← Search

Simon Coste

2 accepted papers

2023

Efficient Training of Energy-Based Models Using Jarzynski Equality

NeurIPS 2023poster

Energy-based models (EBMs) are generative models inspired by statistical physics with a wide range of applications in unsupervised learning. Their performance is well measured by the cross-entropy (CE) of the model distribution relative to the data distribution. Using the CE as the objective for tr…