This talk was about fast computation of low-rank approximations large-scale matrices with the help of tensor networs, in particular tensor train. You can fond the slides here