Optimizing Unnormalized Statistical Models Through Compositional Optimization

W, Jiang; J, Qin; L, Wu; C, Chen; T, Yang; L, Zhang

doi:10.1109/tpami.2025.3621320

Public

Optimizing Unnormalized Statistical Models Through Compositional Optimization.

Published in IEEE transactions on pattern analysis and machine intelligence • Feb 1, 2026

NobleIDNI9P51W76R35S52

Authors:

Jiang W

,

Qin J

,

Wu L

Abstract

Learning unnormalized statistical models (e.g., energy-based models) is computationally challenging due to the complexity of handling the partition function. To eschew this complexity, noise-contrastive estimation (NCE) has been proposed by formulating the objective as the logistic loss between the ...

Discussions