|
| |||||||||||
DeepLearn 2027 : 14th International School on Deep Learning | |||||||||||
| Link: https://deeplearn.irdta.eu/2027/ | |||||||||||
| |||||||||||
Call For Papers | |||||||||||
|
SCOPE
DeepLearn 2027 will be a research training event with a global scope, aimed at updating participants on the most recent advances in the rapidly evolving field of deep learning. It continues a successful series of international schools organized since 2017 in several European countries. Deep learning has become one of the principal driving forces of artificial intelligence, underpinning an extraordinary range of scientific advances, technological developments, and industrial applications. Its impact extends across computer vision, natural language processing, speech and audio technologies, robotics, autonomous systems, biomedicine and healthcare, drug discovery, neuroscience, human-computer interaction, recommender systems, finance, cybersecurity, communications, climate and environmental sciences, physics, materials science, genomics, social sciences, mathematics, and many other domains. Recent advances in foundation models, large language models, generative AI, multimodal learning, self-supervised learning and representation learning, reinforcement learning, and AI agents have profoundly transformed the field. They have opened new research avenues in reasoning, planning, decision-making, scientific discovery, embodied intelligence, and the integration of different modalities of information. At the same time, deep learning continues to raise fundamental scientific and technological questions. These concern, among others, the theoretical foundations of learning and generalization, optimization, data efficiency, computational and energy requirements, robustness, reliability, interpretability, explainability, safety, privacy, fairness, and the responsible deployment of increasingly capable AI systems. The interaction between theory, algorithms, applications, and societal considerations makes deep learning a particularly rich and dynamic research area, in which interdisciplinary collaboration is becoming increasingly important. DeepLearn 2027 will provide a broad and up-to-date overview of the field, combining established foundations with emerging research directions and open challenges. ADDRESSED TO Graduates, postgraduates, researchers, and industry practitioners will be typical profiles of participants. However, there will be no formal prerequisites for attendance in terms of academic degrees, and individuals at different stages of their academic or professional careers will be welcome. A substantial proportion of participants are expected to be PhD students and researchers with backgrounds in computer science, mathematics, statistics, and engineering. Nevertheless, the school will also welcome participants from life and medical sciences, physics, social sciences, the humanities, and other disciplines in which deep learning is increasingly relevant. Courses will be offered at different levels, including introductory, intermediate, and advanced. Specific knowledge and technical background may be assumed for some of them. Participants will be able to select those best suited to their interests, experience, and research objectives. The opportunity to learn directly from leading researchers and industry innovators, exchange ideas, and establish new collaborations will be particularly valuable. VENUE DeepLearn 2027 will take place in Pula, a historic coastal city on the Istrian peninsula in Croatia, overlooking the Adriatic Sea. The city combines an exceptional cultural and historical heritage with an attractive Mediterranean environment, providing a welcoming setting for an international scientific event. The venue will be: Juraj Dobrila University of Pula Faculty of Economics and Tourism "Dr. Mijo Mirković" Preradovićeva 1 52100 Pula, Croatia https://www.unipu.hr/en STRUCTURE The scientific program will comprise 18 monographic courses, 2 keynote lectures, 1 open session, 1 industrial session, 1 round table, and a hackathon competition. Each course will consist of 4 hours and 30 minutes of instruction, organized into three 90-minute sessions over two consecutive days. Courses will cover a variety of topics and levels, from introductory to advanced. Three courses will run in parallel throughout the event. Participants will be free to choose the courses they wish to attend and to move from one course to another. Participants will have the opportunity to present their own work. Likewise, companies will be able to display their industrial developments. The hackathon will allow participants to work in teams on machine learning challenges, encouraging collaboration, experimentation, and the practical application of acquired knowledge. The event will be fully streamed. Live online participation will be possible. The organizers highlight, however, the importance of face-to-face interaction and networking in this kind of research training event. All lectures will be video-recorded. Participants will