Forward Prediction for Physical Reasoning
This addresses physical reasoning for AI systems, but it is incremental as it builds on existing benchmarks and models.
The study tackled the problem of physical reasoning by evaluating forward-prediction models on the PHYRE benchmark, finding that they improve performance on complex tasks but struggle with generalization to new templates, and set a new state-of-the-art.
Physical reasoning requires forward prediction: the ability to forecast what will happen next given some initial world state. We study the performance of state-of-the-art forward-prediction models in the complex physical-reasoning tasks of the PHYRE benchmark. We do so by incorporating models that operate on object or pixel-based representations of the world into simple physical-reasoning agents. We find that forward-prediction models can improve physical-reasoning performance, particularly on complex tasks that involve many objects. However, we also find that these improvements are contingent on the test tasks being small variations of train tasks, and that generalization to completely new task templates is challenging. Surprisingly, we observe that forward predictors with better pixel accuracy do not necessarily lead to better physical-reasoning performance.Nevertheless, our best models set a new state-of-the-art on the PHYRE benchmark.