Colin Paterson

Naif Alasmari, Radu Calinescu, Colin Paterson et al.

Stochastic models are widely used to verify whether systems satisfy their reliability, performance and other nonfunctional requirements. However, the validity of the verification depends on how accurately the parameters of these models can be estimated using data from component unit testing, monitoring, system logs, etc. When insufficient data are available, the models are affected by epistemic parametric uncertainty, the verification results are inaccurate, and any engineering decisions based on them may be invalid. To address these problems, we introduce VERACITY, a tool-supported iterative approach for the efficient and accurate verification of nonfunctional requirements under epistemic parameter uncertainty. VERACITY integrates confidence-interval quantitative verification with a new adaptive uncertainty reduction heuristic that collects additional data about the parameters of the verified model by unit-testing specific system components over a series of verification iterations. VERACITY supports the quantitative verification of discrete-time Markov chains, deciding which components are to be tested in each iteration based on factors that include the sensitivity of the model to variations in the parameters of different components, and the overheads (e.g., time or cost) of unit-testing each of these components. We show the effectiveness and efficiency of VERACITY by using it for the verification of the nonfunctional requirements of a tele-assistance service-based system and an online shopping web application.

Colin Paterson, Haoze Wu, John Grese et al.

We introduce DeepCert, a tool-supported method for verifying the robustness of deep neural network (DNN) image classifiers to contextually relevant perturbations such as blur, haze, and changes in image contrast. While the robustness of DNN classifiers has been the subject of intense research in recent years, the solutions delivered by this research focus on verifying DNN robustness to small perturbations in the images being classified, with perturbation magnitude measured using established Lp norms. This is useful for identifying potential adversarial attacks on DNN image classifiers, but cannot verify DNN robustness to contextually relevant image perturbations, which are typically not small when expressed with Lp norms. DeepCert addresses this underexplored verification problem by supporting:(1) the encoding of real-world image perturbations; (2) the systematic evaluation of contextually relevant DNN robustness, using both testing and formal verification; (3) the generation of contextually relevant counterexamples; and, through these, (4) the selection of DNN image classifiers suitable for the operational context (i)envisaged when a potentially safety-critical system is designed, or (ii)observed by a deployed system. We demonstrate the effectiveness of DeepCert by showing how it can be used to verify the robustness of DNN image classifiers build for two benchmark datasets (`German Traffic Sign' and `CIFAR-10') to multiple contextually relevant perturbations.

Colin Paterson, Radu Calinescu, Chiara Picardi

Regions of high-dimensional input spaces that are underrepresented in training datasets reduce machine-learnt classifier performance, and may lead to corner cases and unwanted bias for classifiers used in decision making systems. When these regions belong to otherwise well-represented classes, their presence and negative impact are very hard to identify. We propose an approach for the detection and mitigation of such rare subclasses in deep neural network classifiers. The new approach is underpinned by an easy-to-compute commonality metric that supports the detection of rare subclasses, and comprises methods for reducing the impact of these subclasses during both model training and model exploitation. We demonstrate our approach using two well-known datasets, MNIST's handwritten digits and Kaggle's cats/dogs, identifying rare subclasses and producing models which compensate for subclass rarity. In addition we demonstrate how our run-time approach increases the ability of users to identify samples likely to be misclassified at run-time.