Josef KittlerCentre for Vision, Speech and Signal Processing, University of Surrey School of Artificial Intelligence and Computer Science, Jiangnan University, China |
Bio: Josef Kittler is Professor of Machine Intelligence at the Centre for Vision, Speech and Signal Processing, University of Surrey. He is also Distinguished Professor at the School of Artificial Intelligence and Computer Science, Jiangnan University. He received his BA, PhD and DSc degrees from the University of Cambridge. He teaches and conducts research in Machine Intelligence, with a focus on Biometrics, Video and Image Database retrieval, Deep Learning and Cognitive Vision. He published a Prentice Hall textbook on Pattern Recognition: A Statistical Approach and several edited volumes, as well as more than 1000 scientific papers, including more than 300 journal publications. He serves on the Editorial Boards of Pattern Recognition, Pattern Recognition Letters and Springer Nature Computer Science. He was Series Editor of Springer Lecture Notes on Computer Science 2004-2016. Speech Title: Intriguing properties of foundation model representationsAbstract: The ability of foundation models to produce effective representations for immensely complex problems is absolutely staggering. The information content of these representations is beyond human comprehension and is commonly perceived as magic. It has the capacity to be probed by queries and prompts, revealing plethora of detail about the input visual stimulus. However, the black box nature of their operation hinders explainability of decisions, and this in turn may impede the adoption of impressive solutions in practice. It is pertinent to ask, what the underlying essence of deep representation is. This is not only motivated by the natural curiosity of what’s the trick behind the magic? Over and above such human interest, it is a serious scientific question. A deeper knowledge of the nature of deep representations could direct the search for more efficient ways of unlocking and exploiting the rich information content it captures, as well as helping explainability. The talk will explore deep representations generated by AI models and discuss the implications of the findings. |
Hamido FujitaMalaysia-Japan International Institute of Technology (MJIIT), Malaysia (Online Talk) |
Bio: Hamido Fujita Life Senior Member IEEE, is Executive Chairman of i-SOMET Incorporated Association, Japan, Professor at Malaysia-Japan International Institute of Technology of Universiti Teknologi Malaysia, and Distinguished Professor at Iwate Prefectural University, Japan, he is also Research Professor at University of Granada, Spain. He is Highly Cited Researcher in Cross-Field for the year 2019 and in Computer Science for the year 2020, 2021, 2022, and 2023 respectively, by Clarivate Analytics. He received Doctor Honoris Causa from Óbuda University, Budapest, Hungary, in 2013 and received Doctor Honoris Causa from Timisoara Technical University, Timisoara, Romania, in 2018, and a title of Honorary Professor from Óbuda University, in 2011. He is Distinguished Research Professor at the University of Granada, and Adjunct Professor with Taipei Technical University, Taiwan, Harbin Engineering University, China and others. He supervised Ph.D. students jointly with the University of Laval, Quebec City, QC, Canada; University of Technology Sydney; Oregon State University, Corvallis, OR, USA; University of Paris 1 Pantheon-Sorbonne, Paris, France; and University of Genoa, Italy. Dr. Fujita is the recipient of the Honorary Scholar Award from the University of Technology Sydney, in 2012. He was the Editor-in-Chief for Knowledge-Based Systems (Elsevier) (2005-2019) and then Emeritus Editor of Knowledge-Based Systems in 2020~. Since 2020 he is currently the Editor-in-Chief of Applied Intelligence (Springer) and the Editor-in-Chief of International Journal of Healthcare Management (Taylor & Francis). He headed a number of projects including intelligent HCI, a project related to mental cloning for healthcare systems as an intelligent user interface between human-users and computers, and SCOPE project on virtual doctor systems for medical applications. He collaborated with several research projects in Europe, and recently he is collaborating in OLIMPIA project supported by Tuscany region on Therapeutic monitoring of Parkinson disease. He has published more than 480 articles mainly in high impact factor journals. Speech Title: Views on Artificial Intelligence and Machine Learning perspectives: Trustworthiness and Explainability Abstract: The hot topics in training in Machine Learning is a crucial aspect that affects the credibility of the system in terms of performance and is employed for robust applications such as in healthcare systems. Machines or algorithms, in wide challengeable applications in security or vision or health care early predictions, learn from data. Nevertheless, in most cases, the extensive and unbalanced data and noise make it unreliable (needs resilient and robustness) in prediction accuracy. Supervised machine learning is and was one of the aspects of providing artificial intelligence-based solutions. However, this is and was limited due to the difficulty of labelling big data and many crucial problems in weak relations and noise in data. Semi-supervised learning, for example, Multiview learning, could assist in solving these problems. In much published research, there are still problems in providing machine learning models that are unbiased and efficient in terms of robustness and resilience in data-driven systems. Multi-class classification still has problems in terms of clear definition in class classification, bias, imbalance and weak relations, making machine learning for multi-class classification insecure for classification or regression analytics. Also I will highlight the explainable AI and it is role to provide trustworthiness in AI solutions. |
