Misbehaviour prediction for autonomous driving systems
Andrea Stocco, Michael Weiß, Marco Calzana, Paolo Tonella
Abstract
Deep Neural Networks (DNNs) are the core component of modern autonomous driving systems. To date, it is still unrealistic that a DNN will generalize correctly to all driving conditions. Current testing techniques consist of offline solutions that identify adversarial or corner cases for improving the training phase.
Topics & Concepts
Adversarial systemComputer scienceComponent (thermodynamics)Core (optical fiber)Artificial neural networkArtificial intelligenceDeep neural networksMachine learningTelecommunicationsThermodynamicsPhysicsAdversarial Robustness in Machine LearningAnomaly Detection Techniques and ApplicationsAutonomous Vehicle Technology and Safety