[2603.02533] Functional Properties of the Focal-Entropy

Computer Science > Information Theory arXiv:2603.02533 (cs) [Submitted on 3 Mar 2026] Title:Functional Properties of the Focal-Entropy Authors:Jaimin Shah, Martina Cardone, Alex Dytso View a PDF of the paper titled Functional Properties of the Focal-Entropy, by Jaimin Shah and Martina Cardone and Alex Dytso View PDF Abstract:The focal-loss has become a widely used alternative to cross-entropy in class-imbalanced classification problems, particularly in computer vision. Despite its empirical success, a systematic information-theoretic study of the focal-loss remains incomplete. In this work, we adopt a distributional viewpoint and study the focal-entropy, a focal-loss analogue of the cross-entropy. Our analysis establishes conditions for finiteness, convexity, and continuity of the focal-entropy, and provides various asymptotic characterizations. We prove the existence and uniqueness of the focal-entropy minimizer, describe its structure, and show that it can depart significantly from the data distribution. In particular, we rigorously show that the focal-loss amplifies mid-range probabilities, suppresses high-probability outcomes, and, under extreme class imbalance, induces an over-suppression regime in which very small probabilities are further diminished. These results, which are also experimentally validated, offer a theoretical foundation for understanding the focal-loss and clarify the trade-offs that it introduces when applied to imbalanced learning tasks. Comments: ...