Predictions obtained by, e.g., artificial neural networks have a high accuracy but humans often perceive the models as black boxes. Insights about the decision making are mostly opaque for humans. Particularly understanding the decision making in highly sensitive areas such as healthcare or finance, is of paramount importance. The decision-making behind the black boxes requires it to be more transparent, accountable, and understandable for humans. This survey paper provides essential definitions, an overview of the different principles and methodologies of explainable Supervised Machine Learning (SML). We conduct a state-of-the-art survey that reviews past and recent explainable SML approaches and classifies them according to the introduced definitions. Finally, we illustrate principles by means of an explanatory case study and discuss important future directions.

No abstract

No abstract

The field of Explainable Artificial Intelligence (XAI) tries to make learned models more understandable. One type of explanation for such models are counterfactual explanations. Counterfactual explanations explain the decision for a specific instance, the factual, by providing a similar instance which leads to a different decision, the counterfactual. In this work a new approaches around the idea of counterfactuals was developed. It generates a data structure over the feature space of a classification problem to accelerate the search for counterfactuals and augments them with global explanations. The approach maps the feature space by hierarchically dividing it into regions which belong to the same class. It is applicable in any case where predictions can be generated for input data, even without direct access to the model. The framework works well for lower-dimensional problems but becomes unpractical due to high computation times at around 12 to 15 dimensions.

Machine learning and deep learning are widely used in various applications to assist or even replace human reasoning. For instance, a machine learning based intrusion detection system (IDS) monitors a network for malicious activity or specific policy violations. We propose that IDSs should attach a sufficiently understandable report to each alert to allow the operator to review them more efficiently. This work aims at complementing an IDS by means of a framework to create explanations. The explanations support the human operator in understanding alerts and reveal potential false positives. The focus lies on counterfactual instances and explanations based on locally faithful decision-boundaries.