M. Kearns. Efficient noise-tolerant learning from statistical queries. In ACM Symposium on Theory of Computing, pages 392--401, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the Microchoice bound. This improvement is not easily expressed as a simplification of Structural Risk Minimization. First we require some background material in order to state and understand Freund s bound. 4. 1 Preliminaries and Definitions The statistical query framework introduced by Kearns [Kea93] is the same as the PAC framework, except that the learning algorithm can only access its data using statistical queries. A statistical query takes as input a binary predicate, mapping examples to a binary output: X; Y ) f0; 1g. The output of the statistical query is the average of over ....

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, pages 392--401. ACM Press, New York, NY, 1993.

....to determine how well it works in practice. In general, the additional work described in section 6.6 on monitoring is a good direction for future work. The most interesting direction for work on CARD is automatic derivation of dependencies. The idea here is to use either machine learning [AL88, Kea93, BHL91, KL88, Lit89] or association rule mining [AIS93, AS94] techniques to automatically determine dependencies. This approach requires having some monitored values that indicate if a system is up or down. Then, if we can show that any time component 1 is down, component 2 is also down, but not ....

Michael Kearns. Efficient Noise-Tolerant Learning From Statistical Queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 392--401, May 1993.

....work without having to put the classical part of the algorithm into the quantum system. Our main contributions are results that answer this question in the negative, for several natural notions of general . We achieve these results through a connection to the notion of statistical query learning [22] studied in Computational Learning Theory, and in particular to a related notion that we introduce of statistical query sampling. Using techniques from Fourier analysis and cryptography, we show that even in cases where the distribution implied by Q is quite simple, it can be hard to use the EV ....

....hardness result for this problem, that holds for the specific set S used by Simon s algorithm (Theorem 2) 1. 2 Techniques and relation to Statistical Query learning Our results are based on a connection to the Statistical Query (SQ) learning model, first introduced by Kearns [22] as a restricted version of the popular Probably Approximately Correct (PAC) model of Valiant [30] In these learning models, the goal of an algorithm is to learn an approximation to a hidden function f : f0; 1g 7 f0; 1g. In the PAC model, the algorithm has access to an example oracle , ....

[Article contains additional citation context not shown here]

M. Kearns. Efficient noise-tolerant learning from statistical queries. In STOC 1993.

....is much easier to handle than the worst case problem. In fact, with very simple algorithms one can (whp) produce a clustering that is quite close to OPT, much closer than the number of disagreements between OPT and f . The analysis is fairly standard (much like the generic transformation of Kearns [16] in the machine learning context, and even closer to the analysis of Condon and Karp for graph partitioning [11] In fact, this problem nearly matches a special case of the plantedpartition problem of McSherry [18] We present our analysis anyway since the algorithms are so simple. One sided ....

M. Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 392--401, 1993.

....learner would ordinarily see. While a limited number of eificient PAC algorithms had been developed which tolerate classification noise [2, 16, 26] no general framework for eflcient learning 1 in the presence of classification noise was known until Kearns introduced the Statistical Query model [19]. 1Angluin and Laird [2] introduced a general framework for learning in the presence of classification noise. However, their methods do not yield computationally efficient algorithms in most cases. In the SQ model, the example oracle of the standard PAC model is replaced by a statistics oracle. ....

....the maximum number of statistics required, and tolerance, the minimum additive error required. The time and sample complexities of simulating SQ algorithms in the PAC model are directly affected by these measures; therefore, we would like to bound these measures as closely as possible. Kearns [19] has demonstrated two important properties of the SQ model which make it wor thy of study. First, he has shown that nearly every PAC learning algorithm can be cast within the SQ model, thus demonstrating that the SQ model is quite general and imposes a rather weak restriction on learning ....

[Article contains additional citation context not shown here]

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the 25 tn Annual A CM Symposium on the Theory of Computing, pages 392-401, San Diego, 1993.

....to within error ffl the Hamiltonian property on random graphs G(n; p) for any p. We also discuss the connection of this work with a related result [BFF87] All of the algorithms considered in this papers fall in the category of a statistical query algorithm, and hence, by the result of Kearns [K93], are noise tolerant. The paper is organized as follows: Section 2 is devoted to a description of the learning models considered in this paper. Section 3 describes notations and the basic theory of the Fourier transform for Boolean functions. It also contains some basic facts that will be ....

....may be ignored without incurring an error of more than O(1) Now observe that with this simplification, there are at most log variables. This problem reduces to Theorem 8.4. 2 Remark. The learning algorithms discussed so far fit into the statistical query learning model introduced by Kearns [K93]. Hence by Kearns results, these algorithms are robust against classification noise in the example oracle. 9 Acknowledgments This paper would not have existed without Jeff Jackson telling us about all the neat things related to Fourier transform. His enthusiastic lectures during his visit at ....

