S P E A K E R S

Corina FORĂSCU

A. Professor at the Faculty of Computer Science, at Alexandru Ioan Cuza PhD from Faculty of Computer Science, Alexandru Ioan Cuza University of Iași Coordinator of the Erasmus and International Relations
Fullbright-RAF scholarship

E-mail: corina.forascu@gmail.com

Title:

EMPOWERING THE WORLD AND AFRICA
Harnessing Natural Language Processing for Research, Innovation and Sustainability 

 

Abstract

The talk will give an overview on how Natural Language Processing (NLP) is revolutionizing research, entrepreneurship and sustainable development, with a focus on Africa and Tunisia. The participants will be able to discover and maybe better understand what NLP is and, meanwhile, the transformative impact of language resources &technologies – by enabling computers to understand, process, interpret and generate human language – on business innovation, social initiatives, and environmental sustainability across the world and the African continent.
The first part of the talk will be focused on the main LRTs: what they are, what should be the main objectives for a given language on this domain, how can they be created, inter-connected and exploited in a practical mono and/or multi-lingual setting.
Then, the main objective will be on those NLP applications spanning various domains, including social media, business, research, and environmental initiatives, driving entrepreneurship and innovation to the next level, especially in Africa.


Keywords:  Computational Linguistic, Computer science, Entrepreneurship & Innovation in IT, Phonetics, Psycholinguistics, and Sociolinguistics

Biography

Corina Forăscu is an Associate Professor at the Faculty of Computer Science, Alexandru Ioan Cuza University of Iași, responsible for outreach activities, managing the faculty’s social media presence, and recently appointed as the Erasmus and International Relations Coordinator for the faculty.

By nature, Corina is a dedicated volunteer, actively involved in causes close to her heart – ranging from nature conservation, biodiversity, bird protection, and climate change to education and science for all, particularly for teenagers. By training, she is a mathematician, who transitioned into computational linguistics through her master’s and Ph.D. studies. Following an exceptional Fulbright-RAF scholarship and, more recently, two courses at Babson University, Corina has committed to bringing entrepreneurship into various fields: in education – at the Faculty of Computer Science, where she developed the first course on Entrepreneurship & Innovation in IT, in high schools, in the broader community, but most importantly, into the mindset of the people she networks with.


In her free time (if any), Corina enjoys traveling, taking nature photographs, birdwatching, reading, playing the piano, spending time with her pets, or volunteering for the local community.

 

 

Dr. Giansalvo CIRRINCIONE


Senior IEEE Member

HDR at the University of Picardie Jules Verne, Amiens-france.

PhD in Cognitive Science from INPG

Electrical Engineering from Politecnico di Torino, Italy

E-mail: exin@u-picardie.fr

 

Title:

A novel approach to deep clustering for very high dimensional data

Abstract

This talk introduces an innovative integration of deep learning and competitive Hebbian learning, tailored for high dimensional data, even in case of small datasets.

The presentation will begin with a brief introduction to deep learning for clustering, providing afoundationforunderstanding its advantagesand limitations in unsupervised tasks. This setsthestageforthe coreof the talk: a »dualcompetitive layer »trained onthe transposed input matrix, which complements the traditional « vanilla » competitive layer.

Theduallayerisshowntodirectlyoutputdataprototypesandexcelsinhigh-dimensional data contexts, where it overcomes challenges associated with input dimensionality.

Advantages of Perfect Integration with Deep Learning:

  1. Gradient Flow Compatibility: The dual competitive layer seamlessly integrates with backpropagation, preserving the advantages of deep learning architectures.
  2. EnhancedFeatureExtraction: By combining gradient-based and competitive learning, the method retains deep learning’s hierarchical feature mapping while replicating the input data’s topology.
  3. Scalability to Complex Architectures: The dual layer’s design supports integration into multi-layered deep networks, enabling advanced applications requiring intricate clustering mechanisms.
  4. Efficient Training for High-Dimensional Data: The dual layer reduces computational overhead by constraining weight estimation to dataset cardinality rather than dimensionality.
  5. Prototype-Based Learning: Its ability to output data prototypes enhances interpretability and supports applications requiring clear clusterrepresentations.
  6. Broad Applicability: The dual layer’s compatibility with deep architectures extends its utility to diverse domains, including biological data, computer vision, and high-dimensional unsupervised learning challenges.

