2 papers
stat.ML2026
Annealing in variational inference mitigates mode collapse: A theoretical study on Gaussian mixtures
Luigi Fogliani, Bruno Loureiro, Marylou Gabrié
Mode collapse, the failure to capture one or more modes when targetting a multimodal distribution, is a central challenge in modern variational inference. In this work, we provide…
stat.ML2024
A theoretical perspective on mode collapse in variational inference
Roman Soletskyi, Marylou Gabrié, Bruno Loureiro
While deep learning has expanded the possibilities for highly expressive variational families, the practical benefits of these tools for variational inference (VI) are often limite…