Abstract
The classical stochastic approximation methods are shown to yield algorithms to solve several formulations of the PAC learning problem defined on the domain [0,1]d. Under some smoothness conditions on the probability measure functions, simple algorithms to solve some PAC learning problems are proposed based on networks of nonpolynomial units (e.g. artificial neural networks). Conditions on the sizes of the samples required to ensure the error bounds are derived using martingale inequalities.
| Original language | English |
|---|---|
| Pages (from-to) | 516-522 |
| Number of pages | 7 |
| Journal | IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics |
| Volume | 27 |
| Issue number | 3 |
| DOIs | |
| State | Published - 1997 |
Funding
Manuscript received February 27, 1994; revised September 14, 1995. This work was supported by the Engineering Research Program of the Office of Basic Energy Sciences, U.S. Department of Energy, under Contract DE-AC05-840R21400 with Lockheed Martin Energy Systems, Inc. The work of N. S. V. Rao was also supported in part by the National Science Foundation under Grant IRI-9108610. An earlier version of this paper was presented at the IEEE World Congress on Computational Intelligence, Orlando, FL, 1994.