Search Based Repair of Deep Neural Networks

Deep Neural Networks (DNNs) are being adopted in various domains, including safety critical ones. The wide-spread adoption also calls for ways to guide the testing of their accuracy and robustness, for which various test adequacy criteria and input generation methods have been recently introduced. In this paper, we explore the natural subsequent step: given an input that reveals unexpected behaviour in a trained DNN, we propose to repair the DNN using input-output pairs as a specification. This paper introduces Arachne, a novel program repair technique for DNNs. Arachne first performs sensitivity based fault localisation to limit the number of neural weights it has to modify. Subsequently, Arachne uses Particle Swarm Optimisation (PSO) to directly optimise the localised neural weights until the behaviour is corrected. An empirical study using three different benchmark datasets shows that Arachne can reduce the instances of the most frequent misclassification type committed by a pre-trained CIFAR-10 classifier by 27.5%, without any need for additional training data. Patches generated by Arachne tend to be more focused on the targeted misbehaviour than DNN retraining, which is more disruptive to non-targeted behaviour. The overall results suggest the feasibility of patching DNNs using Arachne until they can be retrained properly.