Learning Patterns and Abstractions from Perceptual Sequences
A new research paper explores how humans and models learn from sequences by breaking them into smaller parts, a process called chunking. The research proposes chunking as a rational strategy for discovering recurring patterns and nested hierarchies, enabling efficient sequence factorization. The paper also introduces a model that learns both chunks and abstract variables, uncovering invariant symbolic patterns and showing similarities to human learning. AI
IMPACT Proposes a new computational principle for structured knowledge acquisition in sequences, potentially influencing future AI model architectures.