AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Soft Expert Reward Learning for Vision-and-Language Navigation

Hu Wang, Qi Wu, Chunhua Shen

ECCV 2020 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Vision-language navigation agents · 43 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 Soft Expert Distillation module (ablation)EnvDrop-based agent · R2R Val UnseenSR: 0.49 -> 0.52Improved
add Soft Expert Distillation module (ablation)EnvDrop-based agent · R2R Val UnseenSPL: 0.45 -> 0.48Improved
add Self Perceiving module (ablation)EnvDrop-based agent · R2R Val UnseenSR: 0.49 -> 0.53Improved
add Self Perceiving module (ablation)EnvDrop-based agent · R2R Val UnseenSPL: 0.45 -> 0.46Improved
add SP module on top of SED (ablation)SERL · R2R Val UnseenSR: 0.52 -> 0.56Improved
add SED module on top of SP (ablation)SERL · R2R Val UnseenSR: 0.53 -> 0.56Improved
add both SED and SP modules (ablation)EnvDrop-based agent · R2R Val SeenSR: 0.66 -> 0.69Improved
add both SED and SP modules (ablation)EnvDrop-based agent · R2R Val UnseenNE: 5.49 -> 4.74Improved
use beam search at inference (ablation)SERL · R2R Val UnseenSR: 0.56 -> 0.71Improved
use beam search at inference (ablation)SERL · R2R Val UnseenSPL: 0.48 -> 0.02Hurt
use beam search at inference (ablation)SERL · R2R Val SeenSR: 0.69 -> 0.77Improved
vary reward trade-off weights alpha and beta (ablation)SERL · R2R Val UnseenSRNo 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