Citation:
Bhattacharyya , A , Gadekar , A & Rajgopal , N 2018 , Improved learning of k-parities . in Computing and Combinatorics - 24th International Conference, COCOON 2018, Proceedings . Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) , vol. 10976 LNCS , Springer Verlag , pp. 542-553 , International Conference on Computing and Combinatorics Conference , Qing Dao , China , 02/07/2018 . DOI: 10.1007/978-3-319-94776-1_45
|
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.
|