Special Sessions
2025 ICONIP Special Sessions
We are pleased to advise that the following Special Sessions are confirmed for the 2025 programme and we would like to thank the organisers for the time they took to put forward and manage these sessions.
| No. | Special Session Title | Runtime | Special Session Organiser |
| 1 | Sustainable Smart Villages – Harnessing Edge AI and TinyML for Low-Resource Communities | 1h40m | Prof. Absalom El-Shamir Ezugwu |
| Dr. Diego Oliva | |||
| Dr. Seyed Jalaleddin Mousavirad | |||
| Dr. Mario A. Navarro | |||
| 2 | Generative AI and Deep Learning for Advancing Affective Human-Robot Interaction | 1h40m | Prof. El-Sayed M. El-Alfy |
| Prof. Zeng-Guang Hou | |||
| Dr. Sadam Al-Azani | |||
| Dr. Amir Hussain | |||
| Dr. Hussain Bin Samma | |||
| Dr. Shadi Abudalfa | |||
| 3 | Reliable, Robust and Secure Machine Learning Algorithms | 1h40m | Prof. Monowar Bhuyan |
| Prof. Xuan-Son Vu | |||
| Prof. Harry Nguyen | |||
| 4 | Toward a Society Implementing Brain-Inspired AGI | 2h | Prof. Hiroshi Yamakawa |
| Prof. Akira Taniguchi | |||
| Prof. Takeshi Nakashima | |||
| 5 | Computer Vision and Language Understanding: Transformers in Image Analysis and Text Understanding in Medcine | 2h | Dr Olaide Nathaniel Oyelade |
| Prof Absalom El Shamir Ezugwu | |||
| Prof Hui Wang | |||
| 6 | Rethinking neighborhood through the principles of reverse nearest neighborhood and natural nearest neighborhood | 2h | Dr. Payel Sadhukhan |
| Dr. Sarbani Palit | |||
| 7 | Imitation learning and latent models applicable to real-world robotcis | 2h | Prof. Shin Ishii |
| 8 | Deep Learning Applications for Smart Agriculture | 2h | Prof. Mian M. Awais |
| Prof. El-Sayed M. El-Alfy | |||
| Prof. Derong Liu | |||
| Prof. Jian Wang |
1. Sustainable Smart Villages – Harnessing Edge AI and TinyML for Low-Resource Communities
Led by Prof. Absalom El-Shamir Ezugwu (North-West University, South Africa), Dr. Diego Oliva (Universidad de Guadalajara, Mexico), Dr. Seyed Jalaleddin Mousavirad (Mid Sweden University, Sweden), Dr. Mario A. Navarro (Universidad de Guadalajara, Mexico)
Despite the rapid advancement of artificial intelligence (AI) and the Internet of Things (IoT) across various sectors, rural and low-resource communities continue to face technological disparities due to constrained computing infrastructure, unreliable connectivity, and limited access to skilled expertise. Tiny machine learning (TinyML) and edge artificial intelligence (Edge AI) present transformative opportunities to bridge this digital divide by enabling low-power, cost-effective intelligence directly at the edge. This special session will explore innovative applications, challenges, and future directions for deploying TinyML-based solutions in developing regions to build sustainable smart villages – rural communities that leverage modern technologies, sustainable infrastructure, and innovative governance models to improve residents’ quality of life, economic opportunities, and environmental sustainability. By integrating digital solutions such as smart agriculture, renewable energy, e-governance, and community-driven innovations, smart villages address local challenges while promoting long-term socio-economic growth. TinyML, with its ability to run machine learning models on ultra-low-power microcontrollers, offers promising avenues for enhancing agriculture, healthcare, renewable energy, environmental monitoring, and smart governance. From AI-driven pest detection and precision irrigation to low-cost wearable health monitoring and decentralized renewable energy management, TinyML unlocks new possibilities for scalable, locally adaptable, and resilient solutions. Moreover, by reducing dependency on cloud infrastructure, Edge AI ensures real-time decision-making, enhances security, and minimizes data transmission costs, making it particularly well-suited for remote and off-grid environments. This session invites researchers to share cutting-edge research, case studies, and deployment experiences in harnessing Edge AI and TinyML for socio-economic development, addressing critical technological and societal challenges in developing regions.
