A5075630930- Human Pose and Action Recognition
- AI in cancer detection
- Machine Learning in Healthcare
- Advanced Neural Network Applications
- Medical Image Segmentation Techniques
- Ethics in Clinical Research
- Time Series Analysis and Forecasting
- Single-cell and spatial transcriptomics
- Ethics and Social Impacts of AI
- Explainable Artificial Intelligence (XAI)
- Artificial Intelligence in Healthcare and Education
- Radiomics and Machine Learning in Medical Imaging
- Machine Learning and Data Classification
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
Cornell University
2023-2025
Stanford University
2025
Stanford Medicine
2025
Institute of Behavioral Sciences
2025
Weill Cornell Medicine
2024
Interpretability for machine learning models in medical imaging (MLMI) is an important direction of research. However, there a general sense murkiness what interpretability means. Why does the need MLMI arise? What goals one actually seek to address when needed? To answer these questions, we identify formalize and elements MLMI. By reasoning about real-world tasks common both image analysis its intersection with learning, five core interpretability: localization, visual recognizability,...
Longitudinal imaging data are routinely acquired for health studies and patient monitoring. A central goal in longitudinal is tracking relevant change over time. Traditional methods remove nuisance variation with custom pipelines to focus on significant changes. In this work, we present a machine learning–based method that automatically ignores irrelevant changes extracts the time-varying signal of interest. Our method, called Learning-based Inference imAge Changes (LILAC), performs pairwise...
The convolutional neural network (CNN) is one of the most commonly used architectures for computer vision tasks. key building block a CNN kernel that aggregates information from pixel neighborhood and shares weights across all pixels. A standard CNN's capacity, thus its performance, directly related to number learnable weights, which determined by channels size (support). In this paper, we present \textit{hyper-convolution}, novel implicitly encodes using spatial coordinates....