Composite multiclass losses

MaRDI QIDQ2953650zbMATH OpenFDO

Authors Robert C. Williamson, Elodie Vernet, Mark D. Reid

Publication date 5 January 2017

Full work available at URL http://jmlr.csail.mit.edu/papers/v17/14-294.html

zbMATH Keywords

admissibilityclassification calibrationconvexity and quasi-convexity of losseslink functionsmargin lossesminimaxitymixabilitymulticlass lossesparametrisations and representations of loss functionsproper lossessuperprediction set

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Learning and adaptive systems in artificial intelligence (68T05)

Recommendations

• Composite binary losses
• Convex calibration dimension for multiclass loss matrices
• Multiclass classification, information, divergence and surrogate risk
• Mixability is Bayes risk curvature relative to log loss
• Are Loss Functions All the Same?

Cited in

(10)

• Consistent algorithms for multiclass classification with an abstain option
• Multiclass classification, information, divergence and surrogate risk
• Properization: constructing proper scoring rules via Bayes acts
• Convex calibration dimension for multiclass loss matrices
• Composite binary losses
• scientific article; zbMATH DE number 7255066 (Why is no real title available?)
• Memoryless sequences for general losses
• Multicategory composite least squares classifiers
• Mixability is Bayes risk curvature relative to log loss
• A unified view of forward and backward losses for learning from weak labels

This page was built for publication: Composite multiclass losses

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2953650)