Synthetic DNA as the future of data archiving
Technological advances and challenges
Marc Antonini - I3S CNRS - Université Côte d'Azur - France
Abstract
The amount of digital data created on the planet is constantly increasing: 90% of it has been generated in the last two years, resulting in exponential consumption of scarce resources and energy, with no absolute guarantee of the integrity and durability of its storage. It should be noted that 70% of data is “cold”, i.e. there is virtually no need to use it over a relatively long period (10 years or more). Digital data storage is therefore becoming a challenge for mankind. Recent studies suggest the use of the DNA molecule as a promising new candidate, which could theoretically hold 215 petabytes in a single gram. Any digital information can be synthesized into DNA in vitro and stored in special tiny capsules that offer storage reliability of several hundred years. The stored DNA sequence can be retrieved at any time using special machines called “sequencers”. In this presentation, we will discuss the state of the art in DNA data storage for the efficient encoding of digital data in a quaternary code consisting of the 4 DNA bases A (Adenine), T (Thymine), C (Cytosine) and G (Guanine). We will also present a promising new solution for encoding digital images in synthetic DNA that we have developed at the I3S laboratory, and which takes into account the constraints associated with storing data on DNA while optimizing the compromise between compression quality and synthesis cost.
Bio
Marc Antonini is Research Director at CNRS and is leading the MediaCoding research team at I3S laboratory in Sophia Antipolis (France). He has held numerous leadership positions in the signal theory community. He is a driving force behind the European project “OligoArchive” on the issue of information storage on DNA. Since October 2021, he is the Program Director of the PEPR Exploratoire “MoleculArXiv” on Massive data storage on DNA and artificial polymers, funded by France 2030. Marc Antonini is since 2020 the chair of the international JPEG DNA working group for the definition of an image coding standard specific to storage on synthetic DNA. He is the author of more than 300 papers, 7 book chapters and 13 patents. His research activities cover in particular image and video coding as well as geometric processing and compression of surface meshes and point clouds. He is also interested in the analysis of the information contained by the neural code in the visual system, with bio-inspired applications in image and video compression. Since several years he started an activity on the storage of digital data in synthetic DNA.
His works on wavelet transform have been included in the JPEG2000 image coding standard. From 1995 to 2001, he was involved with CNES Toulouse in the Earth Observation program “Pléiades” for the definition of the on-board image coder. The image analysis solution he developed with his research group was implemented inside the image sensor on board the satellite (first launch in 2011).
He is co-founder and scientific advisor of Cintoo a company he created in July 2013, spin-off from University Côte d’Azur and CNRS. Cintoo develops technologies and solutions for managing and leveraging the 3D data coming from Reality Capture devices in the cloud. The company is growing, has done 3 fundraisings and now has 40 employees with an agency in the United States.
He is co-founder an Scientific Director of PearCode a company he created in October 2022, spin-off from University Côte d’Azur and CNRS. PearCode addresses private and public organizations willing to archive digital data by offering a low-carbon molecular storage solution using synthetic DNA which ensures the durability of storage, data integrity, and security.
He is Associate Editor for the IEEE Transactions on Image Processing (TIP) since 2021 and has been Associate Editor for the Eurasip Journal on Image and Video Processing (JIVP) from 2012 to 2021. He received the medal of Université Côte d’Azur in 2013 and 2021 respectively. He is a member IEEE. He is a member of AFNOR.
Graph-based machine learning in the metaverse era
Laura Toni - Imperial College London - UK
Abstract
A major challenge for the metaverse is to design virtual and augmented reality systems for real-world use cases such as healthcare, entertainment, e-education, and high-risk missions. This requires immersive systems that operate at scale, in a personalized manner, remaining bandwidth-tolerant whilst meeting quality and latency criteria. With this goal in mind, in this talk, we provide an overview of our current research on graph-based data processing for immersive communications and systems. After an overview of the open challenges in virtual and augmented reality systems, and an introduction of graph-based processing (e.g., graph signal processing), we then present our main contributions in two main directions. As first topic, we present our graph-based learning architecture aimed at predicting dynamic point clouds in the challenging case of deformable 3D objects, such as human body motion. To handle deformable shapes, we propose a graph-based approach that learns and exploits the spatial structure of point clouds and learns more representative features. We then show the possible impact of this movement studies on computer graphics as well as point cloud compression. As a second topic, we present our study on the behaviour of interactive users in immersive experiences and its impact on the next-generation multimedia systems, with the ultimate goal in mind that immersive systems have to put the interactive user at the heart of the system rather than at the end of the chain. We present novel graph-based tools for behavioural analysis of users navigating in 3-DoF and 6-DoF systems, we show the impact and advantages of considering user behaviour in immersive systems. We then conclude with open challenges in immersive systems (and beyond) in which graph-based machine learning could help.
