AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Among Us: Adversarially Robust Collaborative Perception by Consensus

Yiming Li, Qi Fang, Jiamu Bai, Siheng Chen, Felix Juefei-Xu, Chen Feng

ICCV 2023 · Adversarial robustness and security › Attacks on text and graphs · approach: Latency and driving-perception attacks · 45 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 ROBOSAC consensus sampling instead of no defense under PGD attackaverage-fusion collaborative detector · V2X-Sim scene 8AP@IoU0.5: 39.7 -> 77.3Improved
increase sampling budget from 1 to 7 (ablation)ROBOSAC · V2X-Sim scene 8AP@IoU0.5: 69.8 -> 77.3Improved
even one sampling per frame vs individual perception only (ablation)ROBOSAC · V2X-Sim scene 8AP@IoU0.5: 63.5 -> 69.8Improved
use ROBOSAC instead of PGD adversarial training when attacker switches to C&Wcollaborative detector · V2X-Sim scene 8AP@IoU0.5: 43.2 -> 74.5Improved
ROBOSAC vs PGD adversarial training against PGD attackcollaborative detector · V2X-Sim scene 8AP@IoU0.5: 75.6 -> 77.9Improved
use temporal consistency instead of individual-perception reference (ablation)ROBOSAC dynamic team · V2X-Sim scene 8FPS / AP@0.5: FPS 14.2 -> 19.2; AP 77.5 -> 75.8Mixed
vary consensus threshold epsilon from 0.1 to 0.5 (ablation)ROBOSAC · V2X-Sim scene 8AP@IoU0.5: 77.5 (0.1), 76.1 (0.5)No clear effect
vary attacker type and iterations (PGD, BIM) (ablation)ROBOSAC · V2X-Sim scene 8AP@IoU0.5: 78.3 (PGD 10) to 77.5 (BIM 15)No clear effect
static team (identify attackers once) vs dynamic team (ablation)ROBOSAC · V2X-Sim scene 8FPS: 14.2 -> 36.8Improved

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