HyllFuzz uses LLMs to solve constraints and generate modified inputs for hard-to-reach branches in hybrid fuzzing, reporting 31-59% higher branch coverage than CoFuzz, Intriguer and QSYM plus seven new bugs found.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Large Language Model assisted Hybrid Fuzzing
HyllFuzz uses LLMs to solve constraints and generate modified inputs for hard-to-reach branches in hybrid fuzzing, reporting 31-59% higher branch coverage than CoFuzz, Intriguer and QSYM plus seven new bugs found.