Contact
Location
Rice 310
85 Engineer's Way
Charlottesville, VA 22904

Xi Peng

Copenhaver Fellow Associate Professor of Computer Science Visiting Investigator of MSK Cancer Center
Contact
Location
Rice 310
85 Engineer's Way
Charlottesville, VA 22904
Homepage Google Scholar DeepREALab

About

*** PhD and Postdoc positions are available (Apply Here)

*** UVA Undergraduate or Master research intern (Apply Here)

I am an Associate Professor in Computer Science at the University of Virginia (UVA) and a Visiting Investigator at Memorial Sloan Kettering Cancer Center (MSK).

My research is driven by a fundamental question: How can we develop AI systems that people can truly trust? I believe trustworthy AI requires more than accurate predictions. An AI system should remain reliable when the world changes and, especially in high-stakes settings, provide evidence for why its predictions should be trusted.

At UVA, I lead DeepREALab (Deep Robust & Explainable AI Lab), where we develop trustworthy machine learning and computer vision methods with applications in health, science, and autonomous systems. Our work spans robust and out-of-distribution learning, interpretable and verifiable AI, vision-language and foundation models, and safe learning-enabled systems.

Our research has been supported by NSF, NIH, DoD, CDC, MSK Cancer Center, Google, Snap, and other organizations, with major recognitions including the NSF CAREER Award, DoD DEPSCoR Award, NIH R21 Award, NSF SLES Award, and Google Faculty Research Award.

Education

PhD in Computer Science, Rutgers University, 2018

MS in Computer Science, Institute of Automation, Chinese Academy of Sciences, 2011

BS in Automation Science, Beihang University, 2008

Research Interests

Machine Learning Develop robust and explainable AI systems that people can trust.
Computer Vision Tackle multimodal changes in health, science, and cyber-physical systems.
Mechanistic Interpretability Understand how AI models reason and ensure reasoning are supported by valid evidence.
Robust Generalization Develop AI models that remain reliable under distribution shifts, rare events, and previously unseen conditions.

Selected Publications

[ICML'26] Inside the Visual Mind: Neuroscience-Motivated Concept Circuits
for Interpreting and Steering Vision Transformers
Paper
[CVPR'26 Highlight] Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings
Paper
[ICML'25] “Why Is There a Tumor?”: Tell Me the Reason, Show Me the Evidence
Paper
[NeruIPS'24] Beyond Accuracy: Ensuring Correct Predictions With
Correct Rationales
Paper
[ICLR'23] Topology-aware Robust Optimization for Out-of-Distribution Generalization
Paper
[AAAI'21] SMIL: Multimodal Learning with Severely Missing Modality
Paper
[CVPR'20] Learning to Learn Single Domain Generalization
Paper

Courses Taught

CS6501 Trustworthy Machine Learning 2026 Fall

Awards

Copenhaver Fellow UVA, 2026
Outstanding Faculty Research Award U Delaware, 2024
NSF CAREER Award 2024
DoD DEPSCoR Award 2023
Google Faculty Research Award 2022
General University Research Award U Delaware, 2022
Research Foundation Award U Delaware, 2021

Featured Grants & Projects

Integrating Radiologist Insights for Safe and Accurate AI-Assisted Prostate MRI Interpretation Sponsor: NIH, MSK Cancer Center
ProstateVLM
SeafloorAI: Large-scale Vision-Language Dataset and Model for Seafloor Geological Survey Sponsor: DoD, ONR, NSF
SearfloorAI
OSLA: Orchestrating Model, System, and Hardware for Safe Learning in Autonomous Vehicles Sponsor: NSF Safe Learning-Enabled Systems
OSLA