EngineeringApplications:

The talk will conclude with proposals for engineering applications of this novel approach. Pote  ntial use cases include:

  • Fault Detection and System Optimization in Power Systems:
    • The ability to handle high-dimensional data makesthis approach ideal for analyzing data from comple Prototype-based clustering can identify typical operational states and anomalies, aiding in predictive maintenance and fault detection.
    • It can assist in optimizing performance and identifying patterns in renewable energy generation (e.g., solar or wind) by clustering high- dimensional sensor data.
  • BiomedicalDataAnalysis:Addressinghigh-dimensionalclusteringchallenges in genomic and proteomic datasets.
  • AutonomousSystems: Improving unsupervised learning capabilities in robotics and AI-driven control systems.

 

By blending theoretical insights, experimental validations, and real-world applications, this talk highlights the potential of this novel clustering framework to advance engineering and scientific domains.

  • xpowergrids, wherenumerousvariablesneed to be monitored.

Keywords:i Computational Linguistic, Computer science, Entrepreneurship & Innovation in IT, Phonetics, Psycholinguistics, and Sociolinguistics Deep Learning, Transformers,

Biography

Dr. Giansalvo Cirrincione is an Associate Professor HDR  at the University of Picardie Jules Verne, Amiens, France, and a Senior IEEE Member. He holds a PhD in Cognitive Science from INPG, France, and a Laurea in Electrical Engineering from Politecnico di Torino, Italy. His research spans neural networks, artificial intelligence, renewable energy systems, and power electronics, with practical applications in diagnostics, control systems, and electrical drives. With over 200 publications, Dr. Cirrincione has made significant contributions to clustering algorithms, system identification, and deep learning. His work bridges theoretical advances with engineering solutions, including applications in renewable energy and medical diagnostics. He has received multiple awards, including Best Paper distinctions, and continues to collaborate internationally as a visiting professor and researcher.

Dr. Sami AYARI

Project Director and Senior Technical Expert

Specializing in IT organization and data transformation at BNP Paribas.

Doctor in Electrical Engineering, Electronics, and Automation; Paris-Saclay

Principal engineer in electrical engineering from ENI of Tunis

E-mail: sami.ayari@tunisian-ai-society.org

 

Title:

Investing in Local AI: A Catalyst for Strengthening Digital Sovereignty

Abstract

For Tunisia, investing in local artificial intelligence (AI) represents a strategic opportunity to strengthen its digital sovereignty and reduce dependency on major global technological powers. This investment relies on the development of a dynamic national ecosystem, supported by an ambitious national strategy for AI. This strategy should encourage innovative startups, promote the training of local talents, and facilitate the creation of solutions tailored to national priorities such as health, agriculture, education, or energy.

By developing AI locally, Tunisia can safeguard its sensitive data while designing technologies customized to its specific needs. This requires joint investments from both the public and private sectors, the establishment of a robust digital infrastructure, and the development of favorable regulatory frameworks.

This approach aims to bolster the country’s technological independence, stimulate the national economy, attract international investments, and position Tunisia as a key regional player in the field of AI. For 2025, priority actions and new technological strategies should be based on a holistic approach. Leveraging the lessons learned from past strategies and mobilizing necessary resources, Tunisia can address technological challenges and fully harness the potential of AI to ensure sustainable development.

Keywords: Artificial Intelligence, Digital Sovereignty, Tunisia, Technological Independence, National Strategy, Local Talent Development, Innovation Ecosystem

Biography

Sami Ayari is a Project Director and Senior Technical Expert specializing in IT organization and data transformation at BNP Paribas. With extensive experience, he has led and contributed to numerous technological transformation and strategic process optimization projects, notably within the Data Factory of BNP Paribas IT Group and other key divisions of the organization.

Holding a doctorate in Electrical Engineering, Electronics, and Automation (EEA) from École Normale Supérieure Paris-Saclay (formerly Cachan), Sami Ayari also earned a Diploma of Advanced Studies (DEA) in Electrical Engineering from CentraleSupélec (formerly Supélec) in France. His DEA thesis, completed in 1996, focused on designing a neural network-based algorithm for detecting and analyzing faults in a power transmission network. This exemplary academic background is rooted in a solid initial education in electrical engineering, achieved at the National Engineering School of Tunis (ENIT), where he graduated as a principal engineer in 1995.

