Our work on hierarchical RL was accepted to ALA Workshop 2022!

Our paper “Feature Specialization and Clustering Improves Hierarchical Subtask Learning” was accepted for presentation at the Adaptive Learning Agents (ALA) Workshop 2022 (AAMAS)! We show that feature specialization and clustering improve hierarchical reinforcement learning (RL) in various spatial planning tasks, through our Specialized Neurons and Clustering (SNAC) method.

Check out the talk and paper!