Introduced originally to biological studies for gene expression data analysis, non-negative matrix factorization (NMF) and tensor decomposition (TD) have become major tools for analyzing genome-wide data and other biological data. NMF and TD techniques are adept pattern recognition tools to predict sample classes and infer temporal dynamics. The advantage of both NMF and TD in this domain is that they model the simultaneous use of genes or gene products in concurrent biological processes. In contrast, analytical techniques such as SVD and PCA (and their tensor counterparts) that require independence between components often identify components with limited biological interpretability. However, current NMF and TD approaches are still limited by their lack of unique solutions mathematically, their difficulty in escaping local maximum solutions, and their computational complexity. This workshop aims to bring together researchers, students, and practitioners to: 1) discuss recent statistical or computational techniques addressing these limitations of NMF and TD; 2) provide the latest developments in the field, including Bayesian methods for matrix or tensor estimation, sparsity constraints for minimizing mathematical degeneracy, encoding of biological constraints to guide inference; and 3) to present the current, novel, and important applications to high-throughput biological data.