Learning the Fourier Spectrum of Probabilistic Lists and Trees

Learning the Fourier Spectrum of Probabilistic Lists and Trees (1991)

Cached

Download Links

[www.cc.gatech.edu]

Other Repositories/Bibliography

DBLP

by William Aiello , Milena Mihailt

Venue:	In Proceedings of the Second Annual ACM-SIAM Symposium on Discrete Algorithms
Citations:	7 - 0 self

Versions

Version History

Metadata Version 1

Datum	Value	Source
TITLE	Learning the Fourier Spectrum of Probabilistic Lists and Trees	INFERENCE
AUTHOR NAME	William Aiello	SVM HeaderParse 0.2
AUTHOR ADDR	291	SVM HeaderParse 0.2
AUTHOR NAME	Milena Mihailt	SVM HeaderParse 0.2
AUTHOR ADDR	291	SVM HeaderParse 0.2
ABSTRACT	method of learning boolean concepts (under uni-form sampling distribution) by reconstructing their Fourier represent ation [LMN89] extends when the concepts are probabilistic in the sense of Kearns and Shapire [KS90]. We show that probabilistic decision lists, and more generally probabilistic decision trees with at most one occurrence of each literal, can be approx-imate ed by polynomially small Fourier represent a-tions, and that the non-negligible Fourier coeffi-cients can be efficiently identified and estimated. Hence, all such concepts are learnable in polynomial time under uniform sampling distribution. This is the first instance where Fourier methods result in polynomial learning algorithms: the polynomiality of our results should be contrasted to the np”lylogn complexities in the analogous cases of [LMN89] and [M90]. The new ingredient of our work that allows us to achieve this polynomiality is that via refined Fourier analysis we are able to isolate the polynomi-ally small set of non-negligible Fourier coefficients that reside in a super-polynomially large area of the spectrum. We further observe that several more gen-eral concept classes have slightly super-polynomial (npolyk)gn) learning algorithms. These classes include all polynomial-size probabilistic decision trees, their convex combinations, etc. A concrete special case which results in polynomial learnabil-	SVM HeaderParse 0.2
YEAR	1991	INFERENCE
VENUE	In Proceedings of the Second Annual ACM-SIAM Symposium on Discrete Algorithms	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	291--299	INFERENCE
CITATIONS	8 found	ParsCit 1.0