The Acquisition of Physical Knowledge in Generative Neural Networks

Luca M. Schulze Buschoff, Eric Schulz, Marcel Binz
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:30321-30341, 2023.

Abstract

As children grow older, they develop an intuitive understanding of the physical processes around them. Their physical understanding develops in stages, moving along developmental trajectories which have been mapped out extensively in previous empirical research. Here, we investigate how the learning trajectories of deep generative neural networks compare to children’s developmental trajectories using physical understanding as a testbed. We outline an approach that allows us to examine two distinct hypotheses of human development – stochastic optimization and complexity increase. We find that while our models are able to accurately predict a number of physical processes, their learning trajectories under both hypotheses do not follow the developmental trajectories of children.

Cite this Paper


BibTeX
@InProceedings{pmlr-v202-schulze-buschoff23a, title = {The Acquisition of Physical Knowledge in Generative Neural Networks}, author = {Schulze Buschoff, Luca M. and Schulz, Eric and Binz, Marcel}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {30321--30341}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/schulze-buschoff23a/schulze-buschoff23a.pdf}, url = {https://proceedings.mlr.press/v202/schulze-buschoff23a.html}, abstract = {As children grow older, they develop an intuitive understanding of the physical processes around them. Their physical understanding develops in stages, moving along developmental trajectories which have been mapped out extensively in previous empirical research. Here, we investigate how the learning trajectories of deep generative neural networks compare to children’s developmental trajectories using physical understanding as a testbed. We outline an approach that allows us to examine two distinct hypotheses of human development – stochastic optimization and complexity increase. We find that while our models are able to accurately predict a number of physical processes, their learning trajectories under both hypotheses do not follow the developmental trajectories of children.} }
Endnote
%0 Conference Paper %T The Acquisition of Physical Knowledge in Generative Neural Networks %A Luca M. Schulze Buschoff %A Eric Schulz %A Marcel Binz %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-schulze-buschoff23a %I PMLR %P 30321--30341 %U https://proceedings.mlr.press/v202/schulze-buschoff23a.html %V 202 %X As children grow older, they develop an intuitive understanding of the physical processes around them. Their physical understanding develops in stages, moving along developmental trajectories which have been mapped out extensively in previous empirical research. Here, we investigate how the learning trajectories of deep generative neural networks compare to children’s developmental trajectories using physical understanding as a testbed. We outline an approach that allows us to examine two distinct hypotheses of human development – stochastic optimization and complexity increase. We find that while our models are able to accurately predict a number of physical processes, their learning trajectories under both hypotheses do not follow the developmental trajectories of children.
APA
Schulze Buschoff, L.M., Schulz, E. & Binz, M.. (2023). The Acquisition of Physical Knowledge in Generative Neural Networks. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:30321-30341 Available from https://proceedings.mlr.press/v202/schulze-buschoff23a.html.

Related Material