Re: [rl-list] Digest for rl-list@googlegroups.com - 6 updates in 6 topics

[Mohammed Alkharraz]

Stop

On Sun, 6 Jul 2025 at 8:48 PM <rl-...@googlegroups.com> wrote:

rl-...@googlegroups.com

Google Groups

Topic digest
View all topics

• [NeurIPS2025 competition] Call for contributions - NeurIPS 2025 E2LM Competition - 1 Update
• Final Call for Application | OxML 2025 - 1 Update
• [Jobs] Fully funded PhD studentship in Sustainable Lifelong Robot Learning at the University of Southampton, UK - 1 Update
• [Journals] CFP: Special Issue on 'Social Robot Navigation – Opportunities, Algorithms, Tools, and Systems' - 1 Update
• Hybrid master-level courses in Machine Learning - 1 Update
• [CFP][meetings] AAAI 2025 Fall Symposium on “Unifying Representations for Robot Application Development” (UR-RAD) - 1 Update

[NeurIPS2025 competition] Call for contributions - NeurIPS 2025 E2LM Competition

Reda Alami <alamir...@gmail.com>: Jul 04 05:08PM +0400 Call for contributions NeurIPS 2025 E2LM Competition: Early Training Evaluation of Language Models       Join us in building benchmarks that capture early-stage reasoning & scientific knowledge in LLMs!   The development of Large Language Models (LLMs) typically begins with a series of ablation experiments, wherein various model architectures, data mixtures, and training hyperparameters are systematically evaluated. This phase is commonly referred to as the early stages of training. During this period, researchers primarily monitor two key metrics: the training loss curve and evaluation scores. However, existing evaluation benchmarks often fail to provide meaningful or discriminative signals during these initial stages where LLMs are trained on a few tokens ~200B tokens, making it challenging to derive conclusive insights from ongoing experiments.   In this competition, we want to build together new benchmarks to effectively capture relevant signals in early training stages of LLMs, specifically for scientific knowledge domain. How to participate   The competition will be hosted on a dedicated Hugging Face organization - to register to the competition please follow this registration link 👉 https://e2lmc.github.io/registration. Participants will have to submit their solutions, which will be based on lm-evaluation-harness <https://github.com/EleutherAI/lm-evaluation-harness> library through a HuggingFace Space. An active leaderboard will be maintained during the competition to track promising submissions.   The size of the models make them easily runnable for everyone, on free-tier Google Colab GPUs. We also provide a comprehensive starting kit <https://e2lmc.github.io/starter_kit> including several notebooks to get started with the competition.     Evaluation metrics   Each submission will be evaluated using three different scores:   - Signal Quality : smooth, meaningful learning curves - Ranking Consistency : stable model rankings across training (until 1 Tera Tokens) - Scientific Compliance: benchmarks should accurately reflect scientific knowledge and reasoning.       *Timeline*   Competition kick-off : July 14 July 14 – August 18: Warm-up phase August 18 – October 27: Development phase October 27 – November 7: Final evaluation December 6/7: Winning solutions presentation @NeurIPS 2025       *💰** Prizes*     🏆 $6,000 – 1st place 🥈 $4,000 – 2nd place 🥉 $2,000 – 3rd place 🎓 $2,000 × 2 – Best student submissions 🎓 Winning entries will be showcased at a dedicated workshop @Neurips 2025   More details on the competition website: https://e2lmc.github.io/ Register for the competition: https://lnkd.in/euqjzJcx Read the competition proposal: https://lnkd.in/eu9TKsVh

