Beyond Random Inputs: A Novel ML-Based Hardware Fuzzing

Paper

First seen 6/24/2026

Last seen 6/24/2026

Evidence 6 chunks

NEIGHBORHOOD

No graph connections found for this entity yet. It may appear in future ingestion runs.

8 connections

PPO uses → 88% 2e

PPO is used during RL-based training stages.

ISA Disassembler uses → 90% 2e

A disassembler is leveraged to penalize invalid instruction generations.

Coverage feedback uses → 93% 2e

Coverage information guides the optimization of generated inputs.

ChatFuzz evaluates → 92% 2e

It measures condition coverage and time to coverage on RocketCore and BOOM.

TheHuzz evaluates → 90% 2e

It reports TheHuzz’s coverage and the time required to reach target coverage.

RL optimizes input generation through coverage-based rewards.

LLMs are central to the proposed ML-based fuzzing approach.

ChatFuzz introduces → 98% 1e

Authors present ChatFuzz as a new ML-based hardware/processor fuzzer.