AlphaDog: No-Box Camouflage Attacks via Alpha Channel Oversight
Qi Xia
Network and Distributed System Security (NDSS) Symposium 2025 · Day 2 · ML Security
This article delves into "AlphaDog," a groundbreaking adversarial attack presented by Qi Xia at the NDSS Symposium. AlphaDog introduces the concept of a **no-box camouflage attack**, a novel method that exploits a fundamental oversight in how computer vision (CV) models process images. Unlike traditional adversarial attacks that require intensive queries, iterative processes, or model-specific tuning, AlphaDog operates with zero queries, is universally adapted across various models, and guarantees 100% confidence and attack success rates.
AI review
Genuinely novel attack surface — exploiting alpha channel stripping as a no-box adversarial vector is a clean, original insight that nobody in the CV security space had formalized before. The math is simple enough to be embarrassing in retrospect, which is the hallmark of a good finding. Slight reservations on threat model realism and defense completeness, but this is real research that will age well.