Claire Vernade
Bio
Claire is a Research Scientist at DeepMind in London UK. She received her PhD from Telecom ParisTech in October 2017, under the guidance of Prof. Olivier Cappé. From January 2018-October 2018, she worked part-time as an Applied Scientist at Amazon in Berlin, while doing a post-doc with Alexandra Carpentier at the University of Magdeburg in Germany.
Her research is on sequential decision making. It mostly spans bandit problems, but Claire's interest also extends to Reinforcement Learning and Learning Theory. While keeping in mind concrete problems -- often inspired by interactions with product teams -- she focuses on theoretical approaches, aiming for provably optimal algorithms.
Publications
News !
ACCEPTED AT AISTATS 2021 : Off-Policy Evaluation for Contextual Bandits: finite-time confidence intervals can be used to help choosing the best policy out of a finite set of candidates.
Ilja Kuzborskij, Claire Vernade, András György, Csaba Szepesvári. "Confident Off-Policy Evaluation and Selection through Self-Normalized Importance Weighting". AISTATS 2021 (arXiv link)
ACCEPTED as an ORAL PRESENTATION AT ICLR 2021: A Game-theoretic approach to PCA to scale it up to larger and larger datasets and distribute computation over servers.
I. Gemp, B. McWilliams, C. Vernade, T. Graepel. "EigenGame: PCA as a Nash Equilibrium". ICLR 2021 (arXiv link)
New preprint: We denoise the gradient updates of EigenGame which results in a significant reduction of the variance:
I. Gemp, B. McWilliams, C. Vernade, T. Graepel. "EigenGame Unloaded: When playing games is better than optimizing". arXiv 2102.04152
Other conference papers
Claire Vernade, Alexandra Carpentier, Giovanni Zappella, Beyza Ermis, and Michael Brueckner. "Contextual bandits under delayed feedback." arXiv:1807.02089 . ICML 2020
Claire Vernade*, Andras Gyorgy*, Timothy Mann. Non-Stationary Delayed bandits with Intermediate Observations. arXiv 2006.02119 ICML 2020
Anne-Gaelle Manegueu, Claire Vernade, Alexandra Carpentier, Michal Valko. Stochastic Bandits with Arm-Dependent Delays. arXiv 2006.10459. ICML 2020
Cindy Trinh, Emilie Kaufmann, Claire Vernade, and Richard Combes. "Solving Bernoulli Rank-One Bandits with Unimodal Thompson Sampling." arXiv preprint arXiv:1912.03074 ALT 2020
Russac, Yoan, Claire Vernade, and Olivier Cappé. "Weighted Linear Bandits for Non-Stationary Environments." In Advances in Neural Information Processing Systems, pp. 12017-12026. 2019.
Achab, Mastane; Clémençon, Stephan; Garivier, Aurélien; Sabourin, Anne; Vernade, Claire; "Max K-armed bandit: On the ExtremeHunter algorithm and beyond" ECML 2017
Vernade, Claire; Cappé, Olivier; Perchet, Vianney; "Stochastic bandit models for delayed conversions" UAI 2017
Kwon, Joon; Perchet, Vianney; Vernade, Claire; "Sparse stochastic bandits". COLT 2017.
Katariya, Sumeet; Kveton, Branislav; Szepesvári, Csaba; Vernade, Claire; Wen, Zheng; "Bernoulli Rank-$1 $ Bandits for Click Feedback". IJCAI 2017
Katariya, Sumeet; Kveton, Branislav; Szepesvari, Csaba; Vernade, Claire; Wen, Zheng; "Stochastic rank-1 bandits". AISTATS 2017.
Lagrée, Paul; Vernade, Claire; Cappé, Olivier; "Multiple-play bandits in the position-based model". NIPS 2016
Vernade, Claire; Cappé, Olivier; "Learning from missing data using selection bias in movie recommendation". IEEE International Conference on Data Science and Advanced Analytics (2015)
Patents
Kveton, Branislav, Zheng Wen, Yasin Abbasi Yadkori, Mohammad Ghavamzadeh, and Claire Vernade. "Multivariate digital campaign content testing utilizing rank-1 best-arm identification." U.S. Patent Application 15/944,980, filed October 10, 2019.
Collaborators and friends
Co-authors
Paul Lagrée
Vianney Perchet
Branislav Kveton
Sumeet Katariya
Zheng Wen
Mastane Achab
Aurelien Garivier
Anne Sabourin
Stephan Clemencon
Cindy Trinh
Joon Kwon
Giovanni Zappella
Beyza Ermis
Michael Brueckner
Selected Talks
March 2020: Workshop on Optimisation and Machine Learning at CIRM (France)
November 26th 2019: Seminaire du departement d'informatique de l'ENS Rennes (in French)
March 2018: Invited talk at University of Washington (invited by Joseph Salmon and Zaid Harchaoui)
October 20th 2017: PhD Defense
June 2016: Workshop on Machine Learning for Online Advertising at ICML 2016
Community
I strongly support community groups, in particular towards better diversity in our extended field (Computer Science and Stats). I have volunteered and area chaired for the WiML workshop many times and I am co-leading with Ruth Urner the WiML-T effort for Women in Learning Theory .