Research Areas

Artificial Intelligence and Machine Learning (AIML)

Research in the AIML Group aims to create a unique synergy of strength in artificial intelligence, machine learning, big data analytics, data mining, intelligent user interfaces, autonomy-oriented computing, and Web intelligence. The group is positioned to develop advanced algorithms and intelligent systems to meet the real-world needs and challenges, e.g., in healthcare and social computing.

Faculty Involved:

• Prof. CHEN, Li
• Prof. CHEN, Yifan
• Prof. CHEUNG, William Kwok Wai
• Prof. CHEUNG, Yiu Ming
• Prof. HAN, Bo
• Prof. HUANG, Longkai
• Prof. LIU, Jiming
• Prof. LIU, Yang
• Dr. YIN, Kejing
• Prof. ZHANG, Eric Lu

Funded Research and Consultancy Projects in the Past Few Years:

Interpretable Representation Learning from Intensive Care Big Data by Incorporating Complex Decision-Making Process
	Staff	Dr. YIN, Kejing
	Objectives	Extracting interpretable computational phenotypes from multimodal critical care data to serve as a unified representation of patient disease states. Building dynamic representation learning models that characterize the interaction between clinical decisions and disease states based on interpretable phenotypes. Incorporating prior medical knowledge to improve the robustness of representation learning in data-sparse regions.
	Grant	National Natural Science Foundation of China (NSFC) Research Fund

Research on Key Technology for Modeling Cross-Domain Sequential Behaviors in Recommender Systems
	Staff	Prof. CHEN, Li
	Objectives	To conduct in-depth analyses and systematic exploration of the cross-domain sequential behaviors (CDSB) problem in recommender systems and the corresponding Transfer Learning (TL) techniques. To develop an unbiased TL framework for modeling CDSB in response to prominent challenges such as the cross-domain fairness issues, the cross-domain biased dependency in user behaviors, and the transfer of sensitive information in cross-domain knowledge. More To develop multiple modules based on graph neural networks, attention mechanisms, and TL with non-behavioral information, and to evaluate the effectiveness of our proposed unbiased transfer learning methods in terms of both recommendation accuracy and fairness.
	Grant	NSFC/RGC Joint Research Scheme

Statistical Modeling of Empirical Distribution Transforms via Wasserstein Metrics
	Staff	Prof. CHEN, Yifan
	Objectives	Propose a new distributional regression model in which both predictors and responses are multivariate empirical distributions; Analyze the population risk for empirical distributions taking non-IID transforms, such as free-support constructions or compressions; Apply the distributional perspective (taking weight matrices as empirical distributions) and techniques to large language model compression for efficient inference; More Assess the proposed results on proper synthetic and real-world benchmark datasets, demonstrating the effectiveness of our proposal.
	Grant	Early Career Scheme (ECS)

Towards Trustworthy Foundation Models under Imperfect Scenarios
	Staff	Prof. HAN, Bo
	Objectives	Robust FMs in reasoning with noisy inputs. Safe FMs in generation with adversarial and OOD prompts. Fair FMs against imbalanced preferences. Reliable FMs in problem-solving with professional knowledge.
	Grant	Young Collaborative Research Grant (YCRG)

Bridging Brain and Machine: Real-Time Emotional Intelligence via Multimodal Large Language Models
	Staff	Prof. LIU, Jiming
	Objectives	Physiology-first perception: Deploy specialized neural modules that extract rich semantic evidence directly from raw brain signals (EEG/fNIRS), facial micro-expressions, vocal prosody, and body language in real time. Human-like cognitive core: A central Multimodal Large Language Model (MLLM) orchestrates these cues through a transparent, step-by-step reasoning pipeline that mirrors human integration of bodily signals and external stimuli. Explainable & evolving insights: Our system not only predicts emotions with state-of-the-art accuracy but also generates natural-language explanations of how and why the emotional state is changing over time, building trust and opening new possibilities in mental health, human-AI interaction, education, and beyond.

Learning with Complex Data: A Mutual-Information-Guided Deep Learning (MIGDL) Framework
	Staff	Prof. LIU, Jiming
	Objectives	To develop a general and adaptive Mutual-Information-Guided Deep Learning (MIGDL) framework, which makes use of mutual information to provide theoretical analysis and quantitative characterization of the learning behaviors and capacity of a deep learning model. To develop a Capacity characterization module (C-Module) to analytically characterize and quantitatively describe the capacity of a deep learning model with respect to a given learning target; to develop a Feature adaptation module (F-Module) to determine what kind(s) of features must be further acquired to effectively complement the extracted/learned information; and to develop a Model adaptation module (M-Module) to determine which part(s) of an existing model must be further adapted to make optimal use of the data newly acquired by the F-Module. More To analyze the algorithmic properties of the developed framework and three of its modules, in terms of: the learning capacity of the framework; the correctness and optimality of the learning behavior of each module; and the superiority of the proposed framework to existing deep learning models, with respect to information extraction and module adaptation for a given learning task. To validate the learning performance of the developed framework and modules by conducting systematic experiments on both synthetic datasets and four real-world learning tasks: spatio-temporal prediction, video action recognition, complex systems behavioral prediction, and multi-modal data analytics.
	Grant	General Research Fund (GRF)

Epidemiology-Guided AI for Pandemic Preparedness: Early Warning of Emerging Infectious Diseases from Heterogeneous & Incomplete Data Building the next generation of intelligent early-warning systems that can detect threats before they explode
	Staff	Prof. LIU, Yang
	Objectives	Dynamic Multi-Layer Networks: Integrate “human–animal/vector–environment” interactions using epidemiology-informed spatiotemporal modeling. Heterogeneous Tensor Forecasting: Design novel tensor-based architectures with epidemiological regularization and knowledge transfer to achieve robust predictions under data scarcity. Early Warning Signal Discovery: Leverage complex systems theory to identify mathematical precursors of outbreaks for proactive intervention.More Real-World Impact: Rigorous validation with public health experts and deployment-oriented testing on real epidemic data.

Unlocking Chaos in the Real World: Invariance-Regularized Mutual Learning (IRML) for Partially Observed Dynamical Systems Revolutionizing how we understand and predict chaotic phenomena with limited data and knowledge
	Staff	Prof. LIU, Yang
	Objectives	Discover and exploit hidden invariance to jointly learn chaotic behaviors, refine/impute missing spatiotemporal data, and uncover unknown physical mechanisms. Deliver strong theoretical guarantees on topological consistency, attractor geometry, and out-of- domain generalization. Bridge chaotic systems theory with modern machine learning for robust, interpretable predictions.
	Grant	National Natural Science Foundation of China (NSFC) Research Fund

Improved Metabolic Network Reconstruction and Metabolite Profiles Prediction using Complete and Strain-Resolved Microbial Genomes
	Staff	Prof. ZHANG, Eric Lu
	Objectives	Generate complete and strain-resolved metagenome-assembled genomes using long-read metagenomic sequencing and available reference genomes. Taxonomic annotation and gene function prediction using genome specific deep learning language models. Predict metabolomic profiles using well-characterized microbiomes and reconstruct high-quality species and multispecies metabolic networks.More Application to identify novel biomarkers associated with diarrhea predominant irritable bowel syndrome (IBS-D).
	Grant	Collaborative Research Fund (CRF)