Adversarially-Regularized Mixed Effects Deep Learning (ARMED) Models Improve Interpretability, Performance, and Generalization on Clustered (non-iid) Data

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 45; no. 7; pp. 8081 - 8093
• Main Authors: Nguyen, Kevin P., Treacher, Alex H., Montillo, Albert A.
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.07.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptation models; Bayes methods; Biological system modeling; biomedical imaging; clinical data; Clusters; Data models; Datasets; Deep learning; Generalization; Image analysis; interpretability; Invariants; Machine learning; mixed effects model; multilevel model; Neural networks; Performance enhancement; Predictive models; Training

Natural science datasets frequently violate assumptions of independence. Samples may be clustered (e.g., by study site, subject, or experimental batch), leading to spurious associations, poor model fitting, and confounded analyses. While largely unaddressed in deep learning, this problem has been ha...