Neural Cellular Automata for ARC-AGI
This work addresses the challenge of using self-organizing systems for complex reasoning tasks in AI, but it appears incremental as it applies an existing method to a new domain without claiming major breakthroughs.
The paper tackled the problem of applying Neural Cellular Automata (NCA) to the Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) to test their capabilities in precise transformations and few-shot generalization, finding that gradient-trained NCA models are a promising and efficient approach for abstract grid-based tasks.
Cellular automata and their differentiable counterparts, Neural Cellular Automata (NCA), are highly expressive and capable of surprisingly complex behaviors. This paper explores how NCAs perform when applied to tasks requiring precise transformations and few-shot generalization, using the Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) as a domain that challenges their capabilities in ways not previously explored. Specifically, this paper uses gradient-based training to learn iterative update rules that transform input grids into their outputs from the training examples and apply them to the test inputs. Results suggest that gradient-trained NCA models are a promising and efficient approach to a range of abstract grid-based tasks from ARC. Along with discussing the impacts of various design modifications and training constraints, this work examines the behavior and properties of NCAs applied to ARC to give insights for broader applications of self-organizing systems.