Differentiable Neural Computers

Joerg Franke, KIT

July 25, 2017

Abstract

Differentiable Neural Computers (DNCs) are a very recent new line of deep neural networks that are enhanced by a memory module. This allows them to read and write information for future use, which is considered a major step towards Artificial General Intelligence.

Jörg Franke (Karlsruhe Institute of Technology) employs DNCs for his work in Natural Language Processing. He will introduce us to these exciting new models and share insights on their training and application.

Material

Here's a collection of material if you'd like to familiarize yourself with the concept beforehand:

A blog entry on DNCs by DeepMind: DNCs

DeepMind's Nature publication: DNC paper

DeepMind's DNC github repository: DNC repo

Slides