In parallel with his professional career, Sami Ayari is deeply committed to initiatives aimed at uniting and showcasing Tunisian talent on a global scale. He is the co-founder and general coordinator of the Tunisian AI Society and Tunisia Cybershield, platforms that bring together leading Tunisian experts in artificial intelligence and cybersecurity worldwide.

He is also the founder and president of the association RECONNECTT, which works to promote the professional success of Tunisian executives both in Tunisia and abroad. The association is also actively involved in advancing Tunisian women in STEM fields (science, technology, engineering, and mathematics), thereby increasing their visibility and strengthening their impact in sectors still largely dominated by men.

Prof. Gérard POISSON


Professor Emeritus at University of Orléans,

Member of Prisme Laboratory,

Bourges Institute, France

Main research activity, robotized tele-echography concept

E-mail: gerard.poisson@univ-orleans.fr

 

Title :

                                    FORTY YEARS OF MEDICAL ROBOTICS:

  • Main developments and current challenges

  • A special look at robotized tele-echography

 

Abstract

In this presentation, we’ll be looking at the evolution of technological solutions, in terms of both mechanics and control, for the development of robotic devices for the medical sector, since robotics first entered the field in the early 1980s. At that time, various projects were devoted to transferring industrial robotics approaches to a large variety of medical applications, particularly in surgery.

One of the first references in this field was the performance, in 1985, of brain biopsies on some twenty patients in California, using a6 degrees-of-freedomindustrial-likerobot.A wide range of robot-assisted medical procedures then emerged in neurosurgery, orthopedics and abdominal surgery. While the first applications were based on the transfer of industrial technologies, dedicated kinematic architectures were developed to meet the specific needs and constraints of the medical sector, with a growing range of applications inminimally invasive abdominal, cardiac or ENT surgery, needle insertion and tele-echography.

Our team was a pioneer in tele-echography, developing and validating, in 1998, the world’s first tele-operated robot for long-distance ultrasound diagnosis. We’ll be taking a closer look at how this technology, initially conceived for space applications, has evolved into tele-operated diagnostics for high-risk areas, to provide diagnostic solutions in secondary hospitals lacking experts, and in the future for cancer detection and monitoring through the fusion of ultrasound and positron emission tomography.

Keywords: Artificial Intelligence, Digital Sovereignty, Tunisia, Technological Independence, National Strategy, Local Talent Development, Innovation Ecosystem

Biography

Prof. Gérard Poisson is a graduate of the Mechanical Agrégation in 1980, as student of École Normale Supérieure deCachan, France.He received the Ph.D. degree in robotics from university of Orléans in 1994, and the Habilitation à Diriger des Recherches (HDR), from the latter university in 2004. He hasover 40 years of teaching in mechanical engineering, automatics, and robotics in the Institute of Technology (IUT) in Bourges, where he was also the head, from 2008 to 2018.He is author of 130 publications in refereed journals and international conferences,andhas more than 100 participationsin Ph.D. defenses.

He was one of the founders, in 1988, of the Vision & Robotics Laboratory, now known as Prisme, where he has been working ever since.His research has focused on several aspects on the themes of robotic design and control, for industrial, agricultural, and mainly medical applications. Its works included perception of the environment, telemetry, image processing and data fusion, wheeled mobile robots design and navigation, poly-articulated and actuated mechanisms design, mechanical optimization, teleoperated and co-manipulated robotscontrol, haptic devices design and control.

Its main research activity concerns the robotized tele-echography concept.He was part of the team that carried out a premiere,in 1998, on the application of very long-distance tele-echography.He has continued to work on this theme ever since, on both mechanical and automatic challenges.

 

Kosai RAOOF

Full professor at “Le Mans University”,

Head of Real Time Embedded Systems, ASTRE, Le Mans University, France. Chief Editor of ‘Wireless Sensor Networks’ Journal

Signal Processing and Instrumentation at LAUM laboratory, CNRS UMR6613 E-mail: kosai.raoof@univ-lemans.fr

 

Title : 

Multi-user and Multi-resolution Indoor localization

Abstract

In this lecture, we propose a new acoustic indoor positioning system that aims to localize simultaneously N sources with different precision based on acoustic time reversal. This method allows the acoustic signal to focus efficiently on the initial source regardless of its position and regardless of the heterogeneity of the propagation medium. Moreover, the width of the focal zone depends on the signal wavelength.

