Last week, I have had the opportunity to participate in the winter school on tensors followed by the Workshop on Tensor Decompositions and Applications (TDA2016). The winter school was very interesting and revealing, in the sense that it served a very smooth yet comprehensive introduction to the basics of tensor decomposition methods, followed by its applications in signal analysis. With the hands-on exercises using TensorLab, we were able to solve some real problems of signal separation, and understand why the underlying tensor methods work for providing a good solution. Though there was little interest on sparse tensor decompositions in overall, I believe with the emerging applications of tensors on big data analysis and machine learning problems this situation is likely to change.
The content of the conference was very rich both in theory and applications, yet the majority of the applications were on signal separation problems using fMRI and ECG. On the computational side, there were not a lot of discussions; I believe this is mostly due to the size of the problems being not too big in general to require sophisticated computational techniques. I hope that with the increasing interest on tensor methods in various fields of applications, the next TDA will be hosted much sooner than usual.