Back to top

Final Call for Application | OxML 2025

OxML Team <con...@oxfordml.school>: Jul 01 03:59AM -0700 Join us this summer for the *Oxford Machine Learning Summer School (OxML)* at the *University of Oxford’s Mathematical Institute* and online.   · *MLx Health & Bio*: 2–5 August 2025   · *MLx Representation Learning & Generative AI*: 7–10 August 2025     *Website*: www.oxfordml.school   *Application Deadline: 07 July 2025*       *About OxML 2025*   The Oxford Machine Learning Summer School (OxML), organised by AI for Global Goals, brings together top researchers and industry leaders to explore the latest advances in machine learning. This year’s programme features two expert tracks, each blending theory with real-world application:       *1.* *MLx Health & Bio*   A deep dive into ML methods for healthcare and biomedical research. Topics include medical imaging, EHR, multi-omics, drug discovery, wearable devices, computational biology, and more.   *Confirmed Speakers & Topics:*   · *Michael Bronstein* (Oxford): Geometric deep learning for drug discovery   · *Arthur Gretton* (UCL, DeepMind): Causal ML   · *Aiden Doherty* (Oxford): Wearables & ML for health outcomes   · *Zeyu Gao* (Cambridge): Multiomics   · *Hoifung Poon* (Microsoft Health): Foundation models in medicine   · *Brian Hie* (Stanford): Biological foundation models (Evo 2)   · *Rahul Krishnan* (Toronto): Time-to-event modeling   · *Vivek Natarajan* (Google): Multimodal foundation models in biomedicine   · *Pierre Masselot* (LSHTM): ML in environmental epidemiology   · *Moritz Kraemer* (Oxford): ML for pandemics     *2.* *MLx Representation Learning & Generative AI*   This track focuses on the theoretical and applied aspects of representation learning and generative AI, including LLMs, CV, RL, multi-modal models, AI alignment, GenAI product development, and more.   *Confirmed Speakers & Topics:*   · *Aymeric Dieuleveut* (École Polytechnique): Conformal prediction   · *Peyman Milanfar* (Google): Denoising in imaging & ML   · *Cheng Zhang* (Meta): TBC   · *Abdul Fatir Ansari* (AWS): Time series representation learning   · *Ashley Edwards* (Runway): Generative AI in images/videos   · *Fazl Barez* (Oxford): AI safety & alignment   · *Haitham Bou-Ammar* (Huawei): TBC   · *Gerhard Neumann* (KIT): TBC   · *Hannaneh Hajishirzi* (UW, AI2): TBC   · *David Salinas* (UW): Tabular foundation models, AutoML   · *Edward Johns* (Imperial): GenAI for robotics   · *Ilia Shumailov* (DeepMind): ML security     *3.* *Practical Workshops (included in both tracks)*   Hands-on sessions on deploying ML to edge devices and robots.   *Open-Source Tools for ML on Edge Devices & Robots*:   · *Vincent Moens* (Meta)   · *Rémi Cadène* (Hugging Face)   · *Xuan-Son Nguyen* (Hugging Face)   *Deploying CV Applications to Embedded Systems*:   · *Mergen Nachin* (Meta)     We welcome applications from PhD students, postdocs, researchers, and professionals in ML and related disciplines.

Back to top

[Jobs] Fully funded PhD studentship in Sustainable Lifelong Robot Learning at the University of Southampton, UK

Burhan Hafez <burhan...@gmail.com>: Jul 01 06:49PM +0100 Dear colleagues,   We're hiring a PhD student to develop a sustainable, lifelong on-device robot learning system at the University of Southampton.   *Project Title*   Resource-efficient lifelong robot learning   *About the PhD Project*   Equipping robots with the ability to learn a growing set of tasks over their operational lifetime—rather than focusing on mastering individual tasks—presents a significant challenge in robot learning. Lifelong learning robots often struggle with catastrophic forgetting when learning from changing input distributions, which causes the robot to forget old knowledge when learning new tasks. They are also expected to leverage previous knowledge to accelerate learning of new tasks without the need for complete retraining. This is often referred to as the stability-plasticity dilemma, where stability denotes the retention of old knowledge, and plasticity refers to the acquisition of new knowledge.   Recent advancements in lifelong and continual learning have proposed three primary strategies to address this dilemma: regularization, dynamic growth, and experience replay. However, these methods typically demand high storage and computational resources, leading to increased energy consumption for data storage, processing, and transmission. Additionally, robots often face limitations in onboard resources, making it difficult to support lifelong learning outside controlled lab environments and to retain and integrate experiences from various environments and tasks. Although some recent approaches have shown promise in improving efficiency in continual robot learning, they often come at the cost of reduced performance compared to single-task models, where each task is learned with a dedicated model.   This project aims to develop a continual on-device robot learning system that improves the trade-off between stability and plasticity while enhancing resource efficiency without compromising performance. The system will be designed for deployment on resource-constrained, non-networked robotic platforms and aims to contribute to sustainability by reducing carbon emissions through improved operational efficiency, including minimizing the need for frequent retraining and optimizing data handling processes.   *Funding Available For*   UK, EU and Horizon Europe Associated Countries <https://www.southampton.ac.uk/study/fees-funding/scholarships/postgraduate-uk/horizon-europe-fee-waiver>   *Entry Requirements*   1. A very good undergraduate degree (at least a UK 2:1 honours degree) or its international equivalent 2. IELTS 6.5 (or equivalent) for non-native English speakers     *How to Apply*   Please send your pre-application package including CV, transcripts, sample publications (if any), two references and cover letter expressing your research vision and interest in the position to burhan...@soton.ac.uk   *Closing Date*   31 August 2025. Applications will be considered in the order that they are received. The position will be considered filled when a suitable candidate has been identified.   For more information, please visit: https://www.southampton.ac.uk/study/postgraduate-research/projects/resource-efficient-lifelong-robot-learning   Best regards,   Muhammad Burhan Hafez --   Dr. Muhammad Burhan Hafez   Assistant Professor   School of Electronics & Computer Science   University of Southampton   Southampton SO17 1BJ, UK   E-mail: burhan...@soton.ac.uk   Homepage: https://www.soton.ac.uk/people/65b6st <https://www.southampton.ac.uk/people/65b6st/doctor-muhammad-burhan-hafez>

