AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Temporal Abstraction in Reinforcement Learning with the Successor Representation

Marlos C. Machado, Andre Barreto, Doina Precup, Michael Bowling

JMLR 2023 · Reinforcement learning › Visual and representation RL · approach: Successor features for transfer · 24 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use eigenoptions (top several SR eigenvectors, broad initiation) instead of covering optionsQ-learning with epsilon-greedy exploration · Four-room (reward maximization)learning speedImproved
add covering options to the agent's action setQ-learning with epsilon-greedy exploration · Four-room (reward maximization)learning speedNo clear effect
use more than one eigenvector per iteration (eigenoptions rather than covering options) (ablation)option discovery from the SR · Four-room (closed form, diffusion time)diffusion timeImproved
give options a broad initiation set instead of a single state (ablation)covering options · Four-room (closed form, diffusion time)diffusion timeMixed
estimate the SR online with TD learning instead of closed formeigenoptions · Four-room (online, diffusion time)diffusion timeNo clear effect
estimate the SR online with TD learning instead of closed formcovering options · Four-room (online, diffusion time)average diffusion timeHurt
add covering options (even a single option) to primitive actionsrandom walk · Four-room (closed form, median diffusion time)median diffusion timeImproved
add few eigenoptions (small option budget)random walk · Four-room (closed form, average diffusion time)average diffusion timeHurt
use SR eigenvectors instead of proto-value functions for covering options (ablation)covering options · Four-room (appendix)diffusion timeNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy