Benjamin Heymann

CONSEQUENCES '25 Workshop - Monday 22 📌 Room D3 Org: @flavianv 🗣️Spotlight Talk: Offline Contextual Bandit with Counterfactual Sample Identification by Alexandre Gilotte, @OtmaneSakhi , @imad_aouali and @BenjaminHe12201 arxiv.org/abs/2509.10520

Non-Linear Counterfactual Aggregate Optimization by @BenjaminHe12201 and @OtmaneSakhi arxiv.org/abs/2509.03438

New day at #NeurIPS2024 Don't miss the poster presentation today @NeurIPSConf 📍West Ballroom A-D #6808 ⏰4:30 pm to 7:30 pm (PST) 🗣️@BenjaminHe12201 @MaximeVono @VianneyPerchet 📝 DU-Shapley: A Shapley Value Proxy for Efficient Dataset Valuation neurips.cc/virtual/2024/p…

Dive into our #NeurIPS paper and join us in advancing fairness in advertising! 🔗 arxiv.org/abs/2407.03059 I am in Vancouver in person this week 🍁 If you attend the conference, more information on the poster at neurips.cc/virtual/2024/p…

🔴 Paper presentation at #WINE2024 today 🔴 Join Alexandre Gilotte at 13:30, Session 7A, to know more about 👇 "Repeated Bidding with Dynamic Value" /cc @BenjaminHe12201 We published a research card about it with all the details 📝 techblog.criteo.com/research-card-…

🔴 #NeurIPS24 is happening this week and we are sent a dream team to the event! 🚀 They will be presenting their latest #ML papers (13 papers, 1 dataset paper and 1 challenge paper), bringing original #AI contributions on #AdTech topics. @NeurIPSConf 👇 ailab.criteo.com/papers-from-th…

🔴New research card is out!🔴 How to optimally account for the user fatigue in the bid? #AdTech #RTB This paper explores the scenario where a bidder's valuation for an item (e.g., ad opportunity) depends on the time elapsed since their last successful bid. techblog.criteo.com/research-card-…

@CriteoAILab A short blog post about this work on medium: techblog.criteo.com/research-card-…

Paper accepted at #Wine2024 /cc @BenjaminHe12201 👏🚀

@CriteoAILab Here is the link to the preprint: arxiv.org/abs/2308.01755

Tomorrow at #RecSys2024, join @flavianv's presentation during #KaRS2024 workshop 👇 📍Friday 18 ⏰ 10AM 📝 Can LLMs reliably reason about user preferences for Recommendation? kars-workshop.github.io/2024/program/

@CriteoAILab @CONSEQUENCES_ws @ACMRecSys @BayesianIn @flavianv @OtmaneSakhi @imad_aouali Be sure to check the associated blog post on medium: techblog.criteo.com/recommender-sy… (with @flavianv )

@CONSEQUENCES_ws @ACMRecSys Monday! “Welfare-Optimized Recommender Systems” - SURE (@BenjaminHe12201 @BayesianIn @flavianv) "Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning" - Consequences Workshop @OtmaneSakhi @imad_aouali

#RecSys2024 is happening next week and a Criteo crew is participating in: * Monday - SURE & @CONSEQUENCES_ws * Wednesday - Industrial Talk at Session 11: Optimisation and Evaluation 1 * Friday - KaRS worshop @ACMRecSys Check the thread to know more about our presentations 👇

Indeed, our paper "Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning" has been accepted to @NeurIPSConf 2024 as a spotlight! Always a pleasure to collaborate with @imad_aouali , @PierreAlquier and Nicolas Chopin! Congrats to my co-authors 🎉

This year's program will have two keynotes and a tutorial, in addition to eight oral presentations of workshop contributions and a poster session. Full abstracts and programme can be found on our website: sites.google.com/view/consequen…

@CriteoAILab @CONSEQUENCES_ws @ACMRecSys @BayesianIn @flavianv @OtmaneSakhi @imad_aouali @flavianv, from @CriteoAILab will be presenting our paper at @ACMRecSys #RecSys2024 #RecSys24

How can we ensure that an ad or a job offer can be seen in a non-discriminatory way by the people they are displayed to? 🤔 #FAIRPLAY: a Research program to Optimize Fair and Private Interactions between ML Agents in AI Marketplaces 👇 @Inria @ENSAEparis techblog.criteo.com/fairplay-resea…

Our paper Welfare-Optimized Recommender Systems is accepted for the workshop on Strategic and Utility-aware REcommendation (SURE) at the 18th ACM Conference on Recommender Systems (RecSys24)!

Our paper Repeated Bidding with Dynamic Value is accepted to WINE 2024.