Resource-Bounded Measure and Learnability

Abstract. We consider the resource-bounded measure of polynomial-time learnable subclasses of polynomial-size circuits. We show that if EXP ≠ MA, then every PAC-learnable subclass of P/poly has EXP-measure zero. We introduce a nonuniformly computable variant of resource-bounded measure and show that, for every fixed polynomial q , any polynomial-time learnable subclass of circuits of size q has measure zero with respect to P/poly. We relate our results to the question of whether the class of Boolean circuits is polynomial-time learnable.

[1]  Dieter van Melkebeek,et al.  Hard Sets Are Hard to Find , 1999, J. Comput. Syst. Sci..

[2]  Jeffrey C. Jackson,et al.  An efficient membership-query algorithm for learning DNF with respect to the uniform distribution , 1994, Proceedings 35th Annual Symposium on Foundations of Computer Science.

[3]  Michael Luby,et al.  Pseudorandomness and cryptographic applications , 1996, Princeton computer science notes.

[4]  Carsten Lund,et al.  Non-deterministic exponential time has two-prover interactive protocols , 2005, computational complexity.

[5]  N. Littlestone.  Learning Quickly When Irrelevant Attributes Abound: A New Linear-Threshold Algorithm , 1987, 28th Annual Symposium on Foundations of Computer Science (sfcs 1987).

[6]  Umesh V. Vazirani,et al.  An Introduction to Computational Learning Theory , 1994 .

[7]  Noam Nisan,et al.  Hardness vs Randomness , 1994, J. Comput. Syst. Sci..

[8]  Temple F. Smith.  Occam's razor , 1980, Nature.

[9]  Eric Allender,et al.  Measure on small complexity classes, with applications for BPP , 1994, Proceedings 35th Annual Symposium on Foundations of Computer Science.

[10]  Johannes Köbler,et al.  Oracles in Sp2 are Sufficient for Exact Learning , 2000, Int. J. Found. Comput. Sci..

[11]  Dana Angluin,et al.  When won't membership queries help? , 1991, STOC '91.

[12]  Jin-Yi Cai,et al.  Pseudorandom generators, measure theory, and natural proofs , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[13]  Silvio Micali,et al.  How to construct random functions , 1986, JACM.

[14]  Jin-Yi Cai,et al.  Constant depth circuits and the Lutz hypothesis , 1997, Proceedings 38th Annual Symposium on Foundations of Computer Science.

[15]  Leslie G. Valiant,et al.  A theory of the learnable , 1984, STOC '84.

[16]  Leonid A. Levin,et al.  Pseudo-random generation from one-way functions , 1989, STOC '89.

[17]  Leonid A. Levin,et al.  A hard-core predicate for all one-way functions , 1989, STOC '89.

[18]  Noam Nisan,et al.  BPP has subexponential time simulations unless EXPTIME has publishable proofs , 1991, [1991] Proceedings of the Sixth Annual Structure in Complexity Theory Conference.

[19]  Jack H. Lutz.  Almost Everywhere High Nonuniform Complexity , 1992, J. Comput. Syst. Sci..

[20]  Sebastiaan Terwijn,et al.  Resource Bounded Randomness and Weakly Complete Problems , 1994, Theor. Comput. Sci..

[21]  Dieter van Melkebeek,et al.  A Generalization of Resource-Bounded Measure, With an Application (Extended Abstract) , 1998, STACS.

[22]  Eric Allender,et al.  Measure on P: Robustness of the Notion , 1995, MFCS.