Publications

This report details the results of a three-fold investigation of sensitivity analysis (SA) for machine learning (ML) explainability (MLE): (1) the mathematical assessment of the fidelity of an explanation with respect to a learned ML model, (2) quantifying the trustworthiness of a prediction, and (3) the impact of MLE on the efficiency of end-users through multiple users studies. We focused on the cybersecurity domain as the data is inherently non-intuitive. As ML is being using in an increasing number of domains, including domains where being wrong can elicit high consequences, MLE has been proposed as a means of generating trust in a learned ML models by end users. However, little analysis has been performed to determine if the explanations accurately represent the target model and they themselves should be trusted beyond subjective inspection. Current state-of-the-art MLE techniques only provide a list of important features based on heuristic measures and/or make certain assumptions about the data and the model which are not representative of the real-world data and models. Further, most are designed without considering the usefulness by an end-user in a broader context. To address these issues, we present a notion of explanation fidelity based on Shapley values from cooperative game theory. We find that all of the investigated MLE explainability methods produce explanations that are incongruent with the ML model that is being explained. This is because they make critical assumptions about feature independence and linear feature interactions for computational reasons. We also find that in deployed, explanations are rarely used due to a variety of reason including that there are several other tools which are trusted more than the explanations and there is little incentive to use the explanations. In the cases when the explanations are used, we found that there is the danger that explanations persuade the end users to wrongly accept false positives and false negatives. However, ML model developers and maintainers find the explanations more useful to help ensure that the ML model does not have obvious biases. In light of these findings, we suggest a number of future directions including developing MLE methods that directly model non-linear model interactions and including design principles that take into account the usefulness of explanations to the end user. We also augment explanations with a set of trustworthiness measures that measure geometric aspects of the data to determine if the model output should be trusted.

Abstract not provided.

Abstract not provided.

The impact of machine learning (ML) explanations and different attributes of explanations on human performance was investigated in a simulated spam detection task. Participants decided whether the metadata presented about an email indicated that it was spam or benign. The task was completed with the aid of a ML model. The ML model’s prediction was displayed on every trial. The inclusion of an explanation and, if an explanation was presented, attributes of the explanation were manipulated within subjects: the number of model input features (3, 7) and visualization of feature importance values (graph, table), as was trial type (i.e., hit, false alarm). Overall model accuracy (50% vs 88%) was manipulated between subjects, and user trust in the model was measured as an individual difference metric. Results suggest that a user’s trust in the model had the largest impact on the decision process. The users showed better performance with a more accurate model, but no differences in accuracy based on number of input features or visualization condition. Rather, users were more likely to detect false alarms made by the more accurate model; they were also more likely to comply with a model “miss” when more model explanation was provided. Finally, response times were longer in individuals reporting low model trust, especially when they did not comply with the model’s prediction. Our findings suggest that the factors impacting the efficacy of ML explanations depends, minimally, on the task, the overall model accuracy, the likelihood of different model errors, and user trust.

Abstract not provided.

Abstract not provided.