The ICIC 2026 Program Committee is inviting proposals for special sessions to be held during the conference (http://www.ic-icc.cn/2026/index.php), taking place on July 22-26, 2026, in Toronto, Canada.

Each special session proposal should be well motivated and should consist of 8 to 12 papers. Each paper must have the title, authors with e-mails/web sites, and as detailed an abstract as possible. The special session organizer(s) contact information should also be included. All special session organizers must obtain firm commitments from their special session presenters and authors to submit papers in a timely fashion (if the special session is accepted) and, particularly, present them at the ICIC 2026. Each special session organizer will be session chair for their own special sessions at ICIC 2026 accordingly. All planned papers for special sessions will undergo the same review process as the ones in regular sessions. All accepted papers for special sessions will also be published by Springer's Lecture Notes in Computer Sciences (LNCS)/ Lecture Notes in Artificial Intelligence (LNAI)/ Lecture Notes in Bioinformatics (LNBI).

All the authors for each special session must follow the guidelines in CALL FOR PAPERS to prepare your submitted papers.

Proposals for special sessions should be submitted in ELECTRONIC FORMAT by http://www.ic-icc.cn/icg/index.php at Special Session.

orders	Title	Organizers	Nationality
Special Session 1	Neural Signals and Intelligent Computing: From Brain Data to Trustworthy Human-AI Collaboration	Ziyu Jia Roger Mark Idris Elbakri	China
Special Session 2	Advances in Graph Machine Learning	Zhipeng Li Ming Li Yun Ding Xuesong Jiang Bo Jiang Zhuhong You	China
Special Session 3	Advances in Multimodal Intelligence for Visual and Medical Data	Meng Xing Yong Su Mingliang Dou Yao Zhang Yude Bai Zehua Zhang	China
Special Session 4	Computational Intelligence Models for Smart Cities	Pengjiang Qian Wenbing Zhao Khin-Wee Lai	China
Special Session 5	Sustainable Intelligent Computing: Efficient Systems, Algorithms, and Applications	Lingjie Li Qiuzhen Lin Ling Wang Zhong Ming	China
Special Session 6	Biomolecular Language Models: Foundation Model, Representation, and Applications	Junkai Ji Jun Zhang Wei Zhou

1. Neural Signals and Intelligent Computing: From Brain Data to Trustworthy Human-AI Collaboration

Organizer:
Ziyu Jia
Institute of Automation, Chinese Academy of Sciences
Email: ziyu.jia.editor@outlook.com

Roger Mark
Massachusetts Institute of Technology
Email: rogermark.mit@gmail.com

Idris Elbakri
Kyrgyz National University
Email: ldris@buu.edu.kg

Scope and Topics:
This session targets the full pipeline of neural signals and intelligent computing, spanning data acquisition and quality control; representation learning and time--frequency and spatiotemporal modeling; cross-subject, cross-session, and cross-device generalization; robustness, uncertainty estimation, and interpretability; and system deployment for online decoding and closed-loop control. Modalities include EEG, MEG, fNIRS, and ECoG, optionally combined with peripheral physiology and behavioral streams such as EDA, ECG, skin temperature, respiration, eye tracking, motion capture, and kinematics. We welcome theoretical and algorithmic advances as well as end-to-end systems, wearables and edge inference, neuromorphic or event-driven sensing and computing, real-time human--computer interaction, and safety/compliance practices. Evaluation should emphasize cross-dataset and cross-protocol validation, robustness under distribution shift, personalization and few-shot adaptation, external validation, and audits for fairness and privacy. Application domains include cognitive and affective state estimation, attention and memory modeling, motor imagery and assistive communication, neurorehabilitation and neuromodulation, driving and industrial safety, surgical guidance, and education and immersive interaction. Strong reproducibility is encouraged through open data and code, standardized benchmarks, clear task definitions with statistical reporting, and risk governance aligned with ethical and legal requirements.

