Boolean Factor Analysis by Expectation-Maximization Method
Boolean factor analysis is one of the most efficient methods to reveal and to overcome informational redundancy of high-dimensional binary signals. In the present study, we introduce new Expectation-Maximization method which maximizes the likelihood of Boolean factor analysis solution. Using the so-called bars problem benchmark, we compare efficiencies of the proposed method with Dendritic Inhibition neural network.
KeywordsFactor Score Information Gain Binary Signal Distorted Version Mixed Factor
This paper has been partly elaborated in the framework of the IT4Innovations Centre of Excellence project, reg. no. CZ.1.05/1.1.00/02.0070, supported by Operational Programme ’Research and Development for Innovations’ funded by Structural Funds of the European Union and state budget of the Czech Republic and partly supported by the projects AV0Z10300504, GACR P202/10/0262, 205/09/1079.
- 1.Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum likelihood from incomplete data via the EM algorithm. J. Royal Stat. Soc. Ser. B (Methodol.) 39(1), 1–38 (1977)Google Scholar
- 3.Frolov, A.A., Husek, D., Polyakov, P.Y.: New measure of boolean factor analysis quality. Adapt. Nat. Comput. Algorithms 6593, 100–109 (2011a)Google Scholar
- 4.Frolov, A.A., Husek, D., Polyakov, P.Y.: Bulevskij faktornyj analiz na osnove attraktornoj nejronnoj seti i nekotoryje ego prilozenija (Boolean factor analysis by means of attractor neural network and some its applications). Neirokomputery: Razrabotka, Primenenie (in Russian ISSN 1999–8554) 1, 25–46 (2011b)Google Scholar
- 5.Frolov, A.A., Husek, D., Polyakov, P.Y.: Expectation-maximization approach to boolean factor analysis. In: Proceedings of International Joint Conference on Neural Network. San Jose, California (July 2011c)Google Scholar