- Construct a large vision model (LVM) for general medical imaging with self-supervised learning algorithm by second-order graph matching and large-scale medical imaging data (NeurIPS main track 2023).- Investigate the robustness of large vision models (Segment Anything, LVM-Med) on the out-of-distribution scenarios (R0-FoMo workshop, NeurIPS 2023).- Construct a Large Vision - Language model (LVLM) for general medical tasks (e.g., classification, segmentation, medical question answering, medical reports summarization) by introducing a fine-tuned large language model on large-scale medical text data with an adapter algorithm to connect between large models.

- Integrating multi-omics and machine learning to predict synthetic lethality in cancer cell lines.- Develop deep vision models for human recognition and human activity tracking for automated air-conditioner controlling problem.- Integrate, deploy and pilot ML models on IoT devices (run via Android OS).

- Detect and categorize 12 different types of lung-based lesions in H. sapiens from Chest X-Rays and Computed Tomography images with Image processing techniques as well as Deep convolutional, Transformers-based, and Graph-based Neural networks.- Apply Linear programming and nature-inspired optimized algorithms to find sub-optimal solutions for the Travelling Salesman Problem family.- Discover properties of deep learning based approaches (e.g., Deep Q Network families) under continuous controlling problems.- Develop multi-objective optimization (from more than 3 contrasting objectives) algorithms and strategies for Training-free Neural Architecture Search.

Teaching assistant for:- IT003: Data Structures & Algorithms (Spring semester - 2020)- IT002: Object-Oriented Programming (Fall semester - 2021)

Selected by Google to be one of 700 GDSC Leads around the world and to join a network of global leaders with the shared aim of empowering people through innovation and technology.