Abstract
We consider the standard problem of learning a concept from random examples. Here a learning curve can be defined to be the expected error of a learner's hypotheses as a function of training sample size. Haussler, Littlestone and Warmuth have shown that, in the distribution free setting, the smallest expected error a learner can achieve in the worst case over a concept class C converges rationally to zero error (i.e., Θ(l/t) for training sample size t). However, recently Cohn and Tesauro have demonstrated how exponential convergence can often be observed in experimental settings (i.e., average error decreasing as e Θ(−t)). By addressing a simple non-uniformity in the original analysis, this paper shows how the dichotomy between rational and exponential worst case learning curves can be recovered in the distribution free theory. These results support the experimental findings of Cohn and Tesauro: for finite concept classes, any consistent learner achieves exponential convergence, even in the worst case; but for continuous concept classes, no learner can exhibit sub-rational convergence for every target concept and domain distribution. A precise boundary between rational and exponential convergence is drawn for simple concept Chains. Here we show that somewhere dense chains always force rational convergence in the worst case, but exponential convergence can always be achieved for nowhere dense chains.
Preview
Unable to display preview. Download preview PDF.
References
S. Amari, N. Fujita, and S. Shinomoto. Four types of learning curves. Neural Computation, 4:605–618, 1992.
R. B. Ash. Real Analysis and Probability. Academic Press, San Diego, 1972.
E. B. Baum and Y.-D. Lyuu. The transition to perfect generalization in perceptrons. Neural Computation, 3:386–401, 1991.
R. Brualdi. Introductory Combinatorics. North-Holland, New York, 1977.
D. Cohn and G. Tesauro. Can neural networks do better than the Vapnik-Chervonenkis bounds? In D. Touretzky, editor, Advances in Neural Information Processing Systems 3. Morgan Kaufmann, San Mateo, CA, 1990.
D. Cohn and G. Tesauro. How tight are the Vapnik-Chervonenkis bounds? Neural Computation, 4:249–269, 1992.
M. Golea and M. Marchand. Average case analysis of the clipped Hebb rule for nonoverlapping Perceptron networks. In Proceedings COLT-93, pages 151–157, 1993.
D. Haussler, M. Kearns, H. S. Seung, and N. Tishby. Rigorous learning curve bounds from statistical mechanics. In Proceedings COLT-94, 1994.
D. Haussler, N. Littlestone, and M. K. Warmuth. Predicting {0,1}-functions on randomly drawn points. In Proceedings COLT-88, pages 280–296, 1988.
D. Haussler, N. Littlestone, and M. K. Warmuth. Predicting {0,1}-functions on randomly drawn points. Technical Report UCSC-CRL-90-54, Computer Research Laboratory, University of California at Santa Cruz, 1990.
R. J. Larsen and M. L. Marx. An Introduction to Mathematical Statistics and its Applications. Prentice-Hall, Englewood Cliffs, NJ, 1981.
M. Opper and D. Haussler. Generalization performance of Bayes optimal classification algorithm for learning a Perceptron. Physical Review Letters, 66(20):2677–2680, 1991.
M. J. Pazzani and W. Sarrett. Average case analysis of conjunctive learning algorithms. In Proceedings ML-90, pages 339–347, 1990.
J. G. Rosenstein. Linear Orderings. Academic Press, New York, 1982.
D. B. Schwartz, V. K. Samalam, S. A. Solla, and J. S. Denker. Exhaustive learning. Neural Computation, 2:374–385, 1990.
H. S. Seung, H. Sompolinsky, and N. Tishby. Learning curves in large neural networks. In Proceedings COLT-91, pages 112–127, 1991.
L. G. Valiant. A theory of the learnable. Communications of the ACM, 27(11):1134–1142, 1984.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Schuurmans, D. (1995). Characterizing rational versus exponential learning curves. In: Vitányi, P. (eds) Computational Learning Theory. EuroCOLT 1995. Lecture Notes in Computer Science, vol 904. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-59119-2_184
Download citation
DOI: https://doi.org/10.1007/3-540-59119-2_184
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-59119-1
Online ISBN: 978-3-540-49195-8
eBook Packages: Springer Book Archive