2. Advances in Graph Machine Learning

Organizers:
Zhipeng Li
Ningbo Institute of Digital Twin
Email: lizhipengqilu@gmail.com

Ming Li
Zhejiang Normal University
Email: mingli@zjnu.edu.cn

Yun Ding
Anhui University
Email: yunding92@163.com

Xuesong Jiang
Qilu University of Technology(Shandong Academy of Sciences)
Email: jxs@qlu.edu.cn

Bo Jiang
Anhui University
Email: jiangbo@ahu.edu.cn

Zhuhong You
Northwestern Polytechnical University
Email: zhuhongyou@nwpu.edu.cn

Scope and Topics:
Special Session on “Advances in Graph Machine Learning” Graph machine learning has become one of the most active research frontiers in artificial intelligence, providing powerful tools for representing, learning, and reasoning over complex relational structures. Graph-based models have demonstrated strong capabilities in diverse domains such as social networks, biological systems, chemistry, recommendation, and multimodal information fusion. With the rapid progress of deep learning, graph neural networks (GNNs), graph transformers, and large language models (LLMs), graph machine learning has evolved beyond traditional graph processing—empowering intelligent agents, multimodal reasoning systems, and large-scale decision-making frameworks. This special session aims to bring together researchers and practitioners from academia and industry to discuss recent advances, emerging theories, scalable algorithms, and impactful applications in graph machine learning. The session encourages cross-disciplinary contributions that link graph learning, neural architectures, and intelligent systems, highlighting both fundamental insights and real-world progress. Topics of Interest Topics of interest include, but are not limited to:
1) Learning theory, expressivity, and generalization of GNNs and graph transformers
2) Spectral, spatial, and multiscale perspectives on graph learning
3) Novel architectures and learning paradigms for GNNs, hypergraph neural networks, and manifold learning
4) Scalable, efficient, and distributed algorithms for large and dynamic graphs
5) Self-supervised, contrastive, or foundation models for graph representation learning
6) Graph-based reasoning and planning for intelligent agents and multi-agent systems
7) Graph-structured environments for task decomposition, collaboration, and control
8) Integration of graph knowledge into LLMs and multimodal reasoning frameworks
9) Graph learning in molecular discovery, drug design, and bioinformatics Objective This special session provides a timely platform to exchange the latest findings, foster interdisciplinary collaboration, and identify future directions in graph machine learning. It welcomes both theoretical and application-oriented studies that advance the understanding of relational, geometric, and structured data learning.

3. Advances in Multimodal Intelligence for Visual and Medical Data

Organizer:
Meng Xing
Ningbo Institute of Digital Twin, Eastrn Institute of Technology
Email: mxing@idt.eitech.edu.cn

Yong Su
Tianjin Normal University
Email: suyong@tju.edu.cn

Mingliang Dou
Taiyuan University of Technology
Email: doumingliang@tyut.edu.cn

Yao Zhang
Tianjin University
Email: zzyy@tju.edu.cn

Yude Bai
Tiangong University
Email: baiyude@tiangong.edu.cn

Zehua Zhang
Scientific and Technological Innovation Center
Email: zehua_new@yeah.net

Scope and Topics:
Multimodal intelligent computing has emerged as a key direction in modern artificial intelligence, leveraging heterogeneous data from diverse sources. These sources include visual data (images, videos, depth maps, RGB–IR, etc.), textual information, audio, sensor measurements, and clinical or biomedical data. Integrating and reasoning over these heterogeneous modalities enables richer understanding, improved perception, and more robust decision-making. This special session seeks high-quality contributions on advances in multimodal intelligence, covering both traditional visual applications and medical/clinical data scenarios. By highlighting the synergy between method innovation and application, this session provides a platform for cross-domain insights, bridging foundational multimodal learning techniques with practical visual and medical use cases. Topics of Interest (include but not limited to): 1) Joint representation learning and cross-modal embedding
2) Generative models for multimodal data (diffusion models, VAEs, autoregressive models)
3) Object detection, segmentation, classification, captioning, and reasoning with multimodal cues
4) Multimodal datasets, benchmarks, and evaluation protocols
5) Medical imaging combined with clinical text, omics, pathology, or physiological signals
6) Disease diagnosis, prognosis modeling, treatment planning, and clinical decision support
7) Multimodal image reconstruction, enhancement, segmentation, and detection
8) Report generation, visual–textual grounding, and structured reasoning
9) Digital health, wearable sensors, remote monitoring, and real-world clinical validation
10) Robustness, interpretability, safety, and fairness of multimodal medical AI

