This paper, co-authored by Lichao Wu, Mohamadreza Rostami, Huimin Li, and Ahmad-Reza Sadeghi, tackles the growing challenge of securing increasingly complex hardware systems—where traditional verification methods often fall short.

What makes HFL unique?
- Leverages Long Short-Term Memory (LSTM) models to understand test case semantics and predict hardware coverage.
- Applies Reinforcement Learning to dynamically enhance test generation strategies within the fuzzing loop.
- Successfully identified all known vulnerabilities and discovered four previously unknown high-severity issues across three RISC-V cores—using less than 1% of test cases required by leading fuzzers.

HFL marks a powerful step forward in intelligent, efficient hardware vulnerability detection.

Congratulations to the entire team for this impressive work!

For more details, please visit https://www.date-conference.com