Yudong ZhangSoutheast University, China |
Bio: Dr. Yudong Zhang is a Chief Professor and Doctoral Supervisor at Southeast University, and a recipient of a national-level high-level talent award. He was a postdoctoral researcher from Columbia University. He has held full-time positions at universities in seven countries. He has co-authored over 100 papers published in CAS Q1 journals. His Google citation is over 52k with an h-index of 123. He has served as (co-)principal investigator for numerous domestic and international academic projects, including national key R&D projects, National Institutes of Health (NIH) projects, and BBSRC projects. Speech Title: AI-Enabled Colorectal Cancer Research: Progress and Emerging Innovations Abstract: Colorectal cancer (CRC) remains a formidable challenge within the global healthcare landscape, necessitating urgent and systemic improvements in its management paradigms. This report outlines the latest advancements in CRC research and clinical practice, focusing on the transformative role of integrated innovations across diagnosis, treatment, and translational medicine—specifically highlighting the pivotal supportive value of AI and image processing technologies. The report underscores the critical strategic significance of technological breakthroughs—including advanced diagnostic methods, computer-aided solutions, AI and image processing technologies, and novel therapeutic strategies—in advancing the diagnosis and treatment of CRC. In particular, AI and image processing technologies facilitate the precise identification of early-stage lesions, the accurate differentiation between benign and malignant conditions, and the dynamic monitoring of disease progression through the intelligent analysis of colorectal endoscopic and pathological images; this significantly enhances diagnostic efficiency and accuracy, thereby providing highly effective support for clinical decision-making. Finally, the report delineates key directions for future research—including the clinical validation, safety assessment, and preclinical translational investigation of AI and image processing technologies—while emphasizing the synergistic interplay between basic research and clinical application, thereby offering a strategic outlook for alleviating the global burden of colorectal cancer. |
Christian SacareaBabes-Bolyai University, Romania |
Bio: Christian Săcărea, PhD, is currently holds the position of Vice-Rector. His extensive research interests and contributions lie primarily in the fields of Formal Concept Analysis, Knowledge Representation, Artificial Intelligence, and Data Security. As a scientist, Dr. Săcărea is dedicated to extracting meaningful patterns and structures from complex data, with significant work in knowledge discovery and visualization in domains such as healthcare. He coordinates the Formal Concept Analysis research group at his university, exploring mathematical frameworks for data organization and retrieval. His proficiency in Graph Algorithms and Semantic Web technologies also provides a robust foundation for addressing challenges in 3-D and Surface Reconstruction as well as Computer Vision and Machine Intelligence. Speech Title: Concept Lattices - The Missing Semantics of Visual Models. A conceptual framework for interlinking Formal Concept Analysis with Graphics and Image Processing Abstract: Graphics and image processing (GIP) and Formal Concept Analysis (FCA) both work with pictures, but in opposite directions. GIP turns images into descriptions (analysis) and descriptions into images (synthesis). FCA turns data tables into hierarchies of concepts and presents those hierarchies as diagrams for people to reason with. The two fields have more in common than an interest in visual artefacts: both rest on the theory of complete lattices, Galois connections and closure operators, the same theory that underlies mathematical morphology. Each field also lacks something the other has. GIP builds visual models whose conceptual structure, meaning which concepts a model distinguishes and how they are ordered, usually stays implicit. FCA has an explicit, verifiable theory of concepts, but laying out, rendering and exploring its diagrams are still open problems. This keynote develops two things: a theoretical framework assembled from four established bodies of work, and a researcher-developed conceptual framework, the Visual–Conceptual Bridge (VCB). For each ICGIP track we show how the framework guides data collection, analysis and interpretation, and we state the assumptions, limitations and competing perspectives that bound it. |
