Feel free to contact us! | secretariat@2023.ic-dsp.org

2023 24th International Conference on
Digital Signal Processing

11-13 June 2023 · Island of Rhodes · Greece

Special Session Title: Recent Advancements in Federated Learning and Applications

Organisers: Dimitris Ampeliotis, Ionian University, Greece
Aris Lalos, Industrial Systems Institute, “ATHENA” Research Center, Greece


The scope of Federated Learning (FL) is to train a machine learning model, for instance a deep neural network, using multiple datasets that each resides on a local computer/node in a connected network, without explicitly exchanging data samples. In a typical FL scenario, each local node trains a model using their local dataset and in the sequel all such local models are transmitted to a central node termed as the parameter server. The parameter server utilizes a proper combination rule to compute a global model that, hopefully, incorporates all information from the local models, and this global model is sent back to the local nodes to initiate a new iteration. Thus, FL allows self-interested data owners to train machine learning models collaboratively and facilitates end-users to become co-creators of AI solutions.

The applications of Federated Learning span a wide area of industries, ranging from connected and autonomous vehicles, telecommunications and IoT to pharmaceutics and computer assisted diagnosis systems.

Federated learning offers also the possibility to overcome some limitations associated with the limited computing and storage resources of local computing nodes, thus enabling the computation of models effectively trained over massive datasets. On the other hand, several challenges such as the operation of the FL paradigm when the local datasets have different statistical properties (e.g., non i.i.d. case) or data privacy and trustworthiness issues are raised.
In this special session, we aim to publish the latest advances in trustworthy federated learning which are developed for or evaluated in the field of signal, image, multimedia processing and analysis. It aims to stimulate interdisciplinary thinking and promote developments that can help steer this field towards an even more socially responsible trajectory.
The special session solicits articles in, but not limited to, the following area topics:
● Novel combination rules in federated learning
● Personalized Federated Learning
● Privacy and/or security issues in federated learning
● Decentralized Federated Learning
● Applications of Federated Learning in Image, Video and Multimedia Analysis
● Fairness-Aware Federated Learning
● Client Selection in Federated Learning
● Data Selection in Federated Learning
● Federated Learning in cyber physical systems
● Homomorphic Encryption based Federated Learning

Biography of the organizers:

Dr Dimitris Ampeliotis (ampeliotis@ionio.gr) is an Assistant Professor at the Digital Media and Communication Department, Ionian University, Greece, since March 2021. He received the Diploma degree in computer engineering and informatics, the M.Sc. degree in signal and image processing systems, and the Ph.D. degree in computer engineering and informatics / signal processing, all from the University of Patras, Greece, in 2002, 2004 and 2009, respectively. During the period 2010-20, he worked as an adjunct faculty member at the University of Patras and the Technological Educational Institute of Missolonghi/Western Greece. Dr. Ampeliotis has also worked in several European and National research projects, at the Research Academic Computer Technology Institute (RACTI), the University of Patras, the National Observatory of Athens and the National and Kapodistrian University of Athens. His research interests are in the areas of statistical signal processing, machine learning and optimization theory, with an emphasis on distributed approaches. Dr. Ampeliotis is a member of the Technical Chamber of Greece, a member of the European Association for Signal Processing (EURASIP) and a member of the IEEE Signal Processing Society. More information is available at https://dimitris-ampeliotis.github.io/

Dr. Aris S. Lalos (lalos@athenarc.gr) received the Diploma, M.A.Sc., and Ph.D. degrees from the Computer Engineering and Informatics Department (CEID), School of Engineering (SE), University of Patras (UoP), Patras, Greece, in 2003, 2005, and 2010, respectively. He was a Research Fellow with the Signal Processing and Communications Laboratory, CEID, SE, UoP, from 2005 to 2010. He was a Research Fellow in signal theory and communications (TSC) with the Department of the Technical University of Catalonia (UPC), Barcelona, Spain, from October 2012 to December 2014. From October 2011 to October 2012, he was a Telecommunication Research Engineer with Analogies S.A., an early stage start-up. Since 2015, he has been a Research Fellow with the Visualization and Virtual Reality Group. In May 2018, he was elected as a Principal Researcher (an Associate Research Professor Level with tenure) at Industrial Systems Institute, “ATHENA” Research Centre. He is the author of 130 research papers in international journals (43), conferences (82), and book chapters (five). His current research interests include digital communications, adaptive filtering algorithms, geometry processing, wireless body area networks, and biomedical signal processing. He has participated in more than 18 European projects related to the ICT and eHealth domain as a Project Coordinator (one), a Technical Coordinator (one), a WP Leader (eight), a Senior Researcher (15), and a Researcher (three). He received the Best Demo Award from IEEE CAMAD 2014, the Best Paper Award from IEEE ISSPIT 2015, and the World’s FIRST 10 K Best Paper Award from IEEE ICME 2017. In January 2015, he was nominated as an Exemplary Reviewer of the IEEE Communications Letters. He is a regular reviewer of several technical journals. After 2018 he formed his own group ( http://mips.isi.gr ) at ISI that now counts 10 active members (post-docs, PhD students, and programming engineers) in research and development projects. After 2020 he serves as a president of the Scientific Council of Industrial Systems Institute.