Peng Cao is a PhD candidate who has demonstrated innovative use of MathWorks tools to make fundamental contributions to machine learning and digital health. Specifically, Peng’s research focuses on sensing systems, data analysis, and radio signals. MATLAB is an essential tool in this work, for numerical optimization, complex signal processing, and analysis and manipulation of radio signals and electromagnetic fields. The MathWorks Fellowship will support Peng’s cutting-edge research in several domains. She has developed new approaches to learning from crowdsourcing labels, including robust information-theoretic loss functions that enable learning from multiple labels and noisy labels, as well as semi-supervised, multimodal learning. Peng has also made notable advances in human-motion modeling with a transformer-based deep learning model for generative modeling of 3-D human motion. Another project yielded a new approach to enable robots to move quickly in indoor environments without colliding with people—one of the first such projects to demonstrate indoor localization around corners using radio signals. Most recently, Peng developed the first contactless system to monitor blood oxygen by employing a neural network to assess oxygen levels within 2 to 3 degrees. This work, which was validated with Covid-19 patients, has enormous potential for telehealth applications.