In recent years, many predictive models have been successfully applied to predictive process monitoring, enabling tasks such as predicting the next activity, remaining time, or the future state of a process instance (case). However, recent developments have shown the vulnerability of these models to adversarial attacks, causing algorithms to make incorrect predictions. This paper addresses this issue by leveraging adversarial examples to evaluate the predictive performance of predictive process monitoring models in the face of adversarial threats. Although augmenting training data with adversarial examples has proven effective in defending against specific adversarial attacks, it often remains insufficient in mitigating vulnerabilities to other types of attacks. Our proposed approach explores the use of manifold learning techniques to restrict these examples within the range of data on which the model is trained. By learning from these specifically engineered (hidden) attacks, we seek to develop models that maintain accuracy on new, unseen data while effectively improving adversarial robustness against potential threats.

As machine and deep learning models are increasingly leveraged in predictive process monitoring, the focus has shifted towards making these models explainable. The successful adoption of a model is dependent on whether decision-makers can trust the predictions and explanations made. However, recent studies have shown that deep learning models are vulnerable to adversarial attacks -small perturbations to the inputs-which trick deep learning algorithms into making incorrect predictions. An additional crucial property is that the explanations are robust against these adversarial attacks when the model decision was not affected. Therefore, this paper introduces a robustness assessment framework by investigating the impact of adversarial attacks on the robustness of predictive accuracy and explanations used in the field of predictive process monitoring. First, adversarial examples of cases in the independent test set are generated to examine the robustness of the predictive model against intentionally manipulated data. Next, the predictive models are compared with similar models trained on data imputed with adversarial attacks. We monitor the impact on predictive performance in terms of AUC at different stages of the case execution. Finally, the robustness of the explanations is calculated as the distance between the original explanations and the explanations extracted from the model trained on attacked data. We test multiple machine and deep learning techniques, namely the transparent logistic regression, random forests with Shapley values, and LSTM neural networks with attention. Results show that especially neural networks suffer from adversarial attacks, and the former two are mostly robust in terms of both predictive accuracy and explanations.