CORTEXA
← Browse

Alexander Greenhalgh

1 paper indexed

arxivcs.CVcs.AI2026-07-19

ALLUDE: A Unified Evaluation System for Configurable Attacks in Differentiable Environments

Mansi Phute, Alexander Greenhalgh, Matthew Hull, Haoran Wang, Alec Helbling, ShengYun Peng, et al.

Adversarial attacks against vision models like object detectors are often evaluated under limited conditions, leaving their performance under-characterized. Bridging simulation and differentiable rendering enables more robust, end-to-end evaluation of these adversarial attacks, y…

View free PDFSource page