DeepMathAI
Research Group · Naresuan University
Deep Mathematical Modeling for Machine Learning & AI
Interdisciplinary Research Group on Deep Mathematical Modeling for Machine Learning and Artificial Intelligence
Our Vision
Bridging Mathematics & AI
DeepMathAI is founded on the belief that truly advanced AI technology must be built on a solid mathematical foundation. In the era of Big Data and highly complex deep learning models, we focus on creating new theoretical knowledge alongside developing practically applicable models — transparent, explainable, and generalizable across diverse, large-scale datasets.
Our multi-disciplinary team integrates expertise from pure mathematics, applied mathematics, statistics, computer science, and engineering to develop AI models that are not just powerful, but provably correct, interpretable, and deployable in real-world settings including healthcare and energy systems.
Research Focus
Four Research Pillars
Our research integrates mathematical foundations with cutting-edge AI development across four interconnected areas.
High-Performance Optimization Algorithms
Developing large-scale optimization algorithms for training complex models on medical and engineering data.
- Nonconvex & large-scale optimization for deep learning
- Stochastic optimization with convergence guarantees
- SVM with novel loss functions (Generalized Pinball, Rescaled)
- Neurodynamic neural network methods
Provable Deep Learning Architectures
Designing new deep learning architectures with provable mathematical properties — convergence, stability, and generalization bounds.
- Mathematically-grounded deep architectures
- Convergence and stability proofs for DL models
- Generalization bounds for medical imaging models
- Geometry of neural networks and kernel methods
Explainable & Trustworthy AI
Making AI transparent and interpretable using mathematical tools — differential geometry, spectral theory, and information geometry.
- Spectral graph theory for model interpretability
- Information geometry for understanding neural networks
- Model cards and reproducibility standards
- Auditable and deployable AI systems
Real-World Applications: Medical & Energy AI
Applying mathematical AI to real problems — medical imaging diagnosis, renewable energy systems, and smart grid optimization.
- Glaucoma detection from retinal fundus images
- Bone mineral density prediction from MRI/CT/X-ray
- Diabetic retinopathy recognition
- Wind turbine control and solar energy optimization
- Smart grid and microgrid system analysis
Our People
Multi-Disciplinary Team
7 core researchers from 4 departments, supported by 7 international collaborators.
Prof. Dr. Rabian Wangkeeree
Professor
Assoc. Prof. Dr. Chatchai Sirisamphanwong
Associate Professor
Assoc. Prof. Dr. Rattanaporn Wangkeeree
Associate Professor
Assoc. Prof. Dr. Kasamsuk Ungchittrakool
Associate Professor
Asst. Prof. Dr. Kotchaporn Karoon
Assistant Professor
Dr. Limpapat Bussaban
Lecturer
Dr. Rataporn Ngoenmeesri
Lecturer
International Research Network
Prof. Lam Quoc Anh
Can Tho University
Vietnam
Prof. Ovidiu Bagdasar
University of Derby
United Kingdom
Prof. Jein-Shan Chen
National Taiwan Normal University
Taiwan
Prof. Gue Myung Lee
Pukyong National University
South Korea
Prof. Alireza Nazemi
Shahrood University of Technology
Iran
Asst. Dr. Yirga Abebe Belay
Aksum University
Ethiopia
Dr. Pham Thi Vui
Can Tho University
Vietnam
Research Keywords
Towards a National Center of Excellence
DeepMathAI aims to become the leading “Mathematical AI” research group in Thailand — producing transparent, trustworthy, and deployable AI with strong theoretical foundations.