Transfer Nonnegative Matrix Factorization for Image Representation

Abstract

Abstract. Nonnegative Matrix Factorization (NMF) has received considerable attention due to its psychological and physiological interpretation of naturally occurring data whose representation may be partsbased in the human brain. However, when labeled and unlabeled images are sampled from different distributions, they may be quantized into different basis vector space and represented in different coding vector space, which may lead to low representation fidelity. In this paper, we investigate how to extend NMF to cross-domain scenario. We accomplish this goal through TNMF -a novel semi-supervised transfer learning approach. Specifically, we aim to minimize the distribution divergence between labeled and unlabeled images, and incorporate this criterion into the objective function of NMF to construct new robust representations. Experiments show that TNMF outperforms state-of-the-art methods on real datasets.

Hierarchical dinucleotide distribution in genome along evolution and its effect on chromatin packing

“…Non-negative matrix factorization (NMF)(Lee and Seung 1999) is a decomposition method for a matrix (i.e. multivariate data) that has been used widely in signal processing, image recognition(Wang et al 2016; Du and Swamy 2019) and computational biology(Devarajan 2008). The aim of non-negative matrix factorization is to reproduce the observed data by combining a limited number of basis components.…”

Hierarchical dinucleotide distribution in genome along evolution and its effect on chromatin packing

“…Nonnegative matrix factorization NMF (Lee & Seung, 1999) is a decomposition method for a matrix (i.e., multivariate data) that has been used widely in signal processing, image recognition (Wang et al, 2016;Du & Swamy, 2019) and computational biology (Devarajan, 2008). The aim of nonnegative matrix factorization is to reproduce the observed data by combining a limited number of basis components.…”