Developmental Chunking

University Of Southern California, Los Angeles CA

Investigators

Abstract

This small grant for exploratory research aims to improve the performance of a bottom-up chunking algorithm by incorporating top-down knowledge. The approach is to evaluate chunks in terms of the operations on chunks - at the level of chunks - that the new chunks make possible. The research is expected to provide an algorithmic foundation for chunking, which is ubiquitous and important in cognition. The impact of this work should be felt broadly in machine learning.