BertRLFuzzer: A BERT and Reinforcement Learning Based Fuzzer

We present a novel tool BertRLFuzzer, a BERT and Reinforcement Learning (RL) based fuzzer aimed at finding security vulnerabilities for Web applications. BertRLFuzzer works as follows: given a set of seed inputs, the fuzzer performs grammar-adhering and attack-provoking mutation operations on them to generate candidate attack vectors. The key insight of BertRLFuzzer is the use of RL with a BERT model as an agent to guide the fuzzer to efficiently learn grammar-adhering and attack-provoking mutation operators. In order to establish …

applications attack attack vectors bert cs.cr cs.lg cs.se fuzzer inputs insight key novel operations security seed the key tool vulnerabilities web web applications