Abstract:Non-negative matrix factorization (NMF) approximates a non-negative matrix $X$ by a product of two non-negative low-rank factor matrices $W$ and $H$. NMF and its extensions minimize either the Kullback-Leibler divergence or the Euclidean distance between $X$ and $W^T H$ to model the Poisson noise or the Gaussian noise. In practice, when the noise distribution is heavy tailed, they cannot perform well. This paper presents Manhattan NMF (MahNMF) which minimizes the Manhattan distance between $X$ and $W^T H$ for modeling the heavy tailed Laplacian noise. Similar to sparse and low-rank matrix decompositions, MahNMF robustly estimates the low-rank part and the sparse part of a non-negative matrix and thus performs effectively when data are contaminated by outliers. We extend MahNMF for various practical applications by developing box-constrained MahNMF, manifold regularized MahNMF, group sparse MahNMF, elastic net inducing MahNMF, and symmetric MahNMF. The major contribution of this paper lies in two fast optimization algorithms for MahNMF and its extensions: the rank-one residual iteration (RRI) method and Nesterov's smoothing method. In particular, by approximating the residual matrix by the outer product of one row of W and one row of $H$ in MahNMF, we develop an RRI method to iteratively update each variable of $W$ and $H$ in a closed form solution. Although RRI is efficient for small scale MahNMF and some of its extensions, it is neither scalable to large scale matrices nor flexible enough to optimize all MahNMF extensions. Since the objective functions of MahNMF and its extensions are neither convex nor smooth, we apply Nesterov's smoothing method to recursively optimize one factor matrix with another matrix fixed. By setting the smoothing parameter inversely proportional to the iteration number, we improve the approximation accuracy iteratively for both MahNMF and its extensions.

Upper Bound of Bayesian Generalization Error in Non-negative Matrix Factorization

Bayesian mean-parameterized nonnegative binary matrix factorization

Robust Bayesian Nonnegative Matrix Factorization with Implicit Regularizers

Factor-Bounded Nonnegative Matrix Factorization

Upper Bound of Real Log Canonical Threshold of Tensor Decomposition and its Application to Bayesian Inference

Bilinear Form-Based Non-Negative Matrix Set Factorization

Non-negative Matrix-Set Factorization

Diversified Bayesian Nonnegative Matrix Factorization

Convergence Analysis of Graph Regularized Non-Negative Matrix Factorization

On the Performance of Manhattan Nonnegative Matrix Factorization

Adaptive graph regularized nonnegative matrix factorization for data representation

Error Bound for Compound Wishart Matrices

MahNMF: Manhattan Non-negative Matrix Factorization

Correntropy induced metric based graph regularized non-negative matrix factorization

Kernel Nonnegative Matrix Factorization Without the Curse of the Pre-image - Application to Unmixing Hyperspectral Images

Truncated Cauchy Non-Negative Matrix Factorization.

Automatic Relevance Determination in Nonnegative Matrix Factorization with the β-Divergence

Nonnegative Discriminant Matrix Factorization

Cauchy balanced nonnegative matrix factorization

Optimization and expansion of non-negative matrix factorization

Local Learning Regularized Nonnegative Matrix Factorization.