Search papers, labs, and topics across Lattice.
This paper introduces TransFuzz, a novel fuzzing technique that utilizes LLMs to transfer context-aware bug patterns from historical bug reports to semantically related APIs in deep learning libraries, enabling the detection of silent bugs. The approach extracts bug patterns and oracle designs from known buggy APIs, matches APIs based on functionality embeddings, and synthesizes test cases with customized oracles. TransFuzz discovered 79 previously unknown bugs, including 12 confirmed CVEs, across PyTorch, TensorFlow, and MindSpore, demonstrating its effectiveness in proactively identifying silent bugs.
LLMs can proactively unearth silent bugs in deep learning libraries by transferring bug patterns from historical reports to similar APIs, finding 79 previously unknown bugs.
Deep learning (DL) libraries are widely used in critical applications, where even subtle silent bugs can lead to serious consequences. While existing DL fuzzing techniques have made progress in detecting crashes, they inherently struggle to detect silent bugs due to the lack of effective test programs and corresponding oracles. Building on the observation that historical bug reports contain rich, underutilized information about silent bugs, we leverage large language models (LLMs) to perform versatile yet controlled bug transfer for silent bug fuzzing. Specifically, our approach uses LLMs to extract context-aware bug patterns from historical issues, match semantically related Application Programming Interfaces (APIs) using functionality-based embeddings, and synthesize test cases with customized oracles. This enables proactive detection of silent bugs by transferring high-risk contexts and oracle designs from known buggy APIs to functionally similar target APIs. To ensure the reliability of our context-aware bug transfer, we introduce an LLM-powered self-validation module that systematically evaluates the validity of each transferred bug instance. We implement this methodology in a tool named TransFuzz and evaluate it on three mainstream DL libraries: PyTorch, TensorFlow, and MindSpore. TransFuzz successfully discovers 79 previously unknown bugs (12 confirmed as Common Vulnerabilities and Exposures (CVEs)) in 10 bug types, demonstrating its effectiveness and generalizability in migrating DL library bug discovery capabilities.