The main topics that are of interest to this special session include, but are not limited to, the following:
• Tiny machine learning for precision agriculture and food security
• TinyML for disaster prediction and early warning systems
• Wearable and mobile healthcare innovations for rural areas
• Energy-efficient edge AI solutions for off-grid communities
• Tiny machine learning-driven smart infrastructure for efficient water supply and waste management
• Smart governance and e-governance solutions
• Education and capacity building with AI
• Challenges in scaling tiny machine learning in developing regions
• AI for indigenous knowledge preservation
• Community-driven AI and participatory design
• AI-enabled supply chain and logistics
• Smart villages and the digital preservation of indigenous knowledge
• Real-world deployments and case studies
2. Generative AI and Deep Learning for Advancing Affective Human-Robot Interaction
Led by Prof. El-Sayed M. El-Alfy (King Fahd University of Petroleum and Minerals, Saudi Arabia), Prof. Zeng-Guang Hou (Institute of Automation, Chinese Academy of Sciences), Dr. Sadam Al-Azani (King Fahd University of Petroleum and Minerals, Saudi Arabia), Dr. Amir Hussain (Edinburgh Napier University, UK), Dr. Hussain Bin Samma (King Fahd University of Petroleum and Minerals, Saudi Arabia), Dr. Shadi Abudalfa (King Fahd University of Petroleum and Minerals, Saudi Arabia)
Recent advancements in deep learning and generative AI have significantly enhanced affective computing, enabling robots not only to recognize and interpret human emotions, but also to generate and express emotional cues with greater accuracy and cognitive granularity. This special session serves as forum for interdisciplinary dialogue among expertise from artificial intelligence, deep learning, robotics, affective computing and human-machine interaction. It explores state-of-the-art deep learning and generative AI methodologies designed to improve human-robot interaction (HRI). Potential key topics include facial and speech emotion recognition, physiological signal analysis, multimodal emotion fusion, and real-time affective feedback systems. The session will feature both invited and regular papers presenting novel approaches and real-world applications across various domains, including healthcare, education, and social robotics. Researchers will showcase how deep learning models can improve cognitive emotional intelligence in robotic systems, leading to more natural and intuitive interactions. In addition to technical presentations, the session will facilitate expert discussions on challenges and future directions in affective computing for robotics. By bridging the gap between theoretical advancements and practical implementations, this session aims to foster interdisciplinary collaboration between academia, industry, and AI practitioners, paving the way for emotionally intelligent robots in real-world settings.
This special session include, but are not limited to the following:
• Multimodal emotion fusion to integrate visual, auditory, and physiological signals for emotion recognition
• Generative AI for Emotionally Expressive Robotic Responses
• Multimodal transformer networks for emotion-aware educational and social robotics
• Emotional AI for assistive robotics in healthcare
3. Reliable, Robust and Secure Machine Learning Algorithms
Led by Prof. Monowar Bhuyan (the Department of Computing Science, Umeå University, Sweden), Prof. Xuan-Son Vu (Lund University), Prof. Harry Nguyen (the School of Computer Science and Information Technology, University College Cork – National University of Ireland, Cork, Ireland)
The wider adoption of machine learning (ML) and artificial intelligence (AI) make several applications successful across societies, such as healthcare, finance, robotics, transportation and industry operations by inducing intelligence in real-time. Designing, developing and deploying reliable, robust, and secure ML algorithms are desirable for building trustworthy systems that offer trusted services to users with high-stakes decision-making. Moreover, building trustworthy AI systems requires lots of research efforts in addressing different mechanisms and approaches that could enhance user and public trust. This special session aims to draw together state-of-the-art machine learning (ML) advances to address challenges for ensuring reliability, security and privacy in trustworthy systems. The challenges in different learning paradigms include but are not limited to (i) robust learning, (ii) adversarial learning, (iii) stochastic, deterministic and non-deterministic learning, and (iv) secure and private learning. Nonetheless, all aspects of learning algorithms that can deal with reliable, robust and secure issues are the focus of the special session. It will focus on the robustness, performance guarantee, consistency, transparency and safety of AI, which is vital to ensure reliability. Original contributions and comparative studies among different methods are welcome, along with an unbiased literature review.
Topics of the special session include (reliable/robust/secure learning methods), including but not limited to:
• Robustness of machine learning/deep learning/reinforcement learning algorithms and trustworthy systems in general.
• Confidentiality, consistency, and uncertainty in model predictions for reliability beyond robustness.
• Transparent AI concepts in data collection, model development, deployment and explainability.
• Adversarial attacks – backdoor, bit-flip, evasion, poisoning, extraction, inference, and hybrid.
• New solutions to make a system robust, secure, and private to novel or potentially adversarial inputs; to handle model misspecification, corrupted training data, addressing concept drifts, data shifts, missing/manipulated data instances, backdoored triggers.
• Theoretical and empirical analysis of reliable/robust/secure ML methods.
• Comparative studies with competing methods without reliable/robust certified properties.
