Prof. Samson Lasaulc
Khalifa University, UAE
Bio: Samson Lasaulce is a CNRS Director of Research with
CRAN at Nancy. Over 2023-2025, he has been a Chief Research
Scientist in AI with Khalifa University (KU), Abu Dhabi
where he was also the holder of the TII 6G Chair on Native
AI. He has been the holder of the RTE Chair on the Digital
Transformation of Electricity Networks at CentraleSupélec.
He has also been a Professor with the Department of Physics
at Ecole Polytechnique. Before joining CNRS he has been
working for five years in private R&D companies (Motorola
Labs and Orange Labs). Dr. Lasaulce is the recipient of
several awards such as the SEE Blondel Medal. Dr. Lasaulce
has been serving as an Associate Editor for several journals
such as the IEEE Transactions on Signal Processing and
Springer Nature Discover Artificial Intelligence. His
current research interests lie in distributed networks with
a focus on game theory, information theory, learning,
distributed optimization, network control for communication
networks, energy networks, and social networks. He is a
co-author of the book "Game Theory and Learning for Wireless
Networks: Fundamentals and Applications".
Speech Title: Towards LLMs on Device: The Case of LLM Model Compression
Abstract: In this task we will briefly review possible techniques to deploy large language models on small devices such as mobile phones. We will put more emphasis on the most recent quantization and adaptation techniques.
Prof. Guillermo De Ita Luna
Autonomous University of Puebla, Mexico
34 years as a Full Professor and researcher in the Computer
Science department at the Autonomous University of Puebla.
(BUAP), México. Currently, a member of the Mexican System of
Researchers at level 3 (the highest level). Guillermo De Ita
has made research stances in Texas A&M, Chicago University,
Lille – Inria France, as well as several Universities in
Mexico. He was the principal of the Computer Science Dept
(BUAP) from 1999 to 2003. He designed the engineering
program in computer science and participated as a founding
member of the master's and doctoral programs in Computer
Science at the Computer Science
Department from BUAP. He has supervised 59 thesis projects;
32 in bachelor ‘s level and 23 in posgrade level. He has
published 140 research articles in journals and conference
proceedings that underwent
rigorous double-blind peer review, along with 30 book
chapters. Additionally, he contributed as an author to the
publication of 5 books.
Prof. Antonios Saravanos
New York University (NYU), USA
Bio: Dr. Antonios Saravanos is Clinical (Full)
Professor of Information Systems at New York University
(NYU). He holds two doctorates, the first from Columbia
University (New York, USA) and the second from Bocconi
University (Milan, Italy), as well as graduate degrees from
the University of Oxford and the University of Cambridge in
the United Kingdom. A Senior Member of the Association for
Computing Machinery, Dr. Saravanos’s research examines the
drivers and barriers of technology adoption, trust, and
sustained engagement, with particular emphasis on
intelligent machines and “good tech”, examining the
influence of prosocial motivation and warm glow on user
intentions and continued use. He also collaborates on
applied AI projects that study learning dynamics in modern
models and develop predictive methods, including work in
medical imaging. In addition to his research, Dr. Saravanos
has played a significant role in curriculum development,
academic leadership, and faculty governance at NYU. He led
the creation of NYU’s first undergraduate "big data" degree,
the Bachelor of Science in Applied Data Analytics and
Visualization, served as its program coordinator from 2016
to 2022, and has been recognized for his efforts with the
NYU School of Professional Studies Outstanding Service Award
(2016) and the Teaching Excellence Award (2019).
Prof. Farid Nait-Abdesselam
Université Paris Cité, France
Bio: Farid Naït-Abdesselam is a Full Professor of Computer Science at Université Paris Cité, with more than two decades of experience in research and higher education. His research lies at the intersection of networking, cybersecurity, and distributed systems, with a particular focus on secure communication systems, network resilience and optimization, intrusion and malware detection, adversary-aware machine learning, and adaptive defense strategies. His work addresses the design of robust and trustworthy mechanisms for securing complex, constrained, and heterogeneous networked environments.
He has authored over 180 peer-reviewed publications, edited two scientific books, and contributed several book chapters on topics including network security, malware forensics, blockchain technologies, and advanced networked systems. His research combines theoretical foundations with the design, implementation, and experimental evaluation of practical security solutions, with applications spanning mobile and wireless systems, vehicular and drone networks, sensor networks, distributed infrastructures, and emerging Internet architectures. His recent research interests include intelligent and adaptive cyber threats, adversarial machine learning, AI-assisted malware analysis, and resilient defense mechanisms for next-generation networked systems.
Speech Title: Rethinking DDoS Detection: The Challenge of Partial Network Visibility
Abstract: Distributed Denial-of-Service (DDoS) attacks continue to evolve in scale, distribution, and sophistication, posing significant challenges to modern network infrastructures. While considerable progress has been made in DDoS detection, many existing approaches rely on the availability of sufficient traffic or host-level observations. In operational networks, however, monitoring is often incomplete, and only a fraction of attack activity may be directly observable.
This invited talk examines DDoS detection from the perspective of partial network visibility. It discusses how incomplete telemetry can affect detection effectiveness and potentially lead to overly optimistic assessments of detector robustness. The talk explores the value of combining observations collected at different levels of the network and discusses visibility-aware detection strategies that adapt to the amount of information available to the defender. The presentation concludes with open challenges and research directions toward more robust, adaptive, and operationally realistic DDoS defense mechanisms.
