Modeling and Discovering Direct Causes for Predictive Models
This work addresses the need for interpretable and reliable predictive models, which is crucial for data collection and evaluation in various domains, though it appears incremental in building on existing causal discovery methods.
The paper tackles the problem of identifying which input features directly cause predictions in machine learning models, introducing a causal modeling framework and algorithms that theoretically and empirically accelerate this discovery process.
We introduce a causal modeling framework that captures the input-output behavior of predictive models (e.g., machine learning models). The framework enables us to identify features that directly cause the predictions, which has broad implications for data collection and model evaluation. We then present sound and complete algorithms for discovering direct causes (from data) under some assumptions. Furthermore, we propose a novel independence rule that can be integrated with the algorithms to accelerate the discovery process, as we demonstrate both theoretically and empirically.