Tensor-based methods for blind signal separation -- Deep neural networks for supervised speech separation/enhancment -- Joined analysis of multiple datasets, data fusion, and related topics -- Advances in nonlinear blind source separation -- Sparse and low rank modeling for acoustic signal processing.

This book constitutes the proceedings of the 12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICS 2015, held in Liberec, Czech Republic, in August 2015. The 61 revised full papers presented – 29 accepted as oral presentations and 32 accepted as poster presentations – were carefully reviewed and selected from numerous submissions. Five special topics are addressed: tensor-based methods for blind signal separation; deep neural networks for supervised speech separation/enhancement; joined analysis of multiple datasets, data fusion, and related topics; advances in nonlinear blind source separation; sparse and low rank modeling for acoustic signal processing.