4. Computational Intelligence Models for Smart Cities

Organizer:
Pengjiang Qian
Jiangnan University
Email: qianpjiang@jiangnan.edu.cn

Wenbing Zhao
Cleveland State University
Email: w.zhao1@csuohio.edu

Khin-Wee Lai
University of Malaya
Email: lai.khinwee@um.edu.my

Scope and Topics:
Smart city comprehensive adopts the new generation of Internet, big data, Internet of Things, artificial intelligence, cloud computing and other information technologies to realize the intelligence of urban construction, planning, management, and service, forming an innovative and sustainable intelligent city. It integrates a variety of new generation information technologies to complete the automatic perception, collection, integration, analysis and sharing of urban information resources, and realize intelligent medical care, emergency response, environmental protection, education, transportation, etc., thus bringing convenience, high-efficiency, intelligence to people's life and response to their personalized needs. The construction of smart city involves many aspects, from urban road traffic to urban spatial layout and management, which require high technical support, as well as a complete information-based decision-making mechanism to provide a reliable guarantee for urban development.
In recent years, advanced computational intelligence models such as deep learning, active learning, transfer learning and information fusion have brought opportunities for smart city. Computational intelligence models have been successfully applied in many areas of smart city construction, such as urban traffic flow prediction, health monitoring and early warning, mobile intelligent question answering system, intelligent environmental resource deployment, etc. Although the existing computational intelligence models based on single-view data have achieved certain results, their practical application performance still cannot meet the needs of smart city construction. Compared with single-view data, multi-view data can provide more abundant and comprehensive information for the computational intelligence models, thereby further improving the performance of the model. Therefore, it is necessary to study the deep multi-view learning-driven computational intelligence model to overcome the defects existing in the construction of smart cities.
In this special issue, we hope to build a platform for researchers and engineers to explore this field and contribute their experience and wisdom to the development of computational intelligence models for smart cities. Topics of intended papers contain, but not limited to,
1) Advanced computational intelligence models for smart city, such as deep learning, sparse learning, transfer learning, active learning, multi-task learning
2) Smart city information management platform involving artificial intelligence
3) Smart city information decision-making system based on multi-view data
4) Prediction models combined with multi-view features, such as intelligent traffic flow prediction, intelligent medical disease prediction, intelligent weather prediction, signal light warning
5) Visualized human-computer interaction platform for smart city
6) Smart city monitoring system driven by deep multi-view learning
7) Deployment and management for smart cities with unsupervised methods, such as self-training models, clustering algorithms, principal component analysis
8) Data automated management and analysis for smart city
9) Screening and fusion of multi-modal heterogeneous clinical data for smart medical care with supervision methods, such as random forest, decision tree, naive bayes
Design of smart city resource allocation system based on big data analysis

5. Sustainable Intelligent Computing: Efficient Systems, Algorithms, and Applications

Organizer:
Lingjie Li
School of Artificial Intelligence, Shenzhen Technology University
Email: lilingjie@sztu.edu.cn

Qiuzhen Lin
College of Computer Science and Software Engineering, Shenzhen University
Email: qiuzhlin@szu.edu.cn

Ling Wang
Department of Automation, Tsinghua University
Email: wangling@tsinghua.edu.cn

Zhong Ming
College of Computer Science and Software Engineering, Shenzhen University; School of Artificial Intelligence, Shenzhen Technology University
Email: mingz@szu.edu.cn

