Mattia Sabella, Alberto Archetti, Pietro Pinoli, Matteo Matteucci, Cinzia Cappiello
QuAIL is a new training mechanism that improves machine learning model performance on corrupted tabular data by using feature reliability information to guide learning.
Machine learning models often struggle when trained on data that has errors, missing values, or biases. QuAIL is a new approach that helps models learn better from such corrupted data by using information about which data features are more reliable. Instead of fixing the data or giving different importance to each data point, QuAIL adjusts the learning process itself based on the reliability of each feature. Tests on various datasets show that QuAIL improves model performance, especially when there is limited data or systematic biases.
