Research
Vision
The Machine Learning Laboratory research focuses on building human-machine collaborative AI systems that can learn context-aware and explainable models that are robust to both noise and limited supervision. We develop deep multi-source learning techniques including reinforcement, few/zero-shot, relational/graph and self-supervised learning. We explore multi-source learning strategies including knowledge-enhanced learning (prompts, interdependency structure, data augmentation), multimodal learning (text, image, video, audio, sensor), multi-model learning (ensemble, federated), and hybrid learning (data-driven, physical-model-based). We develop our research in the context of natural language processing, time-series prediction, and control. Our research spans applications such as AI ethics, computer vision, cyber security and healthcare.
Research Topics
- Multisource Learning for Natural Language Information Extraction
- Multisource Learning for Timeseries Prediction
- Multisource Learning for Control
Applications
Research Sponsors
- Intelligence Advanced Research Projects Activity (IARPA)
- National Science Foundation (NSF)
- Department of Energy (DoE)
- National Institute of Health (NIH)
- Commonwealth Cyber Initiative (CCI)
- Siemens
- Adobe
- Virginia Tech