New hierarchical reinforcement learning paper links value functions and hidden Markov models
The author describes combining an agent-centric general value function construction with their hierarchical hidden Markov model work from 25 years earlier. The caveat: no experiments yet.
TLDR
Announcing “A note on goal-based hierarchical RL” on September 15, 2026, the author said the mini-paper combines an agent-centric general value function (ACGVF) construction with their own work on hierarchical hidden Markov models from 25 years earlier. They cautioned that there were no experiments yet.
Combined views
12.8K
1 Source, first seen 15d ago