The Acquisition of Physical Knowledge in Generative Neural Networks

2023/10/30 by Luca M. Schulze Buschoff, Eric Schulz, Buschoff, Luca M. Schulze +3
Computer Science · Physics and Astronomy · #Computational Physics and Python Applications #FOS: Biological sciences #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural Networks and Applications #Neurons and Cognition (q-bio.NC) #Statistical Mechanics and Entropy

paper · pdf · doi:10.48550/arxiv.2310.19943

openalex publication_date 2023/10/30 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

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.

Related

• Stochastic Gradient Descent Captures How Children Learn About Physics
• Emulating Human Developmental Stages with Bayesian Neural Networks
• Intrinsic Motivation Driven Intuitive Physics Learning using Deep Reinforcement Learning with Intrinsic Reward Normalization
• From Infants to AI: Incorporating Infant-like Learning in Models Boosts Efficiency and Generalization in Learning Social Prediction Tasks
• The curious case of developmental BERTology: On sparsity, transfer learning, generalization and the brain