Scope and Topics:
1.Introduction The exponential growth of large-scale AI models and computing infrastructure, particularly with the rise of generative AI represented by Large Language Models (LLMs), has triggered unprecedented global concern regarding computational energy consumption, and environmental sustainability. Against the backdrop of surging computational demands and global climate goals, developing energy-efficient, environmentally friendly, and sustainable intelligent computing technologies has become an urgent challenge. This necessitates not only a rethinking of the efficiency and design of AI models themselves but also an exploration of how AI technology can empower the broader green transition of society. This Special Issue aims to systematically gather the latest breakthroughs in energy-efficient AI algorithms, green hardware, and sustainable AI practices. We focus on innovative methods to enhance the energy efficiency of complex systems like LLMs and actively promote interdisciplinary research that leverages AI methods (e.g., Evolutionary Algorithms, Reinforcement Learning, and Continual Learning) to address climate and environmental challenges, ultimately steering the intelligent computing industry towards an environmentally friendly and resource-conserving future. 2.Scope and Topics The scope of this special session covers, but is not limited to: •Next-Generation AI Algorithms and Theory for Sustainability: This topic focuses on the design of novel AI algorithmic paradigms and related fundamental theory. It emphasizes algorithmic innovations in areas such as Evolutionary Computation, Reinforcement Learning, Continual Learning, and Recommender Systems, alongside their theoretical foundations in scalability, data efficiency, lifelong learning, and cross-layer co-optimization. •Efficient AI Models & Algorithms: Optimization techniques for LLMs and other complex models, including model compression, quantization, efficient Neural Architecture Search (NAS), and lightweight network design. •AI-Enabled Green Computing Optimization: Leveraging Evolutionary Algorithms, Reinforcement Learning, Continual Learning, and the planning/reasoning capabilities of LLMs to optimize computing hardware design, data center resource scheduling, task allocation, and system-wide energy management. •Green Hardware & Computing Systems: Specialized hardware for low-power AI computing (especially LLM inference), e.g., neuromorphic chips, in-memory computing architectures, high-efficiency AI accelerators, and green data center cooling technologies. •Distributed Green Computing Paradigms: Edge AI, federated learning, collaborative computing to reduce energy overhead and transmission costs. •AI for Environmental Sustainability: AI applications in smart grids, environmental monitoring & forecasting, climate modeling, renewable energy optimization, biodiversity conservation, circular economy.

6. Biomolecular Language Models: Foundation Model, Representation, and Applications

Organizer:

Junkai Ji
School of Artificial Intelligence，Shenzhen University
Email: jijunkai@szu.edu.cn

Jun Zhang
School of Artificial Intelligence，Shenzhen University
Email: junzhang@szu.edu.cn

Wei Zhou
School of Artificial Intelligence，Shenzhen University
Email: jerryzhou@szu.edu.cn

Scope and Topics:
Biomolecular language models have become a core methodology for learning generalizable representations across biological macromolecules and chemical compounds. These models are pretrained on large corpora of protein, DNA, and RNA sequences, as well as peptides and small molecules. The goal is to capture statistical regularities shaped by biophysical constraints, evolutionary selection, and functional organization. Recent work has moved beyond sequence-only learning. Structure, context, and experimental readouts are increasingly integrated. This trend supports more mechanistic interpretation and stronger downstream utility. This session addresses three coupled themes. The first theme is foundation model development. It includes data construction, tokenization, objective design, scalability, and multimodal or structure-aware training. The second theme is molecular representation. It includes embedding geometry, transferability, uncertainty, robustness to distribution shift, and representation diagnostics. The third theme is applications. It includes annotation, variant effect prediction, structure and interaction modeling, property prediction, and biomolecular design for therapeutics and biotechnology. Emphasis is placed on clear problem definitions, reproducible evaluation, and principled links between learned representations and biological or chemical mechanisms. More details can be found via http://ic-icc.cn/2026/index.php. Topics include (but are not restricted to): lFoundation model architectures and scaling for biomolecular language models lData construction, curation, and contamination control lTokenization and input representations for sequences, strings, and graphs lPretraining objectives and scalable training strategies lStructure-aware modeling and 3D representation learning lMultimodal learning with structure, context, and experimental readouts lRepresentation analysis, interpretability, and diagnostics lTransfer learning and cross-modality generalization lBenchmarking protocols, split strategies, and evaluation under distribution shift Applications in annotation, variant effects, interaction modeling, property prediction, and design