Improved learning of k-parities

Arnab Bhattacharyya, Ameet Gadekar*, Ninad Rajgopal

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

1 Citation (Scopus)

Abstract

We consider the problem of learning k-parities in the online mistake-bound model: given a hidden vector (Formula Presented) where the hamming weight of x is k and a sequence of “questions” (Formula Presented), where the algorithm must reply to each question with (Formula Presented), what is the best trade-off between the number of mistakes made by the algorithm and its time complexity? We improve the previous best result of Buhrman et al. [BGM10] by an (Formula Presented) factor in the time complexity. Next, we consider the problem of learning k-parities in the PAC model in the presence of random classification noise of rate (Formula Presented). Here, we observe that even in the presence of classification noise of non-trivial rate, it is possible to learn k-parities in time better than (Formula Presented), whereas the current best algorithm for learning noisy k-parities, due to Grigorescu et al. [GRV11], inherently requires time (Formula Presented) even when the noise rate is polynomially small.

Original languageEnglish
Title of host publicationComputing and Combinatorics - 24th International Conference, COCOON 2018, Proceedings
Pages542-553
Number of pages12
DOIs
Publication statusPublished - 1 Jan 2018
MoE publication typeA4 Article in a conference publication
EventInternational Computing and Combinatorics Conference - Qing Dao, China
Duration: 2 Jul 20184 Jul 2018
Conference number: 24

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10976 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceInternational Computing and Combinatorics Conference
Abbreviated titleCOCOON
CountryChina
CityQing Dao
Period02/07/201804/07/2018

Fingerprint Dive into the research topics of 'Improved learning of k-parities'. Together they form a unique fingerprint.

Cite this