3 Hits in 1.4 sec

MTFuzz: Fuzzing with a Multi-Task Neural Network [article]

Dongdong She, Rahul Krishna, Lu Yan, Suman Jana, Baishakhi Ray
2020 arXiv   pre-print
In this paper, we address these issues by using a Multi-Task Neural Network that can learn a compact embedding of the input space based on diverse training samples for multiple related tasks (i.e., predicting  ...  Our results show that MTFuzz uncovers 11 previously unseen bugs and achieves an average of 2x more edge coverage compared with 5 state-of-the-art fuzzer on 10 real-world programs.  ...  We demonstrate that MTFuzz can transfer a NN learnt on one program to other similar programs. CONCLUSION This paper presents MTFuzz, a multi-task neural-network fuzzing framework.  ... 
arXiv:2005.12392v1 fatcat:46luwyy3mja5naldfaz2lkcula

Refined Grey-Box Fuzzing with SIVO [article]

Ivica Nikolic and Radu Mantu and Shiqi Shen and Prateek Saxena
2021 arXiv   pre-print
We design and implement from scratch a new fuzzer called SIVO that refines multiple stages of grey-box fuzzing.  ...  First, SIVO refines data-flow fuzzing in two ways: (a) it provides a new taint inference engine that requires only logarithmic in the input size number of tests to infer the dependency of all program branches  ...  MTFuzz [30] trains a multiple-task neural network to infer the relationship between program inputs and different kinds of edge coverage to guide input mutation.  ... 
arXiv:2102.02394v2 fatcat:3zwqfzx53jhmjmhyakl3sjqtce

Fine Grained Dataflow Tracking with Proximal Gradients [article]

Gabriel Ryan, Abhishek Shah, Dongdong She, Koustubha Bhat, Suman Jana
2021 arXiv   pre-print
Dataflow tracking with Dynamic Taint Analysis (DTA) is an important method in systems security with many applications, including exploit analysis, guided fuzzing, and side-channel information leak detection  ...  We introduce proximal gradient analysis (PGA), a novel, theoretically grounded approach that can track more accurate and fine-grained dataflow information.  ...  NEUZZ, MTFuzz and Neutaint train neural networks to predict program branch behavior and use the network's gradients to guide the mutation algorithm [43] [44] [45] .  ... 
arXiv:1909.03461v6 fatcat:2kuubtwdfbadpj7r5j37rz4gp4