Missing values imputation is often evaluated on some similarity measure between actual and imputed data. However, it may be more meaningful to evaluate downstream algorithm performance after than the itself. We describe a straightforward unsupervised algorithm, minimax approach based optimal recovery, derive probabilistic error bounds non-negative matrix factorization (NMF). Under certain geome...

Nonnegative Matrix Factorization (NMF) is a standard tool for data analysis. An important variant is the Sparse NMF problem. A natural measure of sparsity is the L0 norm, however its optimization is NP-hard. Here, we consider a sparsity measure linear in the ratio of the L1 and L2 norms, and propose an efficient algorithm to handle the norm constraints which arise when optimizing this measure. ...

In this paper, we propose to cluster and model network-level traffic states based on a geometrical weighted similarity measure of network-level traffic states and locality preservative non-negative matrix factorization. The geometrical weighted similarity measure makes use of correlation between neighboring roads to describe spatial configurations of global traffic patterns. Based on it, we pro...