The idea is to assign to each source a unique code that is used to scramble two carrier frequencies in the audible band. The resulted acoustical signals are emitted by sources (loud speakers), recorded by four transducers (microphones), time-reversed and then re-emitted simultaneously into the medium. A receiver that searches to locate one of the existing sources demodulates he received signals in function of the desired precision and then descrambles the resulted signal with the corresponding source code.

Simulations have shown that the proposed method provides location estimates of better than 2 cm accuracy with 95 % precision in the case of four sources. An accuracy of 1 cm with 97% precision can be achieved when the system aims to localize only one source.

Keywords: smart sensor networks, agents for intelligent spaces, Indoor Localization, acoustic time reversal, accuracy, crambled-descrambled.

Biography

Kosai Raoof is the head of Real Time Embedded Systems, ASTRE Department of the ENSIM college of Engineering, Le Mans University, Le Mans France. He obtained his MSc and Ph.D from Grenoble University in 1990 and 1993 respectively; in 1998 he obtained the Habilitation à Diriger des Recherches Degree (HDR). He was invited to join Laboratoire des Images et Signaux (LIS) in 1999, to participate in the founding of telecommunication research group. His research interest was first focalized on advanced MIMO systems and joint CDMA synchronization; he studied and introduced polarized diversity MIMO systems in the research group. In 2007 he joined GIPSA-LAB to continue his research on MIMO antenna selection systems. Since 2011, he is full professor at the University of Maine, and establish a new group of research on signal processing and instrumentation at LAUM laboratory. He supervised more than 20 Ph.D. and MSc students in different fields of applied signal processing and telecommunications. He is editor of two books and chief editor of ‘Wireless Sensor Networks’ Journal first published in 2008. His recent interest is localization for smart sensor networks and distributed agents for intelligent spaces.

Biography :Prof. Kosai RAOOF

  1. Academic career

1994-2003Associate professor with ISTG, LIME and Laboratoire LIS

2004-2010 Associate professor at the UFR de Physique, GIPSA-LAB, UMR CNRS 5216

2011 University Full Professor, ENSIM, Le Mans University, Researcher at LAUM, UMR CNRS 6613

  1. Titles and diplomas

1981-1985 : Electronics engineer – Baghdad engineering college 1990 : DEA Signal Image Speech – CEPHAG, INPG

1993 : Doctoral thesis in Experimental Physical Methods, LIME-UJF Laboratory 1998 : Accreditation to Direct Research (HDR), in real-time signal processing, UJF Qualified in sections CNU Prof 61 (2003,2007) and Prof 63 (2007, 2010)

Holder of the PEDR Doctoral Supervision and Research Award in 2008 and PES since 2012

  1. Teaching

Course and assembly of practical work in Signal and Image Processing at ISTG and Polytech’Grenoble, as well as courses and practical work in Digital Telecommunication, Industrial Computing at the UFR de Physique IUP-Master- Licence EEA, ENSERG School of Engineering -INPG Grenoble, ENSIM Engineering School – Le Mans University.

  1. Pedagogical responsibilities

2003 – 2007: Responsible for IUP3 + MEEA Master, UJF Grenoble 2006 – 2009: Head of GEII License (ETCOM & SEEC), UJF Grenoble 2009 – 2011: Head of the ISTRE Master, UFR Physique Grenoble

since 2012: Head of ASTRE department – ENSIM Engineering School, Le Mans

  1. Research

Real-time signal processing in the biomedical field (LIME) and digital telecommunications (GIPSA-LAB). In 30 years of activity, more than 100 scientific articles and reports including:

  • Book edition (Advanced MIMO Systems), Scientific Research Publishing,
  • Edition of the book (Cognitive Radio Theory & Applications), Scientific Research Publishing,
  • 9 chapters in books,
  • 40 articles from international journals with reading committee,
  • 80 international congresses with proceedings and reading committee,
  • 10 guest lectures,
    • Supervision:22 Doctoral Theses.
    • Juries and thesis reviewer: 62 doctoral theses, 4 HDR Habilitations
    • Editor-in-Chief:Wireless Sensor Network (WSN) Journal, Scientific Research Publishing,