Interaction Networks for Learning about Objects, Relations and Physics

TLDR

Reasoning about objects, relations, and physics is central to human intelligence and a key goal of artificial intelligence. The study introduces the interaction network, a model that reasons about object interactions to support dynamical predictions and infer abstract system properties. The interaction network processes graph‑structured inputs, performs object‑ and relation‑centric reasoning akin to a simulation using deep neural networks, and is tested on n‑body, rigid‑body collision, and non‑rigid dynamics tasks. The interaction network accurately simulates trajectories of dozens of objects over thousands of time steps, estimates abstract quantities such as energy, generalizes to varying system configurations, and represents the first general‑purpose, learnable physics engine for reasoning about objects and relations.

Abstract

Reasoning about objects, relations, and physics is central to human intelligence, and a key goal of artificial intelligence. Here we introduce the interaction network, a model which can reason about how objects in complex systems interact, supporting dynamical predictions, as well as inferences about the abstract properties of the system. Our model takes graphs as input, performs object- and relation-centric reasoning in a way that is analogous to a simulation, and is implemented using deep neural networks. We evaluate its ability to reason about several challenging physical domains: n-body problems, rigid-body collision, and non-rigid dynamics. Our results show it can be trained to accurately simulate the physical trajectories of dozens of objects over thousands of time steps, estimate abstract quantities such as energy, and generalize automatically to systems with different numbers and configurations of objects and relations. Our interaction network implementation is the first general-purpose, learnable physics engine, and a powerful general framework for reasoning about object and relations in a wide variety of complex real-world domains.