Abstract:Deep Learning systems (DL) based on Deep Neural Networks (DNNs) are more and more used in various aspects of our life, including unmanned vehicles, speech processing, and robotics. However, due to the limited dataset and the dependence on manual labeling data, DNNs often fail to detect their erroneous behaviors, which may lead to serious problems. Several approaches have been proposed to enhance the input examples for testing DL systems. However, they have the following limitations. First, they design and generate adversarial examples from the perspective of model, which may cause low generalization ability when they are applied to other models. Second, they only use surface feature constraints to judge the difference between the adversarial example generated and the original example. The deep feature constraints, which contain high-level semantic information, such as image object category and scene semantics are completely neglected. To address these two problems, in this paper, we propose CAGFuzz, a Coverage-guided Adversarial Generative Fuzzing testing approach, which generates adversarial examples for a targeted DNN to discover its potential defects. First, we train an adversarial case generator (AEG) from the perspective of general data set. Second, we extract the depth features of the original and adversarial examples, and constrain the adversarial examples by cosine similarity to ensure that the semantic information of adversarial examples remains unchanged. Finally, we retrain effective adversarial examples to improve neuron testing coverage rate. Based on several popular data sets, we design a set of dedicated experiments to evaluate CAGFuzz. The experimental results show that CAGFuzz can improve the neuron coverage rate, detect hidden errors, and also improve the accuracy of the target DNN.

Format-aware Learn&Fuzz: Deep Test Data Generation for Efficient Fuzzing

FA-Fuzz: A Novel Scheduling Scheme Using Firefly Algorithm for Mutation-Based Fuzzing

MorFuzz: Fuzzing Processor Via Runtime Instruction Morphing Enhanced Synchronizable Co-simulation.

Data Augmentation by Fuzzing for Neural Test Generation

LAFuzz: Neural Network for Efficient Fuzzing

Optimizing Seed Inputs in Fuzzing with Machine Learning

Not all bytes are equal: Neural byte sieve for fuzzing

NESTFUZZ: Enhancing Fuzzing with Comprehensive Understanding of Input Processing Logic

DRLFCfuzzer: fuzzing with Deep-Reinforcement-Learning under Format Constraints

FuzzInMem: Fuzzing Programs Via In-memory Structures

FuzzCoder: Byte-level Fuzzing Test via Large Language Model

AIFORE: Smart Fuzzing Based on Automatic Input Format Reverse Engineering

FDFuzz: Applying Feature Detection to Fuzz Deep Learning Systems

FormatFuzzer: Effective Fuzzing of Binary File Formats

Reinforcement Learning-Based Fuzzing Technology.

CAGFuzz: Coverage-Guided Adversarial Generative Fuzzing Testing of Deep Learning Systems

A Fuzzing Method for Security Testing of Sensors

High-performance Directional Fuzzing Scheme Based on Deep Reinforcement Learning

SpeedNeuzz: Speed Up Neural Program Approximation with Neighbor Edge Knowledge

Understanding Programs by Exploiting (Fuzzing) Test Cases

Deep Learning-Based Hybrid Fuzz Testing