AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Generalizing Reinforcement Learning through Fusing Self-Supervised Learning into Intrinsic Motivation

Keyu Wu, Min Wu, Zhenghua Chen, Yuecong Xu, Xiaoli Li

AAAI 2022 · Reinforcement learning › Visual and representation RL · approach: Learned state and action representations · 5 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
add the self-supervised-loss-derived intrinsic reward (ablation)SIM on DMControl color hard tasks · color hardaverage episode reward: 673±85 -> 803±33; 926±6 -> 940±2; 835±8 -> 841±13; 952±10 -> 953±7; 954±20 -> 960±6Improved
add the self-supervised-loss-derived intrinsic reward (ablation)SIM on DMControl video easy tasks · video easyaverage episode reward: 779±65 -> 861±33; 958±15 -> 963±5; 749±55 -> 770±18; 704±184 -> 820±135; 801±34 -> 815±38Mixed
add the self-supervised-loss-derived intrinsic reward (ablation)SIM on DMControl video hard tasks · video hardaverage episode reward: 438±104 -> 459±67; 810±36 -> 827±24; 390±50 -> 367±47; 237±165 -> 287±39; 353±18 -> 362±9Mixed
use SIM rather than the strongest listed baselineSAC-based RL on DMControl color hard tasks · color hardaverage episode reward: 692±68 -> 803±33; 893±12 -> 940±2; 805±28 -> 841±13; 949±19 -> 953±7; 803±72 -> 960±6Improved
use SIM rather than the strongest listed baselineSAC-based RL on DMControl video easy tasks · video easyaverage episode reward: 768±38 -> 861±33; 955±13 -> 963±5; 758±62 -> 770±18; 875±56 -> 820±135; 695±97 -> 815±38Mixed
use SIM rather than the strongest listed baselineSAC-based RL on DMControl video hard tasks · video hardaverage episode reward: 381±72 -> 459±67; 771±83 -> 827±24; 429±64 -> 367±47; 327±100 -> 287±39; 302±41 -> 362±9Mixed

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