Portfolio
Here you’ll find summaries of my major academic projects as well as the tutorial I presented in MICCAI conference, which are representative of my skills and interests. All projects utilize the deep learning model to solve specific image analysis problem, both in medical and geophysical fields. Specifically, I focus on solving the less-label issue of deep learning by using meta-learning or self-supervised learning strategies. The tutorials here provides the basic knowledge about Bayesian learning and Bayesian meta-learning, as well as some popular approaches and our solutions in terms of the medical applications.
For all academic projects, you can find the related publication in my resume. There are also links to documents at the bottom of each project, which offer more depth. If you’re curious about technical details I’ve glossed over, feel free to contact me.
Adaptable Lung Nodule Classification
A memory-augmented capsule network for the rapid adaptation of CAD models to new domains.
AGILE: Few Shot Brain Cell Classification
A tAsk-auGmented actIve meta-LEarning (AGILE) brain cell type classifier.
DECAPS: Detail-Oriented Capsule Networks
A tAsk-auGmented actIve meta-LEarning (AGILE) brain cell type classifier.
Multimodal Breast Lesion Classification
Breast lesion risk estimation using both mammograms and clinical reports.
Phasetime: Nuclei Detection
A deep learning approach to detect nuclei in time lapse phase images.
Self-supervised Seismic Interpolation
A self-supervised learning approach for anti-aliasing seismic data interpolation.
First Arrival Picking
A robust first-arrival picking workflow using convolutional and recurrent neural networks.
Unsupervised Seismic Deblending
Seismic deblending by self-supervised deep learning with a blind-trace network.