Assoc. Prof. Kezhi Mao
Nanyang Technological University, Singapore
Bio: Dr. Mao received his PhD from the University of Sheffield in 1998. He is currently an Associate Professor in the School of Electrical and Electronic Engineering at Nanyang Technological University (NTU), Singapore. His research spans several areas of artificial intelligence (AI), including machine learning, computer vision, natural language processing, and information fusion.
As a lead Principal Investigator, Dr. Mao has led and successfully completed more than a dozen externally funded research projects covering a broad spectrum of AI technologies and applications. He has published approximately 140 papers in refereed international journals and conferences, attracting nearly 13,000 citations on Google Scholar. A strong advocate of translational research, he has led the development and deployment of AI-enabled systems and tools for government agencies and industry, bridging fundamental research with real-world operational needs.
Beyond academic research, Dr. Mao has advised multinational corporations on the adoption and practical application of state-of-the-art AI technologies. He is currently an Action Editor of Neural Networks and an Associate Editor of Neurocomputing. He has also contributed extensively to the international research community through leadership and invited roles at international conferences, including serving as General Chair and Conference Co-Chair, as well as a keynote/invited speaker.
Speech Title: "Image-Based Depth Estimation: From Robust Self-Supervision to Foundation-Model-Assisted Depth Learning"
Abstract: Depth estimation from images is fundamental to 3D scene understanding, robotics, autonomous systems, and intelligent sensing. Despite rapid advances in deep learning, reliable depth estimation in real-world environments remains difficult due to textureless regions, ambiguous object boundaries, illumination variations, sensor noise, and incomplete depth measurements.
In this talk, I will present our recent work addressing these challenges from three perspectives. First, GAM-Depth develops a self-supervised approach for indoor depth estimation, combining gradient-aware supervision with semantic constraints to improve depth prediction in textureless regions and at object boundaries. Second, LumiDepth addresses monocular depth estimation under spatially varying illumination, where shadows, specular highlights, and exposure changes can significantly distort visual cues. It exploits probabilistic pseudo supervision and frequency-aware consistency and distillation to improve both depth accuracy and stability across illumination conditions. Finally, HFD-Teacher investigates RGB-guided depth completion by distilling high-frequency geometric information from depth foundation models through adaptive multi-scale decomposition and topological constraints.
Together, these studies explore how visual, semantic, frequency, and foundation-model knowledge can be exploited to achieve more accurate and robust depth estimation under challenging real-world conditions.
Assoc. Prof. Ismail Bennis
University of Haute Alsace, France
Bio: Ismail Bennis earned in 2009 a bachelor's degree in mathematics and computer science from the Université Mohammed V in Rabat, Morocco. 2011, he received a master's degree in Computer Networks and Telecommunications from the same university. He completed his PhD in 2015 under joint supervision between the Université Mohamed V in Rabat, Morocco and the Université de Reims Champagne-Ardenne in France. From 2015 to 2017, he worked as a temporary professor for research and teaching (A.T.E.R) at the University of Reims. Between 2017 and 2020, he worked as an associate professor at La Rochelle University. His research interests include routing protocols with quality of service over wireless sensor networks, IoT and outlier detection. Since September 2020, he has worked as an associate professor at the University of Haute Alsace.
Speech Title: "Towards Smarter LoRaWAN Networks: Insights from Multi-Gateway Deployments, AI-Driven Optimization and V2X Applications"
Abstract: LoRaWAN has established itself as a leading LPWAN technology for large-scale IoT deployments. However, achieving reliable and scalable communications remains challenging in dense networks, urban environments, and emerging mobility scenarios. This talk presents a synthesis of our research contributions on LoRaWAN performance analysis and optimization. We discuss the impact of gateway deployment, traffic patterns, and protocol parameters on network performance, and present solutions developed to improve reliability, scalability, and resource allocation. The talk also introduces reproducible simulation tools designed to facilitate LoRaWAN experimentation and optimization. Finally, we explore new research directions, including AI-driven network optimization and the use of LoRaWAN in vehicular and intelligent transportation systems. The presentation concludes by highlighting the opportunities and open challenges that will shape the next generation of LPWAN-based communications.
Dr. Douglas Schmidt
William & Mary, USA
Bio: Dr. Douglas C. Schmidt is the Dean of the School
of Computing, Data Sciences & Physics at William & Mary,
where he leads initiatives at the intersection of artificial
intelligence, software engineering, and institutional
transformation. An internationally recognized researcher and
educator, his work explores how generative AI reshapes
software development, testing, and human-computer
collaboration, with a particular focus on intent-driven and
human-centered AI systems.
Before joining William & Mary, Schmidt held senior
leadership and faculty roles at Vanderbilt University and
Carnegie Mellon University’s Software Engineering Institute.
He has also collaborated extensively with industry and
government partners developing and testing large-scale,
software-reliant systems. He is a frequent speaker, author,
and advisor on the opportunities and risks of deploying AI
in real-world organizations, education, and critical
infrastructure. In his keynote, Schmidt examines how AI is
not just automating tasks, but redefining expertise, agency,
and the future of knowledge-driven institutions.