QuIK Workshop
The first Quantum Information Knowledge (QuIK) workshop was co-organized by Priya J. Nadkarni, Narayanan Rengaswamy and Bane Vasić. The long-term vision for this workshop is to establish a platform for active discussions on problems in quantum information. The goal of this first edition was to provide foundational knowledge in quantum error correction (QEC) for fault-tolerant quantum computing (FTQC), complement that with exciting talks by invited speakers working in this area, foster discussions on key open problems, both foundational and practical ones, and discuss some of the latest results in the field. The workshop started with a tutorial by the organizers introducing fundamental concepts in QEC and FTQC, targeted at an audience with background in classical coding theory. No prior familiarity with quantum mechanics or quantum computing was assumed. A written tutorial was shared with the participants before the workshop and published at:. The tutorial was followed by a keynote as well as seven invited talks by a diverse set of researchers who are pushing the boundaries of QEC and FTQC. The keynote speaker was Liang Jiang, and the invited speakers were Valentin Savin, Nithin Raveendran, Priya J. Nadkarni, Anirudh Krishna, Armanda O. Quintavalle, Gretchen Matthews and Shayan Srinivasa Garani. The workshop concluded with a stimulating panel discussion involving all speakers, followed by a poster session featuring novel research and key open problems in QEC and FTQC. The workshop was received very positively by the 60 participants who came from various institutions across the world. There were seven accepted papers in the workshop, which were presented as posters and published on 鶹ýAVXplore. Besides, there were three more poster-only submissions presented at the workshop. Further details about the workshop can be found on its website:. Those who are interested in future editions can sign up to the mailing list there.
Workshop on Coding Theory and Algorithms for DNA-based Data Storage
This workshop was jointly organized by academia, represented by Rawad Bitar, Antonia Wachter-Zeh and Eitan Yaakobi and industry, represented by Dave Landsman from the DNA Data Storage Alliance and Western Digital. It invited researchers from the IT and sibling communities and from industry to brainstorm on the pressing research directions pushing further the technology of DNA-based data storage.
The workshop was attended by around 60 participants from around the globe. It featured a keynote by Prof. Robert Grass from ETH Zurich, two invited talks from academia given by Prof. Olgica Milenkovic from UIUC and Prof. Zohar Yakhini from the Technion and Reichman University, and two invited talks from industry given by Daniel Bedau from Western Digital and James Diggans from Twist Bioscience. A key highlight of the workshop was the surprising attendance of the 2024 鶹ýAV President and CEO Thomas Coughlin. In addition, the workshop included four contributed talks and a session of 16 poster presentations during which the participants exchanged information and discussed ongoing research topics. The last part of the workshop was designed as a “think-tank” session: four groups were formed, having the background set by the keynote speaker, each group discussed an open problem posed by one of the invited speakers. Each group presented a summary of their discussion and the workshop was concluded by a social dinner at a Michelin star restaurant allowing the participants to have a more informal chat. The organizers express their gratitude to the 鶹ýAV IT Society and the DNA Data Storage Alliance for their generous financial and logistical support.
As a follow-up, the organizers together with Olgica Milenkovic and Moshe Schwartz are planning a JSAIT special issue on “Information and Coding Theory Aspects of DNA-based Data Storage.”
Participants and interested researchers are warmly encouraged to submit their original research and inquire for potential tutorial papers on the topic.
The feedback from the participants was overwhelmingly positive. The organizers cordially thank all participants for contributing to the success of this workshop. In particular, we thank the invited speakers who put a lot of effort into perfectly tuning their talks to the diverse audience, and all the participants, especially from Biomemory and the JPEG standardization committee, who freed their busy schedules and traveled to Athens to participate in this workshop. We are grateful to everyone who contributed a talk and/or a poster at the workshop.
A full program with a list of invited speakers, poster presenters, titles of the talks and posters, and abstracts are available
Workshop on IT-ML
The inaugural Information-Theoretic Methods for Trustworthy Machine Learning (IT-TML) workshop was successfully co-organized by Shahab Asoodeh, Flavio Calmon, Oliver Kosut, and Lalitha Sankar. The workshop served as a collaborative platform for researchers, engineers, and practitioners to come together and explore the challenges and opportunities involved in deploying machine learning systems responsibly, particularly in applications with significant societal impact. The discussions placed a strong emphasis on key topics such as privacy and fairness, which are critical pillars for building trustworthy AI systems.
The workshop featured two insightful tutorials delivered by Peter Kairouz and Kush Varshney. Following the first tutorial, attendees enjoyed three engaging invited talks by Sanghamitra Dutta, Catuscia Palamidessi, and Mario Diaz, complemented by three lightning spotlight presentations that provided concise overviews of emerging ideas. The second tutorial was followed by two more invited talks, given by Haewon Jeong and Ananda Suresh Theerta, both of whom shared valuable insights on current challenges and solutions in the field.