Michael Kearns. Efficient Noise Tolerant Learning from Statistical Queries. In Proceedings of the Twenty Fifth Annual ACM Symposium on the Theory of Computing, pages 392-- 401, 1993.

....learner would ordinarily see. While a limited number of efficient PAC algorithms had been developed which tolerate classification noise [2, 16, 26] no general framework for efficient learning in the presence of classification noise was known until Kearns introduced the Statistical Query model [19]. Angluin and Laird [2] introduced a general framework for learning in the presence of classification noise. However, their methods do not yield computationally efficient algorithms in most cases. In the SQ model, the example oracle of the standard PAC model is replaced by a statistics oracle. ....

....the maximum number of statistics required, and tolerance, the minimum additive error required. The time and sample complexities of simulating SQ algorithms in the PAC model are directly affected by these measures; therefore, we would like to bound these measures as closely as possible. Kearns [19] has demonstrated two important properties of the SQ model which make it worthy of study. First, he has shown that nearly every PAC learning algorithm can be cast within the SQ model, thus demonstrating that the SQ model is quite general and imposes a rather weak restriction on learning ....

[Article contains additional citation context not shown here]

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 392--401, 1993.

.... one or m keys be selected from S at random ; our intent is that Alice would build the designated key or keys, and then assist Bob in building the same one(s) The statistical idea used to solve the problems resembles, albeit on a smaller scale, the methods for learning discrete distributions in [15, 16]. A second point of interest in our work is that it gives a new twist on muchstudied distributed consensus problems. In a typical consensus problem involving k processors, each processor is given some value, and the goal is for the processors to agree on one of the values that were initially ....

....the fault free case, it remains to ask whether the same lower bounds still apply. The introduction of either noise or systematic faults in the communicated sampling data leads to a completely new problem. The combination of ideas in our work, the noise tolerant learning model developed by Kearns [15], and the work on distributed protocols described in the Introduction may have important consequences. ....

M. Kearns. Efficient noise-tolerant learning from statistical queries. In Proc. 25th STOC, pages 392--401, 1993.

.... a polynomial time randomized algorithm using membership queries to learn DFA s with high rates of random persistent errors in the answers to the membership queries [21] Algorithms that use membership queries to estimate probabilities (in the spirit of the statistical queries defined by Kearns [17]) are generally not too sensitive to small rates of random persistent errors in the answers to queries. For example, Goldman, Kearns, and Schapire give polynomial time algorithms for exactly learning read once majority formulas and read once positive NAND formulas of depth O(log n) with high ....

M. Kearns. Efficient noise-tolerant learning from statistical queries. In Proc. 25th Annu. ACM Sympos. Theory Comput., pages 392--401. ACM Press, New York, NY, 1993.

No context found.

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, May 1993.

No context found.

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, May 1993.

No context found.

Michael J. Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the TwentyFifth Annual ACM Symposium on Theory of Computing, pages 392--401, 1993.

No context found.

M. Kearns. Efficient noise-tolerant learning from statistical queries. In ACM Symposium on Theory of Computing, pages 392--401, 1993.

No context found.

M. J. Kearns. Efficient noise-tolerant learning from statistical queries. Journal of the ACM, 45(6): 983--1006, 1998.

No context found.

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, pages 392--401. ACM Press, New York, NY, 1993.

No context found.

Kearns, M. (1993). Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 392--401.

No context found.

M. Kearns. Efficient Noise-Tolerant Learning from Statistical Queries. Journal of the ACM, 45(6):983--1006, November 1998.

No context found.

M. Kearns. Efficient noise-tolerant learning from statistical queries. In ACM Symposium on Theory of Computing, pages 392--401, 1993.

No context found.

M. Kearns. Efficient Noise-Tolerant Learning from Statistical Queries. Journal of the ACM, 45(6):983--1006, November 1998.

No context found.

M. Kearns, "Efficient noise-tolerant learning from statistical queries," in Proceedings of the ACM Symposium on Theory of Computing, 1993.

No context found.

Kearns, M. Efficient noise-tolerant learning from statistical queries. Journal of the ACM, 45:6, 983-1006, 1998.

No context found.

Michael Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, pages 392-- 401, 1993.

No context found.

M. Kearns. Efficient noise-tolerant learning from statistical queries. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 392--401, 1993.

No context found.

M. Kearns. Efficient noise-tolerant learning from statistical queries. In STOC 1993.

No context found.

M. Kearns. Efficient noise-tolerant learning from statistical queries. In Proc. 25th Annu. ACM Sympos. Theory Comput., pages 392--401. ACM Press, New York, NY, 1993.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC