AutoFR - Automated Filter Rule Generation for Adblocking
Explore a groundbreaking conference talk from USENIX Security '23 that introduces AutoFR, an innovative reinforcement learning framework designed to automate the creation and evaluation of filter rules for adblocking. Learn how this system utilizes multi-arm bandits to generate effective filter rules, balancing ad blocking capabilities with minimal visual breakage. Discover the efficiency and effectiveness of AutoFR as it generates rules in minutes, blocks 86% of ads compared to EasyList's 87%, and demonstrates strong generalization to new sites. Gain insights into how this technology could revolutionize the adblocking community by enabling large-scale filter rule generation and potentially assisting filter list authors in their curation efforts.