Online Learning for Two Novel Latent Topic Models

摘要

Latent topic models have proven to be an efficient tool for modeling multitopic count data. One of the most well-known models is the latent Dirichlet allocation (LDA). In this paper we propose two improvements for LDA using generalized Dirichlet and Beta-Liouville prior assumptions. Moreover, we apply an online learning approach for both introduced approaches. We choose a challenging application namely natural scene classification for comparison and evaluation purposes.