Back to top

[Journals] CFP: Special Issue on 'Social Robot Navigation – Opportunities, Algorithms, Tools, and Systems'

Allan Wang <allanwa...@gmail.com>: Jul 01 01:33AM -0700 Dear fellow roboticists,   We are bringing key social navigation researchers together in a focused special issue on “Social Robot Navigation – Opportunities, Algorithms, Tools, and Systems”, in Frontiers in Robotics and AI.   We cordially invite you to contribute a submission to our special issue. Please find all information about the opportunity in our special issues's website.: https://www.frontiersin.org/researchtopic/68960 If you plan to contribute, please register your interest by clicking "Participate in this topic".   Timeline:   The Special Issue operates on a rolling submission basis, through Fall 2025:   - Last abstract submission for the year: *7* November 2025   - Last manuscript submission for the year: 14 November 2025   Topic Editors:   - Allan Wang (Miraikan)   - Phani Teja Singamaneni (Laboratoire d'analyse et d'architecture Des Systèmes (LAAS))   - Jonathan Francis (Bosch Center for AI + Carnegie Mellon University)   - Dražen Brščić (Kyoto University)       Please feel free to let us know if you have any questions.     Best regards,   Allan

Back to top

Hybrid master-level courses in Machine Learning

Yevgeny Seldin <yevgeny...@gmail.com>: Jul 02 10:48AM +0200 The coming academic year the Department of Computer Science at the University of Copenhagen offers four master-level 7.5 ECTS courses in Machine Learning in hybrid format. The courses offer hybrid lectures that are streamed on Zoom and recorded, and a choice of physical and online exercise classes. The courses support fully remote participation. The courses provide strong theoretical foundations in the field of Machine Learning and welcome participants from universities and industry.   The courses offered are:   *Machine Learning A* - https://kurser.ku.dk/course/ndak22000u/2025-2026 1 September 2025 - 9 November 2025 Course organizer - Sadegh Talebi <sadegh...@di.ku.dk>   *Advanced Topics in Machine Learning* (this year the focus is on *Privacy in Machine Learning*) - https://kurser.ku.dk/course/ndak15014u/2025-2026 1 September 2025 - 9 November 2025 Course organizer - Amartya Sanyal <am...@di.ku.dk>   *Online and Reinforcement Learning* - https://kurser.ku.dk/course/ndak21003u/2025-2026 2 February 2026 - 12 April 2026 Course organizer - Sadegh Talebi <sadegh...@di.ku.dk>   *Machine Learning B* - https://kurser.ku.dk/course/ndak22001u/2025-2026 20 April 2026 - 21 June 2026 Course organizer - Nirupam Gupta <ni...@di.ku.dk>   "Machine Learning A" is the entry point for the other three courses (they assume that the students have successfully passed this course). The other three courses do not depend on each other and can be taken in any order. *All the four courses assume solid mathematical background.*   For further details and registration check https://science.ku.dk/english/courses-and-programmes/other-study-opportunities/hybrid-and-online-courses/   If you have questions concerning individual courses, please, contact the corresponding course organizer.

