Cophy: Counterfactual Learning of Physical Dynamics

Primary tabs

Citation Author(s):
Fabien
Baradel
INSA Lyon, LIRIS
Natalia
Neverova
Facebook AI Research
Julien
Mille
INSA Val de Loire, LIFAM
Greg
Mori
SFU, Borealis AI
Christian
Wolf
INSA Lyon, LIRIS
Submitted by:
Fabien Baradel
Last updated:
Mon, 01/27/2020 - 10:07
DOI:
10.21227/ps5q-8m55
Links:
License:
55 Views
Categories:
0
0 ratings - Please login to submit your rating.

Abstract 

Abstract: 

Understanding causes and effects in mechanical systems is an essential component of reasoning in the physical world. This work poses a new problem of counterfactual learning of object mechanics from visual input. We develop the COPHY benchmark to assess the capacity of the state-of-the-art models for causal physical reasoning in a synthetic 3D environment and propose a model for learning the physical dynamics in a counterfactual setting. Having observed a mechanical experiment that involves, for example, a falling tower of blocks, a set of bouncing balls or colliding objects, we learn to predict how its outcome is affected by an arbitrary intervention on its initial conditions, such as displacing one of the objects in the scene. The alternative future is predicted given the altered past and a latent representation of the confounders learned by the model in an end-to-end fashion with no supervision. We compare against feedforward video prediction baselines and show how observing alternative experiences allows the network to capture latent physical properties of the environment, which results in significantly more accurate predictions at the level of super human performance.

Instructions: 

Please follow the instructions given in the github repo: https://github.com/fabienbaradel/cophy

s