Nadav Merlis

✨Exciting News from CREST!✨ We’re proud to announce that 19 papers authored by CREST researchers and PhD students have been selected for the #neurips2024 Conference in Vancouver!🌎 The NeurIPS conference is one of the most prestigious events in #AI! linkedin.com/feed/update/ac…

It's been a while without updates. Let's cover some of the recent publications from the various RL groups at the @TechnionLive ! Thread.

The first Israeli RL Day will take place on Apr. 7th! Program is packed with great speakers, plus keynotes by @AvivTamar1, @MannorShie and Yishay Mansour. Schedule details at israel-rl-day.github.io. #IRLD2021

"Tight Lower Bounds for Combinatorial Multi-Armed Bandits" Recent work by @NadavMerlis and Shie Mannor was accepted to #COLT2020 ! Read their paper here: arxiv.org/abs/2002.05392

excited to announce a new series of virtual seminars on ~~~REINFORCEMENT LEARNING THEORY~~~ we've set this up with @CiaraPikeBurke and @CsabaSzepesvari to keep track of all the advances of this fast-paced field. hope others will also find it useful! sites.google.com/view/rltheorys…

New blog post by Prof. Shie Mannor on why Reinforcement learning doesn't work in many practical applications and what open questions need answering before we can trust RL in the industry. rlrl.net.technion.ac.il/2020/01/27/why…

My first tweet ! On Tuesday evening (7-9pm, West 217-219), we organize a #NeurIPS2019 Social Event "Learning Theory". Let's have a beer and a bite between theoretical learners :-) Thanks to #Criteo for sponsoring it!

@TechnionLive Batch-Size Independent Regret Bounds for the Combinatorial Multi-Armed Bandit Problem Nadav Merlis and Shie Mannor COLT 2019 We introduce a new smoothness criterion, which that takes into account both the nonlinearity of the reward and concentration properties of the arms. (5/n)

@NeurIPSConf Tight Regret Bounds for Model-Based Reinforcement Learning with Greedy Policies Yonathan Efroni*, Nadav Merlis*, Mohammad Ghavamzadeh, Shie Mannor NeurIPS 2019 We show that local models are sufficient and there is no need to perform full planning! (2/n)