• Applications of reliable/robust machine learning algorithms in domains such as healthcare, biomedical, finance, computer vision, natural language processing, big data, and all other relevant areas.
• Unique societal and legal challenges are facing reliability for trustworthy AI systems.
• Secure learning from data having high missing values, incompleteness, and noise
• Assessing security and privacy of Large Language Models (LLMs)
• Private learning from sensitive and protected data
* Website: https://sites.google.com/view/reliablemldl2025/home
4. Toward Safe Brain-Inspired AGI
Led by Dr. Hiroshi Yamakawa (The University of Tokyo / The Whole Brain Architecture Initiative), Dr. Akira Taniguchi (Ritsumeikan University), Mr. Takeshi Nakashima (Ritsumeikan University)
Modern deep learning systems have achieved remarkable successes but face critical challenges in interpretability, trustworthiness, and alignment with human values. Research at the interface of computational neuroscience and machine learning suggests that incorporating key principles from the human brain—such as sparse, modular network architectures—can enable the development of more interpretable and controllable AI models. At the same time, ensuring these systems are genuinely beneficial requires robust frameworks for AI alignment. These frameworks draw on ethics, policy, and advanced neurotechnology to detect latent internal states and guide system behavior toward socially acceptable outcomes.
This special session will highlight recent progress in brain-inspired mechanistic interpretability, ethical alignment strategies, and neurotechnology-driven cognitive understanding of AI. We will explore how biologically motivated network structures enable clearer insight into internal representations, how alignment mechanisms can be integrated without sacrificing performance, and how leveraging cognitive neuroscience tools can expand our understanding of human and artificial minds.
Through invited talks (spanning academic and industrial perspectives), contributed presentations, and an interdisciplinary panel discussion, attendees will gain a holistic overview of:
• Brain-Inspired AI: Sparse and modular architectures, biologically plausible learning rules, and reverse-engineering neural computations.
• AI Alignment: Balancing advanced capabilities with safety, ethical standards, and user trust.
• Neurotechnology Applications: Using neuroimaging and cognitive modeling to interpret AI systems and refine human–machine interaction.
* Website:https://wba-initiative.org/en/25532/
5. Computer Vision and Language Understanding: Transformers in Image Analysis and Text Understanding in Medcine
Led by Dr Olaide Nathaniel Oyelade (Department of Engineering, Computing and Mathematics, University of Chichester, West Sussex, Engand United Kingdom), Prof Absalom El Shamir Ezugwu (Unit for Data Science and Computing, North-West University, Potchefstroom 2520, South Africa), Prof Hui Wang (School Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast)
Recent advances in transformer architectures have inspired new ways to apply artificial intelligence (AI) to medicine, healthcare, pharmacology, and biomedical sciences. Core to this advancement and application is the design of novel techniques to address the limitation of traditional machine learning and deep learning architectures in computer vision and language processing. On the other hand, demonstrating the outstanding application of this progress has a tendency to promote the widespread use of vision transformers in the medical field in the context of model interpretability, model complexity, and the need for extensively annotated datasets. They promise to advance capturing complex chemical features in drug design, and also evolvement of drug design landscapes. The models’ capability to capture long-range dependencies of complex relationships in medical imaging, protein sequences, and gene expression data, points to the prospect of further research in this area. Novel techniques are now required to tackle real-world challenges in medical image and language processing. In this ICONIP2025 Special Session, we encourage researchers to contribute their work, focusing on the design of algorithms, architectural innovative techniques, and application of the transformer and associated deep learning architectures. We also welcome studies demonstrating the combination of these models with other computational methods.
The following are the topics authors may consider when submitting their manuscripts for the special session, though not limited to these:
• Medical image analysis in segmentation or classification
• Multimodal medical image integration and fusion
• Image feature to automated medical report generation
• Predictive modeling
• Clinical natural language processing: analyzing clinical text from patient records
• Biomolecular sequences: analyzing protein sequences, gene expression data, and
• Identifying potential drug targets.
