Research Fields

Our laboratory designs machine learning solutions for multi-scale biological frameworks and computational intelligence systems. Our research activities are divided into seven core streams:

Manifold learning visualization used for single-cell RNA-seq analysis

Single-Cell RNA-Seq, Genomics and Multi-Omics

Developing robust mathematical models to identify cell-type annotations and handle transcriptomics distributions. Projects focus on pattern-aware gene selection models, manifold learning algorithms, and privacy-preserving framework clusters.

Check out our related publications.
SEGCECO pipeline for cell-cell communication prediction

Cell-Cell Communication and Molecular Interactions

Formulating subgraph embedding methods to evaluate cell-to-cell communication networks. We investigate contrastive learning platforms to predict protein-protein interaction interfaces and map sequence constraints.

Check out our related publications.
Multi-level thresholding used in biomedical image segmentation

Biomedical Imaging and Image Analysis

Designing lightweight graph machine learning models for nodule segmentation and tracking on computed tomography (CT) images. We implement positional convolutions, Delaunay triangulations, and superpixel partitions to decrease false positives.

Check out our related publications.
Self-organizing map model used in recommender system research

Recommender Systems

Integrating Graph Attention Networks with modern Large Language Model (LLM) agents to enhance end-to-end personalization. Additionally, we explore adversarial systems like neural network detectors for review spam filtering.

Check out our related publications.
Network model diagram related to hypergraph neural network research

Hypergraph Neural Networks

Exploring structural hypergraph neural network systems to track spatial domains from single-cell transcriptomics data where relationships extend past basic pairwise connections.

Check out our related publications.
Kernel PCA pattern separation used in associative memory research

Associative Memory Models

Utilizing modern Hopfield neural network architectures to advance out-of-distribution (OOD) processing security, data pattern recovery, and latent prototype discovery.

Check out our related publications.
Breast cancer subtypes classification tree

Cancer Research and Clinical Disease Diagnostics

Developing multi-omic data integration pipelines to identify clinical biomarkers and mutations across colorectal, bladder, and breast cancer subtypes.

Check out our related publications.