Jean-Christophe Pesquet (IEEE Fellow 2012, EURASIP Fellow 2022) received the engineering degree from Supélec in 1987, the Ph.D. and HDR degrees from the University Paris-Sud in 1990 and 1999, respectively. From 1991 to 1999, he was a Maître de Conférences at the University Paris-Sud. From 1999 to 2016, he was a Professor with the University Paris-Est, and from 2012 to 2016, he was the Deputy Director of the CNRS Laboratoire d’Informatique of the university. He is currently a Distinguished Professor with CentraleSupélec, University Paris-Saclay and the Director of the CVN (Inria team). He was also a senior member of the Institut Universitaire de France from 2016 to 2021. In 2005, J.-C. Pesquet was technical chairman of ICASSP and he is also technical chairman of ICIP 2022. He was a member of the SPTM technical committee (2000-2005 and 2011-2016) and served as an associate editor for IEEE SPL (2004-2006). He was an associate editor for IEEE TSP (2009-2013), a senior area editor for the same journal (2010-2015), and a member of the committee for the best paper award of EURASIP JASP (2007-2019). He is now an associate editor of SIAM Journal on Imaging Sciences. His research interests are focused on optimization methods in data science.
Fixed Point Strategies in Signal and Image Processing
Fixed point strategies provide a simplifying and unifying framework to model, analyze, and solve a great variety of problems in signal and image processing. They constitute a natural environment to explain the behavior of advanced convex optimization methods. In addition, an increasing number of problems go beyond optimization since their solutions are not optimal in the classical sense of minimizing a function but, rather, satisfy more general notions of equilibria. Among the formulations that fall outside of the realm of standard minimization methods, we can mention variational inequality and monotone inclusion models, game-theoretic approaches, neural network structures, and plug-and-play methods. This talk will provide an overview of the main tools of fixed point theory and discuss some of their applications to machine learning and inverse problems. The prominent role played by the forward-backward algorithm will be emphasized.
Dr. Ahmed Tewfik is currently at Apple. He served as the ECE as department of electrical and computer engineering chair at the University of Texas at Austin from Fall 2010 through 2019. Tewfik combines a distinguished academic research record, entrepreneurial insight as the founder and CEO of a software company and leadership experience gained through various positions with IEEE. His education includes a B.S.E.E. from Cairo University, Cairo Egypt, in 1982, an M.S.E.E. in 1984 and Sc.D. in 1987, both from the Massachusetts Institute of Technology. Prof. Tewfik is a Fellow of the IEEE. He received several awards and honors, including the IEEE Signal Processing Society Technical Achievement Award in 2018, IEEE third Millennium award in 2000 and Distinguished Lecturer of the IEEE Signal Processing Society in 1997 – 1999. He was elected to the post of President-elect of the IEEE Signal Processing Society in 2017 and served as President in 2020 and 2021.
The age of the omnipresent and omnipotent virtual assistant
Accelerating hardware and machine learning advances are ushering the age of ever more present and capable virtual assistants in all aspects of human life, from home automation to driver and surgery assistants. This talk will review recent progress and technology trends. It will discuss opportunities and challenges that this new era is bringing to the profession. What problems and tasks are best handled by machines? What is the optimal role of the human in the symbiotic virtual assistant-human partnership? Do most scientist and engineers in general have the right education and background to design human-machine systems?
Olgica Milenkovic is the Franklin W. Woeltge endowed chair of the Electrical and Computer Engineering department at the University of Illinois, Urbana-Champaign (UIUC). She obtained her MS degree in Mathematics in 2001 and PhD in Electrical Engineering in 2002, both from the University of Michigan, Ann Arbor. Her scholarly contributions have been recognized by multiple awards, including the NSF Faculty Early Career Development (CAREER) Award, the DARPA Young Faculty Award, the Dean’s Excellence in Research Award, and several best paper awards. In 2013, she was elected a UIUC Center for Advanced Study Associate and Willett Scholar while in 2015 she became a Distinguished Lecturer of the Information Theory Society. She was elevated to the level of IEEE Fellow in 2018. During her academic career, she served as Associate Editor of the IEEE Transactions of Communications, the IEEE Transactions on Signal Processing, the IEEE Transactions on Information Theory and the IEEE Transactions on Molecular, Biological and Multi-Scale Communications. She was the Guest Editor in Chief of a special issue of the IEEE Transactions on Information Theory on Molecular Biology and Neuroscience and the special issue of the IEEE Transactions on Information Theory in Memory of V. I. Levenshtein. She was also the plenary speaker at a number of conferences and workshop, including the recent Information Theory Workshop in Visby, Sweden (2019) and International Symposium on Information Theory in Los Angeles, USA (2020).
Signal processing, Learning and Coding for DNA-Based Data Storage
Molecular storage, and in particular, DNA-based data storage, is an emerging synthetic biology paradigm that offers ultradense nanoscale information recording and unique in-memory computation capabilities. The use of DNA molecules as storage media also ensures extreme robustness/durability, ease of information copying/replication and random access via standard polymerase chain reactions. The main issues that stand in the way of practical deployment of DNA-based storage platforms are the high cost of information recording (i.e., DNA synthesis) and the large latency of both the read and write processes. We present an overview of our recent work on two-dimensional and DNA flash storage platforms that utilize both the sequence and topological dimensions of the molecular media. These systems also limit the use of costly solid-state synthesis and enable new length-modulation schemes via enzymatic chain growth and positional encoding functions. The development of such multidimensional molecular storage technologies is coupled with many new signal processing, learning and coding challenges. To describe how to address these challenges, we first provide an overview of the various signals encountered during content readout, such as analog raw current signals and stochastic “population tail signals” generated by nanopores, as well as noisy reads generated by next generation sequencers. We then proceed to outline how to perform signal extraction, identification and classification, and reconstruction of image data through the use of new learning and computer vision techniques and low-redundancy error-control codes. We conclude with a list of experimental results and present open problems in the field.
Regina Barzilay is a School of Engineering Distinguished Professor for AI and Health in the Department of Electrical Engineering and Computer Science and a member of the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology. She is an AI faculty lead for Jameel Clinic, an MIT center for Machine Learning in Health. Her research interests are in natural language processing and applications of deep learning to chemistry and oncology. She is a recipient of various awards including the NSF Career Award, the MIT Technology Review TR-35 Award, Microsoft Faculty Fellowship and several Best Paper Awards at NAACL and ACL. In 2017, she received a MacArthur fellowship, an ACL fellowship and an AAAI fellowship. In 2021, she was awarded the AAAI Squirrel AI Award for Artificial Intelligence for the Benefit of Humanity, the AACC Wallace H. Coulter Lectureship Award, and the UNESCO/Netexplo Award. She received her PhD in Computer Science from Columbia University, and spent a year as a postdoc at Cornell University. Prof. Barzilay received her undergraduate degree from Ben-Gurion University of the Negev, Israel.
Challenges and Opportunities in Clinical AI
The first generation of clinical AI tools comprised direct applications of existing machine learning algorithms to clinical data. While many of the resulting models can successfully analyze patient data, there are many unresolved methodological challenges that impede translation of this technology into the healthcare systems. Examples of such challenges include confidence estimation, recognition of distributional shifts in patient data, and robustness in different application scenarios. In my talk, I will exemplify the need in addressing these challenges and discuss new algorithms we developed to address them.