• Identifying genetic variations associated with diseases
• Multimodal language processing
• Knowledge graph in large language model for drug design
• Application of AI to ageing control
* For questions and further information, please contact Dr Olaide Nathaniel Oyelade/o.oyelade@chi.ac.uk
6. Rethinking neighborhood through the principles of reverse nearest neighborhood and natural nearest neighborhood
Led by Dr. Payel Sadhukhan (Techno Main, Salt Lake, Kolkata, India), Dr. Sarbani Palit (Indian Statistical Institute, Kolkata, India)
The principles of reverse nearest neighborhood (RkNN) and natural neighborhood (NaN) provide intuitive and pliable ways to identify adjacency relationships in feature spaces, mimicking human-like reasoning about proximity and connectivity. RkNN defines asymmetric neighborhoods, capturing the idea that a point may be in the neighborhood of another without the reverse being true, which allows for adaptive neighborhood sizes even at a fixed k. This asymmetry is particularly useful for detecting outliers and handling disconnected points, as it permits zero-neighborhood assignments—making it ideal for open-set classification and anomaly detection tasks. On the other hand, the modus operandi of principle of Natural Neighborhood eliminates the need for predefined parameters like k by dynamically determining neighborhood structures based on local density and connectivity, making it highly adaptable to datasets with unknown or varying configurations.
Despite their robustness and interpretability, these methods remain underutilized in contemporary machine learning research. Their ability to model complex, non-uniform data structures can enhance the outcomes of clustering algorithms, improve graph-based learning, and refine manifold learning techniques. For instance, integrating RkNN and NaN into graph neural networks (GNNs) could lead to more accurate message-passing mechanisms by considering asymmetric and density-aware relationships. Additionally, these principles could strengthen semi-supervised learning by improving pseudo-labeling strategies through better neighborhood estimation.
We call upon the ML community to further explore and apply these techniques in real-world scenarios, such as biomedical data analysis, fraud detection, and recommendation systems, where nuanced neighborhood definitions are critical. Research papers investigating theoretical extensions, scalable implementations, or novel applications of RkNN and NaN are highly encouraged. By leveraging these principles, we can develop more adaptive, interpretable, and efficient learning systems that better reflect the inherent structure of complex datasets.
The main topics that are of interest to this special session include, but are not limited to, the following:
• Applications of the paradigms in handling real-world data and different data types like text, image, etc.
• Application of the paradigms in extant methodologies and schemes.
• Investigation of the properties of these paradigms.
• Efficient ways of computing reverse nearest neighborhood and nearest neighborhood.
• Approximate ways of computation of the same.
• Related applications.
7. Imitation learning and latent models applicable to real-world robotcis
Led by Prof. Shin Ishii (Graduate School of Informatics, Kyoto University)
Imitation learning is a machine learning framework that enables learning agents to behave similarly to demonstrator agents. Recently, modern AI-based methods that include latent models have demonstrated great success in real-world applications. Among those, foundation model (FM)-based approaches equipped with latent models have also gained attention, offering the potential for broad applicability across a wide variety of robotic systems, architectures, and environments in a unified manner. However, simply applying FM-based methodology that has achieved great success in large language models to imitation learning by real-world robots is not trivial, given the relative scarcity of available data in the robotics domain. In this session, we will invite a leading researcher who has successfully applied imitation learning methodologies to real-world robotics and welcome several full-paper submissions as well as invited-paper submissions. This session aims to provide valuable insights into the future of imitation learning technologies, including FM-based approaches, in real-world robotics, and to explore the future direction of data-driven robotics.
Topic covered in the session:
• Imitation learning methods applicable to robotics
• Foundation model-based approrches to robotics
• Scalable machine learning methods for real-world robotics
• Real world robotics motivated by human/aminal motor controls
8. Deep Learning Applications for Smart Agriculture
Led by Dr. Mian M Awais (Lahore University of Management Science (LUMS), Pakistan), Prof. El-Sayed M. El-Alfy (King Fahd University of Petroleum and Minerals, Saudi Arabia), Prof. Derong Liu (Southern University of Science and Technology, Shenzhen, China, and University of Illinois, Chicago, USA), Prof. Jian Wang (China University of Petroleum, East China), and Prof. Paul Pang (Federation University, Australia)
Smart agriculture is rapidly transforming with advancements in deep learning, enabling precision farming, automated monitoring, and intelligent decision-making. This session explores cutting-edge research and applications of deep learning in agriculture, focusing on crop yield prediction, pest detection, smart irrigation, and future of smart agriculture. The session will feature invited and regular papers showcasing innovative methodologies, including convolutional neural networks (CNNs) and transformers, image analysis, and AI-driven automation. Through expert presentations, interactive discussions, and a panel forum, attendees will gain insights into the challenges and future trends of AI-driven agriculture. This session aims to foster collaboration among researchers, industry professionals, and policymakers, contributing to the evolution of sustainable and efficient farming practices.
The special session includes topics related to:
• Future trends and challenges in smart agriculture
• Deep learning for precision agriculture
• Deep Learning Models for Crop Yield Prediction
• Automated Pest Detection using Convolutional Neural Networks AI-driven Smart Irrigation Systems
• Satellite Imagery and Deep Learning for Soil Health Monitoring