Theory And Rationale Of Interpretable All-in-one Pattern Discovery And Disentanglement System

Machine Learning (ML) has been used for decades to obtain associative and discriminative information from relational datasets. However, existing ML models are not able to meet the critical needs of healthcare, judicial system, and human resource recruitment, such as transparency, knowledge organization, improved classification results, label discrepancies, and rare and new classes. To meet these needs, ML models must be able to visualize the causal relationship between an input and an output, understand the inference and evidence produced by the model, and trust the results.