Timezone: »

 
Spotlight
A simple neural network module for relational reasoning
Adam Santoro · David Raposo · David Barrett · Mateusz Malinowski · Razvan Pascanu · Peter Battaglia · Timothy Lillicrap

Wed Dec 06 03:25 PM -- 03:30 PM (PST) @ Hall A
in Reinforcement Learning, Deep Learning »

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 dynamical 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. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.

Author Information

Adam Santoro (DeepMind)
David Raposo (DeepMind)
David Barrett (DeepMind)
Mateusz Malinowski (DeepMind)

Mateusz Malinowski is a research scientist at DeepMind, where he works at the intersection of computer vision, natural language understanding, and deep learning. He was granted PhD (Dr.-Ing.) with the highest honor (summa cum laude) at Max Planck Institute for Informatics in 2017 in computer vision for his pioneering work on visual question answering, where he proposed the task and developed methods that answer questions about the content of images. Prior to this, he graduated with honors from Saarland University in computer science. Before that, he studied computer science at Wroclaw University in Poland.

Razvan Pascanu (Google DeepMind)
Peter Battaglia (DeepMind)
Timothy Lillicrap (Google DeepMind)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors