The quality of Machine Learning models strongly depends on the quality of the input data, as such generating high-quality features is often required to improve the predictive accuracy. This process is referred to as Feature Engineering. However, manual Feature Engineering is time-consuming and requires case-specific domain knowledge. In addition, with the proliferation of ML-powered systems, especially in critical contexts, the need for interpretability and explainability becomes increasingly important. To tackle this problem, we introduce SMART, an automated Feature Engineering approach that uses semantic technologies to guide the generation of interpretable features through a two-step process: Exploitation and Exploration. The former uses a reasoning algorithm over a Knowledge Graph (KG) to infer domain-specific features, while the latter exploits the KG to conduct a guided exploration of the search space through Deep Reinforcement Learning. Our experiments on public datasets demonstrate that SMART significantly improves prediction accuracy while ensuring a high level of interpretability. This work highlights the potential of combining KGs and reasoning within Feature Engineering, paving the way for interpretable ML models.
Bouadi et al. (Fri,) studied this question.