Center of Complex Systems
Special mission: understanding emergent behavior, nonlinear dynamics, and adaptive systems through computational modeling and data-driven analysis.
Focus
- Medical Imaging: Deep learning pipelines for segmentation, classification, and anomaly detection in clinical imaging data (MRI, CT, X-ray).
- Agricultural Imaging: Remote sensing and computer vision for crop health monitoring, disease detection, and yield estimation.
- Graph Theory: Graph-theoretic modeling of biological, social, and infrastructural networks to analyze structure, flow, and resilience.
- Topology: Topological data analysis (TDA) and persistent homology for extracting shape-based features from high-dimensional complex datasets.
- Federated Learning: Privacy-preserving distributed machine learning across decentralized data sources in medical and agricultural domains.
- Modeling and simulation of nonlinear dynamical systems and emergent phenomena.
- Agent-based modeling and self-organizing adaptive systems.
Research Domains
Medical Imaging
AI-assisted analysis of MRI, CT, and X-ray data. Developing segmentation models, disease classification pipelines, and clinical decision support tools using deep convolutional and transformer architectures.
Agricultural Imaging
Satellite and drone-based imaging for precision agriculture. Crop stress detection, field mapping, and growth stage classification using multispectral and RGB imagery.
Graph Theory
Structural analysis of complex networks using spectral graph theory, community detection, and flow optimization. Applications in biological pathways, infrastructure, and social systems.
Topology
Topological Data Analysis (TDA) using persistent homology and Mapper algorithms to identify hidden structure and shape-based signatures in high-dimensional data.
Federated Learning
Decentralized model training across hospitals, farms, and institutions without sharing raw data. Focus on robust aggregation, communication efficiency, and fairness under heterogeneous data distributions.
Dynamical Systems
Computational frameworks for simulating nonlinear, chaotic, and emergent system behavior. Agent-based and evolutionary models for studying self-organization and adaptation.
Active Projects
FedMed
Federated learning framework for multi-site medical image analysis with differential privacy guarantees and heterogeneity-robust aggregation.
TopoAgri
Topological and graph-based analysis of agricultural field data for early stress detection and precision intervention mapping.
GraphNet Bio
Graph neural network models for biological interaction networks — protein-protein, gene regulatory, and metabolic pathway analysis.