MR
Mihaela Rosca
cs.LGstat.MLcs.AIcs.CLeess.SPmath.OC
On Valency
published · living versionsW_5vz5yxu5·v1 · currentpublished
Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step
with William Fedus, Balaji Lakshminarayanan, Andrew M. Dai, Shakir Mohamed +1
1 version
Preprints & journals
14 papers in the corpus · 2016–2023On discretisation drift and smoothness regularisation in neural network training2310.14036v1 · Mihaela Claudia Rosca2023 · 0 citationsarXiv
Implicit regularisation in stochastic gradient descent: from single-objective to two-player games2307.05789v1 · Mihaela Rosca, Marc Peter Deisenroth2023 · 0 citationsarXiv
Investigating the Edge of Stability Phenomenon in Reinforcement Learning2307.04210v1 · Rares Iordan, Marc Peter Deisenroth, Mihaela Rosca2023 · 0 citationsarXiv
Why neural networks find simple solutions: the many regularizers of geometric complexity2209.13083v2 · Benoit Dherin, Michael Munn, Mihaela Rosca et al.2022 · 12 citationsarXiv
A case for new neural network smoothness constraints2012.07969v3 · Mihaela Rosca, Theophane Weber, Arthur Gretton et al.2020 · 16 citationsarXiv
Discretization Drift in Two-Player Games2105.13922v2 · Mihaela Rosca, Yan Wu, Benoit Dherin et al.2021 · 0 citationsarXiv
Spectral Normalisation for Deep Reinforcement Learning: an Optimisation Perspective2105.05246v1 · Florin Gogianu, Tudor Berariu, Mihaela Rosca et al.2021 · 12 citationsarXiv
Monte Carlo Gradient Estimation in Machine Learning1906.10652v2 · Shakir Mohamed, Mihaela Rosca, Michael Figurnov et al.2019 · 81 citationsJournal of Machine Learning Research, 21(132):1-62, 2020
Distribution Matching in Variational Inference1802.06847v4 · Mihaela Rosca, Balaji Lakshminarayanan, Shakir Mohamed2018 · 47 citationsarXiv
Deep Compressed Sensing1905.06723v2 · Yan Wu, Mihaela Rosca, Timothy Lillicrap2019 · 58 citationsarXiv
Learning Implicit Generative Models with the Method of Learned Moments1806.11006v1 · Suman Ravuri, Shakir Mohamed, Mihaela Rosca et al.2018 · 6 citationsarXiv
Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step1710.08446v3 · William Fedus, Mihaela Rosca, Balaji Lakshminarayanan et al.2017 · 159 citationsarXivon Valency
Variational Approaches for Auto-Encoding Generative Adversarial Networks1706.04987v2 · Mihaela Rosca, Balaji Lakshminarayanan, David Warde-Farley et al.2017 · 241 citationsarXiv
Sequence-to-sequence neural network models for transliteration1610.09565v1 · Mihaela Rosca, Thomas Breuel2016 · 53 citationsarXiv
Career total: 16 works. 14 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.