Intrinsic Physical Concepts Discovery with Object-Centric Predictive Models
This work addresses the challenge of enabling AI systems to autonomously learn abstract physical concepts, which is crucial for advancing machine intelligence in domains like robotics and simulation, though it builds incrementally on existing object-centric representation methods.
The paper tackles the problem of discovering intrinsic physical concepts like mass and charge from visual observations without supervision, introducing PHYCINE, which infers these concepts and shows they align with real-world properties and improve performance in causal reasoning tasks, achieving better results on the ComPhy benchmark.
The ability to discover abstract physical concepts and understand how they work in the world through observing lies at the core of human intelligence. The acquisition of this ability is based on compositionally perceiving the environment in terms of objects and relations in an unsupervised manner. Recent approaches learn object-centric representations and capture visually observable concepts of objects, e.g., shape, size, and location. In this paper, we take a step forward and try to discover and represent intrinsic physical concepts such as mass and charge. We introduce the PHYsical Concepts Inference NEtwork (PHYCINE), a system that infers physical concepts in different abstract levels without supervision. The key insights underlining PHYCINE are two-fold, commonsense knowledge emerges with prediction, and physical concepts of different abstract levels should be reasoned in a bottom-up fashion. Empirical evaluation demonstrates that variables inferred by our system work in accordance with the properties of the corresponding physical concepts. We also show that object representations containing the discovered physical concepts variables could help achieve better performance in causal reasoning tasks, i.e., ComPhy.