The day concluded with a vibrant poster session showcasing 12 posters, where participants actively exchanged ideas, discussed ongoing research, and explored potential collaborations. The diversity of topics and enthusiasm of the presenters made the session an excellent way to wrap up the workshop.
Approximately 50 participants attended the workshop, and the feedback was overwhelmingly positive. The organizers extend their heartfelt thanks to all participants for their invaluable contributions. A special note of gratitude goes to the invited speakers for delivering talks that were perfectly tailored to a diverse and interdisciplinary audience. We also appreciate the efforts of all contributors who presented talks and posters, making this event both engaging and impactful.
For those interested in exploring the workshop in more detail, the full program, including a list of invited speakers, poster presenters, titles, and abstracts, is available
Workshop on Learn to Compress
The "Learn to Compress" Workshop at ISIT 2024, organized by Professor Aaron Wagner fromCornell University, Professor Elza Erkip from NYU T andon, and her PhD student Ezgi Ozyilkan, brought together experts from information theory, ML, and computer science to explore the evolving landscape of learning-based data compression. The workshop saw significant participation from academia and industry, featuring distinguished researchers from Google Research and Google DeepMind, along with attendees from industrial research labs, including Nokia Bell Labs. The presence of first-time ISIT attendees from the ML community underscored the workshop's mission to strengthen connections between the Information Theory Society (ITSoc), industry, and the ML community.
The technical program showcased modern approaches to data compression, including channel simulation/synthesis as an alternative to traditional quantization in practice-oriented applications. Notably, Dr. Ruida Zhou (UCLA) received the best paper award for his work on common randomness in source coding and its relation to perceptual quality. Numerous papers exploring rate-distortion-perception trade-off generated engaging discussions about balancing information-theoretical foundations with ever-present practical implementation challenges. The workshop was timely given the recent shift in lossy compression/source coding research toward ML and generative AI approaches, including variational autoencoders, generative adversarial networks (GANs), and diffusion models. While driven largely by the ML community, the workshop represented a strategic initiative to reconnect this rapidly evolving field with its information-theoretic roots while simultaneously embracing modern AI-driven approaches. The diverse format, which included keynote talks, oral presentations, and a poster session, fostered dialogue between ISIT regulars and first-time attendees from the ML community. This vibrant exchange of ideas represented an important step toward bridging the gap between classical information theory, which has long underpinned compression,, and contemporary ML techniques, which now offer powerful new tools for implementation and optimization.
Building on this momentum, the "Machine Learning and Compression" Workshop at NeurIPS 2024 further fostered connections between the traditionally separate research communities of ML and information theory. Focusing on shared interests in learning-based and modern approaches to data compression and source coding, and their increasing relevance with the rise of LLMs and efficient AI, these two workshops point towards a sustained and growing interaction between these two communities.
Workshop on NeurIT
In fundamental and clinical neuroscience, there is an enormous opportunity for information theory to lay a foundation for understanding of the principles of information representation, communication and processing in the brain, in health and disease. In neuroengineering, the explosion of technologies for high-resolution sensing and stimulation of the brain (a Moore’s law of neuroengineering) is a game-changer for neuroscience and clinical practice. These opportunities require addressing new fundamental challenges in sensing, stimulation, and inference. To address these challenges in neuroscience and neuroengineering, the NeurIT 2024 workshop united computational neuroscientists, neuroengineers, and information theorists. This interdisciplinary event facilitated vibrant discussions on advancing both foundational understanding and technological innovation in these fields.
The workshop opened with a compelling keynote by Nihat Ay from Hamburg University of Technology, who presented an “Information-Geometric Approach to Brain Complexity,” discussing new information measures that arise due to examining these questions. Morning sessions focused on foundational topics, including quantifying brain-region interactions, high-order interdependencies, and reverse-engineering neural circuits, with speakers from the Allen Institute, Aalborg University, Carnegie Mellon, EPFL, and Boston University.
Afternoon sessions emphasized applications, exploring how information theory has informed, and can inform, neuroengineering and inference advancements. Talks highlighted directed information flow in neuron-based computing systems, causal brain-muscle mappings, latent neural population channels, and modeling cognitive decline as faulty computing, with contributions from Yale, Imperial College, UIUC, University of Bern, Dalian University, and Carnegie Mellon.
The workshop concluded with an open and interactive discussion synthesizing insights and charting future directions, including new foundational questions for information theory posed by these applications. By addressing both opportunities and potential misapplications of information theory, NeurIT 2024 underscored its critical role in understanding the brain’s mechanisms and driving neuroengineering innovations with broad societal impact.