October 16, 2014
Brad Hayes presented the paper, “A Developmentally Inspired Transfer Learning Approach to Predicting Skill Durations”, co-authored with labmates Corina Grigore, Alex Litoiu, and Aditi Ramachandran at ICDL 2014, held in Genoa, Italy. This work proposed the decomposition of observed skill durations into multiple signals that could be modeled, including task-specific procedural experience, tool proficiencies, and motor skill proficiencies to achieve better bounds in multi-agent planning scenarios.