Our research focuses on basic and applied research in machine learning. We develop approaches for anomaly detection, uncertainty estimation, out-of-distribution detection, few-shot learning, noisy label learning, as well as hardware-aware algorithms for model compression and efficient design of neural network architectures. Our applications include medical image analysis with problems such as cell counting and segmentation in histology images, organ segmentation from CT and MRI volumes, and multi-label medical image segmentation.
- Few-shot learning for medical images.
- Cell Segmentation.
- Multi-label image classification.