And that interests me because I am to, and have been for years, nay, decades – see, for example, this post from yesterday: Gestalt Switch: From Artificial Intelligence to Artificial Minds. I started out with classical symbolic systems and have worked my way to neural networks. They're going in the opposite direction.
Harshvardhan Gupta has published a useful primer, DeepMind’s Relational Networks — Demystified. That's from July 2017. Why didn't I know about it? Here's the paper Gupta references, A simple neural network module for relational reasoning (PDF):
Abstract: Relational reasoning is a central component of generally intelligent behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
Here's a page where DeepMind lists their work on relational reasoning.
No comments:
Post a Comment