Title :
Sample complexity of Bayesian optimal dictionary learning
Author :
Sakata, A. ; Kabashima, Yoshiyuki
Author_Institution :
Dept. of Comput. Intell. & Syst. Sci., Tokyo Inst. of Technol., Yokohama, Japan
Abstract :
We consider a learning problem of identifying a dictionary matrix D ∈ RM×N from a sample set of M dimensional vectors Y ∈ RM×P = N-1/2DX ∈ RM×P, where X ∈ RN×p is a sparse matrix in which the density of non-zero entries is 0 <; ρ <; 1. In particular, we focus on the minimum sample size Pc (sample complexity) necessary for perfectly identifying D of the optimal learning scheme when D and X are independently generated from certain distributions. By using the replica method of statistical mechanics, we show that Pc ~ O(N) holds as long as α = M/N > ρ is satisfied in the limit of N → ∞. Our analysis also implies that the posterior distribution given Y is condensed only at the correct dictionary D when the compression rate α is greater than a certain critical value αM(p). This suggests that belief propagation may allow us to learn D with a low computational complexity using O(N) samples.
Keywords :
Bayes methods; computational complexity; learning (artificial intelligence); sparse matrices; statistical mechanics; vectors; Bayesian optimal dictionary learning; M dimensional vectors; compression rate; dictionary matrix identification; nonzero entry density; posterior distribution; replica method; sample complexity; sparse matrix; sparse representations; statistical mechanics; Decision support systems; Information theory; Manganese;
Conference_Titel :
Information Theory Proceedings (ISIT), 2013 IEEE International Symposium on
Conference_Location :
Istanbul
DOI :
10.1109/ISIT.2013.6620310