Bio
Laura Toni is an associate professor in the Department of Electronic and Electrical Engineering at University College London (UCL). She received her PhD degree in electrical engineering in 2009 from the University of Bologna, Italy. She was a Post-Doc at the University of California at San Diego (UCSD) from 2011-2012 and at the Swiss Federal Institute of Technology (EPFL), Switzerland from 2012-2016. Her major contributions are in the area of large-scale signal processing for machine learning, graph signal processing, decision-making strategies under uncertainty, and multimedia processing. She has (co)-authored over 60 high-impact publications, and she is co-inventor of 2 patents on low-delay video processing and streaming. She is significantly involved in scientific committees of world-leading conferences/journals (e.g., Program Chair of ACM MM 2022, general chair of ACM MMSys 2019). She recently received the UCL Future Leadership Award and Cisco Academic grant and online optimization on irregular domains with application to smart cities. She also received the Adobe System academic donation on graph-based processing for point clouds. Since at UCL (2016), she has been PI /Co-PI in over 6 projects sponsored by EPSRC, Royal Society, and industrial partners with cumulative funding for my research exceeding £600k as PI, all centred around media processing and online learning on irregular domains.
Advanced blind signal restoration
Moncef Gabbouj, Tampere, University, Tampere, Finland
Abstract
Advanced machine learning pushed further the current state of the art in blind signal restoration. In this talk, we exploit three major applications of blind signal restoration, which aim to restore any signal corrupted by a blend of artifacts with varying types and severities. First, we propose a joint model for blind X-ray image restoration and classification following a Restore-to-Classify Generative Adversarial Networks (R2C-GANs) approach. This is a pioneer approach where the classification improves the restoration performance and vice versa. In other words, it is a “goal-oriented” blind restoration scheme where the aim is to maximize the classification performance after the restoration. The jointly optimized model aims to remove the artifacts while preserving any disease-related features after restoration. A related Cycle-Consistent Generative Adversarial Networks (Cycle-GANs) is used for blind ECG restoration to improve the signal quality to a clinical-level ECG regardless of the type and severity of the artifacts corrupting the signal. Finally, an Operational Generative Adversarial Networks (Op-GANs) with temporal and spectral objective metrics is proposed for blind restoration of real-world audio signals in order to enhance the quality of restored audio signal regardless of the type and severity of the corrupting artifacts. The 1D Operational-GANs exploit the generative neuron model of the Self-Organized Operational Neural Networks and optimized for blind restoration of any corrupted audio signal.
Bio
MONCEF GABBOUJ received his BS degree in 1985 from Oklahoma State University, and his MS and PhD degrees from Purdue University, in 1986 and 1989, respectively, all in electrical engineering. Dr. Gabbouj is a Professor of Information Technology at the Department of Computing Sciences, Tampere University, Tampere, Finland. He was Academy of Finland Professor during 2011-2015. His research interests include Big Data analytics, artificial intelligence, machine learning, pattern recognition, and video processing and coding. Dr. Gabbouj is a Fellow of the IEEE and member of the Academia Europaea and the Finnish Academy of Science and Letters. He is the past Chairman of the IEEE CAS TC on DSP and committee member of the IEEE Fourier Award for Signal Processing. He served as associate editor and guest editor of many IEEE, and international journals and Distinguished Lecturer for the IEEE CASS. Dr. Gabbouj served as General Co-Chair of IEEE ISCAS 2019, ICIP 2020, ICIP 2024 and ICME 2021. Gabbouj is Finland Site Director of the USA NSF IUCRC funded Center for Big Learning and led the Artificial Intelligence Research Task Force of Finland’s Ministry of Economic Affairs and Employment funded Research Alliance on Autonomous Systems (RAAS).