Explaining Naive Bayes and Other Linear Classifiers with Polynomial Time and Delay
This work addresses the need for efficient and scalable explanation methods in machine learning, particularly for interpretability in classification tasks, but it is incremental as it builds on existing PI-explanation concepts.
The paper tackles the problem of efficiently computing PI-explanations for Naive Bayes and linear classifiers, showing that one explanation can be computed in log-linear time and enumeration can be done with polynomial delay, with experimental results demonstrating performance gains over earlier methods.
Recent work proposed the computation of so-called PI-explanations of Naive Bayes Classifiers (NBCs). PI-explanations are subset-minimal sets of feature-value pairs that are sufficient for the prediction, and have been computed with state-of-the-art exact algorithms that are worst-case exponential in time and space. In contrast, we show that the computation of one PI-explanation for an NBC can be achieved in log-linear time, and that the same result also applies to the more general class of linear classifiers. Furthermore, we show that the enumeration of PI-explanations can be obtained with polynomial delay. Experimental results demonstrate the performance gains of the new algorithms when compared with earlier work. The experimental results also investigate ways to measure the quality of heuristic explanations