be able to watch them again for 45 days after the event. KEYNOTE SPEAKERS (to be completed) PROFESSORS AND COURSES (to be completed) Plamen Angelov (Lancaster University), [intermediate] Interpretability and Adaptivity Aspects of Deep Learning: Problems and Solutions Aaron Courville (University of Montreal), [intermediate/advanced] LLMs Reasoning, Agents and Agent Interaction Daniel Cremers (Technical University of Munich), [intermediate] The Dawn of Spatial Intelligence: 3D Computer Vision in the Age of Deep Learning Jennifer Dy (Northeastern University), [intermediate] Interpretable and Explainable AI Brian Kulis (Boston University), [intermediate] Models for Audio and Other Long Time Series Alice O’Toole (University of Texas Dallas), [introductory/intermediate] Using Deep Learning for Face, Body, and Person Recognition in Real World Viewing Conditions Wojciech Samek (Technical University of Berlin & Fraunhofer Institute for Telecommunications), [intermediate] Understanding the Next Generation of AI: From Foundation Models to Autonomous Agents Franco Scarselli (University of Siena), [intermediate] Theory of Graph Neural Networks: State of the Art and Perspectives Nicu Sebe (University of Trento), [intermediate] Cross-Modal Generation, Understanding, and Action: From Multimodal Content to Vision-Language-Action Models René Vidal (University of Pennsylvania), [intermediate] Engineering Trustworthy AI Systems OPEN SESSION An open session will feature voluntary five-minute oral presentations of work in progress by participants. They should submit a half-page abstract containing the title, authors, and summary of the research to david@irdta.eu by July 12, 2027. INDUSTRIAL SESSION A session will be devoted to 10-minute demonstrations of practical applications of deep learning in industry. Companies interested in contributing are welcome to submit a 1-page abstract containing a description of the demonstration and any logistical requirements. People in charge of the demonstration must register for the event. Abstracts have to be submitted to david@irdta.eu by July 12, 2027. ROUND TABLE A round table will bring together experts to discuss current challenges and future directions in deep learning. The discussion will encourage interaction between the panelists and the audience. HACKATHON A hackathon will take place, where participants will voluntarily work in teams to tackle several machine learning challenges. They will be coordinated by Professor Sergei V. Gleyzer (University of Alabama). The challenges, participation rules, evaluation criteria, submission procedure, and calendar will be announced two weeks before the beginning of the school. The submissions will be assessed by a jury, which will select the winning team of each challenge by the end of August 2027. The winning teams will receive a modest monetary prize and the runners-up will get a certificate. ORGANIZING COMMITTEE Sergei V. Gleyzer (Tuscaloosa, hackathon chair) Karlo Griparić (Pula, local chair) Carlos Martín-Vide (Tarragona, program chair) Santiago Montes (Tarragona, web manager) Sara Morales (Luxembourg, financial officer) David Silva (London, organization chair) Diego Sušanj (Pula, member) REGISTRATION Registration is available at https://deeplearn.irdta.eu/2027/registration/ Participants will indicate a tentative selection of six courses during registration. This selection will be non-binding and will be used only to help the organizers estimate demand and plan the allocation of lecture rooms. Participants will be free to attend the courses that best suit their interests. Since the capacity of the venue is limited, registration requests will be processed on a first-come, first-served basis. Registration will close once maximum capacity has been reached. Early registration is therefore strongly recommended. FEES Fees comprise access to all program activities and lunches. Several early registration deadlines are available, with fees depending on the date of payment. The fees for on-site and for online participation are the same. SPONSORS Companies, institutions, and organizations interested in supporting DeepLearn 2027 are welcome to become sponsors of the event. Sponsorship provides an opportunity to contribute to advanced international research training in artificial intelligence and to establish contact with highly motivated students, researchers, and professionals. The sponsorship leaflet, including the available categories and benefits, can be downloaded from: https://deeplearn.irdta.eu/2027/sponsors/ ACCOMMODATION Accommodation suggestions will be made available in due time at https://deeplearn.irdta.eu/2027/accommodation/ CERTIFICATE Participants will receive a certificate documenting 40 hours of academic activities. This certificate should be suitable for participants seeking ECTS or equivalent academic recognition from their home institutions. QUESTIONS AND FURTHER INFORMATION david@irdta.eu ACKNOWLEDGMENTS Juraj Dobrila University of Pula IRDTA – Institute for Research Development, Training and Advice |
|