AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Social GAN: Socially Acceptable Trajectories With Generative Adversarial Networks

CVPR 2018 · Probabilistic and scientific modeling › Molecules and proteins · approach: Multimodal trajectory prediction · 2,462 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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
Train with variety loss and use multiple sampled trajectories rather than a single sample (ablation)SGAN · ETH, HOTEL, UNIV, ZARA1, ZARA2 (AVG)ADE: 0.49 / 0.74 -> 0.39 / 0.58Helped
Draw more test samples without training with variety loss (ablation)SGAN · ETH, HOTEL, UNIV, ZARA1, ZARA2 (AVG)ADE: 0.49 / 0.74 -> 0.45 / 0.67No clear effect
Use the proposed pooling module (ablation)SGAN-20V-20 · ETH, HOTEL, UNIV, ZARA1, ZARA2 (AVG)ADE: 0.39 / 0.58 -> 0.41 / 0.61Hurt
Use the proposed pooling module (ablation)SGAN · Synthetic collision-avoidance scenariossocially acceptable collision avoidanceHelped
Train and evaluate with multiple samples but without variety loss (ablation)SGAN · ETH and UCY datasetsADENo clear effect
Increase the variety-loss sample count to k=100 (ablation)SGAN · ETH and UCY datasetsprediction accuracy: 33%Helped
Pool context once rather than at every decoder time step (ablation)SGAN-P · ETH and UCY datasetsspeed: 16xHelped

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