(Sometimes) Less is More: Mitigating the Complexity of Rule-based Representation for Interpretable Classification
This work addresses the need for model transparency in AI applications where high performance alone is insufficient, though it appears incremental as it builds on existing logic-based networks and regularization techniques.
The authors tackled the problem of interpretability in deep neural networks by adapting a differentiable L0 regularization into a logic-based neural network to reduce the complexity of its interpretable version, the Concept Rule Set, while maintaining performance, achieving results compared to alternative heuristics like Random Binarization.
Deep neural networks are widely used in practical applications of AI, however, their inner structure and complexity made them generally not easily interpretable. Model transparency and interpretability are key requirements for multiple scenarios where high performance is not enough to adopt the proposed solution. In this work, a differentiable approximation of $L_0$ regularization is adapted into a logic-based neural network, the Multi-layer Logical Perceptron (MLLP), to study its efficacy in reducing the complexity of its discrete interpretable version, the Concept Rule Set (CRS), while retaining its performance. The results are compared to alternative heuristics like Random Binarization of the network weights, to determine if better results can be achieved when using a less-noisy technique that sparsifies the network based on the loss function instead of a random distribution. The trade-off between the CRS complexity and its performance is discussed.