Back to top

[CFP][meetings] AAAI 2025 Fall Symposium on “Unifying Representations for Robot Application Development” (UR-RAD)

David Porfirio <david.j....@gmail.com>: Jul 01 10:02AM -0700 What: AAAI 2025 Fall Symposium on Unifying Representations for Robot Application Development (UR-RAD)   When: November 6-8, 2025   Where: Arlington, VA   Web: ur-rad.github.io     —   Dear colleagues,   We invite you to submit a paper and/or participate in the 3rd AAAI Fall Symposium on Unifying Representations for Robot Application Development (UR-RAD). Accepted papers will be presented in person only (though participants who are not presenting may attend remotely).   Capturing a desired task or interaction as a computational artifact (i.e., a representation) has long played a pivotal role in robotics. Many robotic subfields have traditionally employed a variety of different representational techniques, such as LTL, planning languages, social representations, natural language, and many more. These representations, however, lack cohesion in when and how they are applied. The 3rd Symposium on Unifying Representations for Robot Application Development (UR-RAD) therefore aims to increase interaction between junior and senior researchers, support in-progress research, and cultivate collaboration.   Symposium Website: ur-rad.github.io AAAI Fall Symposium Series Website: https://aaai.org/conference/fall-symposia/fss25/     ========================= Submission Topics of Interest =========================   - Representational trends   - Representations for robot learning   - AI planning for robotics - Formal methods in robotics - Natural language as a representation - Novel representations & novel representation uses - Representations for user interfaces - Robot end-user development - Robot programming interfaces & paradigms - Robot runtime/control environments - Opportunities for standardization - Frameworks (e.g., ROS or middleware) - Open-source & collaboration initiatives - Identifying representation requirements   ==================== Important Dates ====================   "Preferred" round, with earlier feedback and the option to be included in the AAAI proceedings:   - Submission deadline: August 7th (hard deadline) - Camera Ready Deadline: August 29th (hard deadline)   "Late" round, without the option to archive in the AAAI proceedings:   - 100-200 word abstract submission encouraged by: August 7 - Submission deadline: August 22 - Camera Ready Deadline: September 5   Regardless of archival plans, authors are encouraged to submit earlier rather than later. All papers submitted by August 7th, regardless of archival plans, will receive preference if a large number of submissions are received.   ==================== Submission Information ====================   We invite the following contributions, formatted using the AAAI-25 author kit <https://aaai.org/authorkit25/>:   - Full Papers (4-8 pages, archival OR non-archival): for novel research, artifact submissions, or strong works in progress. - Short Papers (2-4 pages, archival OR non-archival): for positions, smaller artifacts, and early work in progress. - Abstracts (1 page, non-archival only): for sharing ideas.   In order to encourage interaction between junior and senior researchers, each paper accepted to UR-RAD 2025 (regardless of archival status or contribution type) has the option to be paired with an expert in the field, as a “mentor”. Paper authors who opt in, in addition to junior members of the community, will have the opportunity for extensive interaction with mentors, who will guide discussions about individual papers.     If you intend to submit a paper, please email us at urrad.s...@gmail.com. Use the subject line, “[UR-RAD] Intent to Submit”, and include the following in the body of the message: (1) a tentative title, (2) a tentative author list, and (3) tentative keywords. This step is optional, but will aid in our planning for reviewers. Note that at least one author from each paper will be expected to review 2-3 UR-RAD papers themselves.   =================== Organizing Committee ===================   Ruchen Wen (Co-Chair, Colgate University)   David Porfirio (Co-Chair, Naval Research Laboratory) Saad Elbeleidy (Peerbots) Laura M. Hiatt (Naval Research Laboratory) Ross Mead (Semio) Andrew Schoen (Semio) Willie Wilson (Franklin & Marshall College) Laura Stegner (George Washington University)   Contact urrad.s...@gmail.com for any questions!

Back to top

You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to rl-list+u...@googlegroups.com.