How to put algorithms into neural networks? (2019) [video]

Recorded at the ML in PL 2019 Conference, the University of Warsaw, 22-24 November 2019.

Anton Osokin (Higher School of Economics, Moscow), https://aosokin.github.io/

Slides available at https://docs.mlinpl.org/conference/2019/slides/anton_osokin_...

Abstract: Recently, deep neural nets have shown amazing results in such fields as computer vision, natural language processing, etc. To build such networks, we usually use layers from a relatively small dictionary of available modules (fully-connected, convolutional, recurrent, etc.). Being restricted with this set of modules complicates transferring technology to new tasks. On the other hand, many important applications already have a long history and successful algorithmic solutions. Is it possible to use existing methods to construct better networks? In this talk, we will cover several ways of putting algorithms into networks and discuss their pros and cons. Specifically, we will touch using optimization algorithms as structured pooling, unrolling of algorithm iterations into network layers and direct differentiation of the output w.r.t. the input. We will illustrate these approaches on applications from structured-output prediction and computer vision.