Abstract
We address the problem of interactive feature construction and denoising of binary data. To this end, we develop a variational ICA method, employing a multivariate Bernoulli likelihood and independent Beta source densities. We relate this to other binary data models, demonstrating its advantages in two application domains. © Springer-Verlag Berlin Heidelberg 2006.
Cite
CITATION STYLE
APA
Kabán, A., & Bingham, E. (2006). ICA-based binary feature construction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 3889 LNCS, pp. 140–148). Springer Verlag. https://doi.org/10.1007/11679363_18
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
Already have an account? Sign in
Sign up for free