@rl3polimi

Real Life Reinforcement Learning Research Lab AIRLab - DEIB - Politecnico di Milano

 Milan, Italy
rl.airlab.deib.polimi.it/
Joined February 2023
  • Tweets55
  • Following45
  • Followers244
10Media
RL3 @ Polimi retweeted
Sergey Levine
@svlevineAug 7
Action chunking is a mysteriously effective method. Modern large-scale imitation learning basically doesn't work without it. But why does it actually help? In our new paper, we try to break down the reasons. As the saying goes, what happened next might surprise you...
Andrew Wagenmaker@ajwagenmakerAug 7
Action chunking is a critical component in virtually all modern approaches to imitation learning for robotics. But why is it so critical, and do we really need action chunking? Check out our latest work to find out! (1/n) action-chunking.github.io
16
70
749
83,579
🚀 Poster #4204 is up now! Come by to discuss Online Compatible Reward Identification from Preference Feedback — learning rewards from preferences in an online setting, with guarantees for identifying compatible reward functions. #icml2026 #RL3 #FIS2
2
9
377
🚀 Poster #4203 is up now! Come by to discuss Reusing Trajectories in Policy Gradients Enables Fast Convergence — new ideas on making reinforcement learning more sample-efficient by leveraging past trajectories. #icml2026 #RL3 #FIS2
3
15
857
🚀 Poster #4601 is live! Come by to discuss Learning to Rank from Incomplete Rankings — how to learn robust rankings when feedback is only partial, sparse, or imperfect. #icml2026 #Bandit #Ranking #FIS2
2
7
367
🚀 Poster #4501 at @icmlconf is up now! Come by to discuss Fast Mixing Steady-State Control in Markov Decision Processes — bridging control, learning, and Markov dynamics for faster convergence and better stability. #icml2026 #RL #RL3 #FIS2
4
16
939
📢 I’m looking for 𝐏𝐨𝐬𝐭𝐃𝐨𝐜𝐬 to join my team working on 𝐫𝐞𝐢𝐧𝐟𝐨𝐫𝐜𝐞𝐦𝐞𝐧𝐭 𝐥𝐞𝐚𝐫𝐧𝐢𝐧𝐠! 🧠 Theory- or experiment-oriented profiles 💰 Funded by 𝐅𝐈𝐒2 Starting Grant (Italian ERC equivalent) Details: albertometelli.github.io/job… Please share!
11
26
1,222
We are starting the @arlet_workshop! We have an amazing line-up today
8
37
2,628
🚀 Congratulations to @marcomussi95 for winning the prestigious Dimitris N. Chorafas Prize for his PhD thesis “Online Learning Methods for Pricing and Advertising.” A remarkable achievement! 🎉
4
12
554
At #EWRL2025 in Tübingen, @alberto_metelli and @papinimat presented their contributed talks - great insights on inverse RL and policy gradients shared with the community!
4
16
1,086
RL3 @ Polimi retweeted
ARLET@arlet_workshop18 Aug 2025
We've extended the deadline for our workshop's calls for papers/ideas! Submit your work by August 29 AoE. Instructions on the website: arlet-workshop.github.io/neu…
5
1
13
1,588
RL3 @ Polimi retweeted
ARLET@arlet_workshop5 Aug 2025
The OpenReview link for our calls (for papers and ideas) is available, submit here: openreview.net/group?id=Neur… We look forward to receiving your submissions!
7
1
14
7,124
🚨 Thursday’s Poster @ #ICML2025 🚨 Tomorrow, we will present three papers, (1) "Learning Utilities from Demonstrations in MDPs”, (2) "Sleeping Reinforcement Learning", (3) "Convergence Analysis of Policy Gradient Methods with Dynamic Stochasticity"!
RL3 @ Polimi@rl3polimi14 Jul 2025
The @rl3polimi group will be at #ICML2025 with 6 papers, including 1 position paper! 🎉 We’ll be sharing details about each contribution in the coming days—stay tuned! Looking forward to connecting in Vancouver 🇨🇦
1
3
194
more replies
(2/3)@simonedrago_ ,@marcomussi95 , and @alberto_metelli will present a new RL paradigm for settings where available actions change over time, explore the regret minimization problem and its challenges. 📍West Exhibition Hall B2-B3 #W-707
🕚 Thu 17 Jul, 11:00 AM–1:30 PM PDT
1
3
126
(3/3) @montenegronwski , @marcomussi95 , @papinimat @alberto_metelli will present a new framework explaining why dynamically tuning exploration in PG methods works and guarantees for a common practical approach. 📍West Exhibition Hall B2-B3#W-808
🕓 Thu 17Jul,4:30 PM–7:00 PM PDT
1
1
4
414
🚨 Wednesday’s Poster @ #ICML2025 🚨 Tomorrow, we will present one paper “Enhancing Diversity in Parallel Agents: A Maximum Exploration Story”!
RL3 @ Polimi@rl3polimi14 Jul 2025
The @rl3polimi group will be at #ICML2025 with 6 papers, including 1 position paper! 🎉 We’ll be sharing details about each contribution in the coming days—stay tuned! Looking forward to connecting in Vancouver 🇨🇦
1
1
4
343
@vincenz94, @RiccZamboni and @mirco_mutti will give an original take on how to make the best of parallel data collection for RL, how to enforce behavioral diversity and what this leads to in downstream tasks. 📍West Exhibition Hall B2-B3 #W-601
🕚 Wed 16 Jul,4:30 PM–7:00 PM PDT
1
4
198
🚨Tuesday’s Poster @ #ICML2025🚨 Tomorrow, we will present two papers: "Position: Constants are Critical in Regret Bounds for Reinforcement Learning" (1) & “Towards Theoretical Understanding of Sequential Decision Making with Preference Feedback"(2)!
RL3 @ Polimi@rl3polimi14 Jul 2025
The @rl3polimi group will be at #ICML2025 with 6 papers, including 1 position paper! 🎉 We’ll be sharing details about each contribution in the coming days—stay tuned! Looking forward to connecting in Vancouver 🇨🇦
1
3
176
(1/2)@simonedrago_ ,@marcomussi95,@alberto_metelli will challenge the obsession with asymptotics in RL theory, arguing that *constants matter*. 📍East Exhibition Hall A-B #E-503
🕚 Tue 15 Jul, 11:00 AM–1:30 PM PDT
1
3
78
(2/2) Later, @simonedrago_,@marcomussi95,@alberto_metelli will present a theoretical framework that bridges preference feedback, utilities, and rewards, laying the groundwork for RL from human preferences. 📍West Exhibition Hall B2-B3 #W-1008
🕓 Tue 15 Jul, 4:30–7:00 PM PDT
3
72
Load more