Do Machine Learning Models Learn Statistical Rules Inferred from Data?
This addresses the issue of hidden errors in ML models for practitioners, offering a method to improve model coherence, though it is incremental as it builds on existing logic-based and statistical techniques.
The paper tackles the problem of machine learning models making errors that violate statistical rules inferred from data, proposing a framework called SQRL to derive these rules without supervision and adapt models at test time to reduce violations. The result shows that SQRL uncovers up to 158K rule violations by state-of-the-art models and reduces them by up to 68.7% with performance improvements up to 32%.
Machine learning models can make critical errors that are easily hidden within vast amounts of data. Such errors often run counter to rules based on human intuition. However, rules based on human knowledge are challenging to scale or to even formalize. We thereby seek to infer statistical rules from the data and quantify the extent to which a model has learned them. We propose a framework SQRL that integrates logic-based methods with statistical inference to derive these rules from a model's training data without supervision. We further show how to adapt models at test time to reduce rule violations and produce more coherent predictions. SQRL generates up to 300K rules over datasets from vision, tabular, and language settings. We uncover up to 158K violations of those rules by state-of-the-art models for classification, object detection, and data imputation. Test-time adaptation reduces these violations by up to 68.7% with relative performance improvement up to 32%. SQRL is available at https://github.com/DebugML/sqrl.