DiffGAN: A Test Generation Approach for Differential Testing of Deep Neural Networks for Image Analysis

Keyphrases

• Accuracy-based 16%
• Autonomous Driving 16%
• Behavioral Difference 16%
• Black-box Testing 16%
• Classification Datasets 16%
• Deep Neural Network 100%
• Differential Testing 100%
• Divergence 16%
• Fitness Function 16%
• Generation Approach 100%
• Generative Adversarial Networks 50%
• High Diversity 16%
• Image Analysis 100%
• Image Classification 33%
• Image Classification Tasks 16%
• Input Characteristics 16%
• Input Space 33%
• Internal Structure 16%
• Machine Learning Models 16%
• Model Output 33%
• Model Selection 16%
• Non-dominated Sorting Genetic Algorithm (NSGA-II) 16%
• Order of Accuracy 16%
• Seed Output 16%
• Selection Mechanism 16%
• Smart Model 16%
• Specific Features 16%
• Test Generation 100%
• Test Input 16%
• Testing Approaches 16%
• Voting System 16%
• White Box Method 16%

Computer Science

• Alternative Model 33%
• Application Area 16%
• Autonomous Driving 16%
• Classification Task 16%
• Deep Neural Network 100%
• Fitness Function 16%
• Generalizability 16%
• Generative Adversarial Networks 50%
• Image Analysis 100%
• Image Classification 50%
• Internal Model 16%
• Internal Structure 16%
• Learning System 16%
• Machine Learning 16%
• Model Selection 16%
• Neural Network Model 66%
• Non-Dominated Sorting Genetic Algorithm II 16%
• Test Generation 100%