valiant's Learnability Theory using Random Sets
– Oblow
- 1992
|
|
4
|
Decision theoretic generalizations of the pac model
– Haussler
- 1992
|
|
4
|
Effective Classification Learning
– Schuurmans
- 1995
|
|
1
|
et al. Instance-based learning algorithms
– Aha
- 1991
|
|
1
|
et al. Learnability and the Vapnik-Chervonen. dimension
– Blumer
- 1989
|
|
1
|
Numerical Methods for Unconstrained and Nonlinear Equations
– Dennis, Schnabel
- 1983
|
|
1
|
et al. A general lower bound on the number of examples needed for learning
– Ehrenfeucht
- 1989
|
|
1
|
et al. Results on learnability and the Vapnik-Chervonenkis dimension
– Linial
- 1991
|
|
1
|
et al. Bounding sample size with the Vapnik-Chervonenkis dimension
– Shawe-Taylor
- 1993
|
