Pingkun Yan is receiving an award by the National Science Foundation for his proposal "CAREER: Systematic Mitigation of Deep Learning Adversaries in Medical Imaging". This is a five year award and is considered NSF's flagship award for new investigators. A detailed description of the summary of the proposed work can be found below.
About the NSF CAREER award:
The Faculty Early Career Development (CAREER) Program is a Foundation-wide activity that offers the National Science Foundation's most prestigious awards in support of early-career faculty who have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organization. Activities pursued by early-career faculty should build a firm foundation for a lifetime of leadership in integrating education and research.
Project Abstract:
With enormous amounts of data acquired by large-scale healthcare systems, computational data analysis has become an essential component in healthcare applications to process and extract information. Deep learning, a subspecies of artificial intelligence (AI), has established itself as a paradigm-shifting technology for data analytics due to its powerful ability to extract high-level data representations. However, deep learning is known to be vulnerable to adversaries, which cause algorithms to yield dramatically different results by making very small alterations to regular data samples. Adversaries are particularly hazardous in medical imaging applications where an altered image may lead an AI algorithm to cause medical errors. Thus, there is an urgent need to innovate and build robust healthcare cyberinfrastructure against deep learning adversaries. This project develops novel AI techniques to tackle the unprecedented challenges of adversaries in medical imaging applications from a new systematic standpoint. It brings awareness to potential issues when implementing AI in healthcare and develops new tools to mitigate these issues. This research will bolster confidence in adopting AI to improve healthcare efficiency and will also attract and train the next generation of AI researchers and engineers.
This project aims to develop innovative AI techniques to systematically mitigate deep learning adversaries in medical imaging applications. This project is very timely as deep learning is already widely used in image reconstruction, quality enhancement, computer-aided diagnosis, and image-guided intervention and surgery. Several challenges, including detection and rectification of adversaries as well as robust algorithm training across data domains, must be resolved before achieving robust medical imaging applications. Existing methods are concerned with only the deep learning algorithms themselves and try to build universal blind robustness against arbitrary adversaries, which overlooks the upstream data characteristics and downstream task specificities. This research adopts a holistic approach and is organized around a series of integrated subtopics, including detecting individual adversarial images, differentiating adversarial images from different sources, rectifying adversarial images, determining the transferability of robustness across data domains, and quantifying output uncertainties. The research will provide new insights, accurate yet robust AI techniques, and novel strategies to improve the robustness of medical imaging applications.