Dr. Xi Peng (Peter), 彭曦, AI/ML Scientist & Educator

Welcome! I am a computer science faculty at the University of Delaware (UD):

Email: xipeng at udel dot edu   Tel: (302) 831-2876   

Office: FinTech 416C, 591 Collaboration Way, Newark, DE 19713

Google Scholar Deep-REAL Lab (Deep Robust & Explainable AI Lab)

Short Bio

Dr. Peng is leading the Deep-REAL Lab (Deep Robust & Explainable AI Lab) at the University of Delaware. His research interests primarily focus on two areas: (1) Trustworthy Machine Learning, specifically in building algorithm foundations for robustness, explainability, and reliability; and (2) AI for Sciences, for safety-critical applications in Geo and Bio Sciences. His group publishes on top AI/ML venues such as NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, AAAI, IJCAI, KDD, and TPAMI. According to csrankings.org, Dr. Peng is ranked as the top individual in the CIS department, and the second highest across the entire university. His research work has garnered recognition, including the Best Paper Award at the NeurIPS'21 MLPH workshop and the Best Student Paper Finalist at ECCV'16. His research has received support from NSF, DOD, CDC, industry such as Memorial Sloan Kettering Cancer Center, Google Faculty Research Award, Snap Research Award, and internal research awards such as General University Research Award and University of Delaware Research Foundation Award. He received Ph.D. in Computer Science advised by Chair and Distinguished Professor Dimitris N. Metaxas from Rutgers University in 2018.

Research Interests

Trustworthy Machine Learning (methodology): I am interested in researching and developing principled models, algorithms, and theory to solve fundamental machine learning problems regarding robustness, explainability , and scalability:

AI for Sciences (applications): Building upon established methodologies, my research aims to develop AI/ML systems that are trustworthy, reliable, and safe to use in high-stakes scientific applications:

Top conference and journal papers (by my students) since joined UD in 2019 (A Full List)

[ICCV'23] Learning from Semantic Alignment between Unpaired Multiviews for Egocentric Video Recognition [PDF] [Code]

[CVPR'23] Are Data-driven Explanations Robust against Out-of-Distribution Data? [PDF] [Code]

[ICLR'23] Topology-aware Robust Optimization for Out-of-Distribution Generalization. [PDF] [Code]

[TNNLS'23, IF=14.3] Semi-identical Twins Variational AutoEncoder for Few-Shot Learning. [PDF]

[TPAMI'22, IF=24.3] Out-of-Domain Generalization from a Single Source: An Uncertainty Quantification Approach. [PDF] [Code]

[TMM'22, IF=8.2] Region-aware Arbitrary-shaped Text Detection with Progressive Fusion. [PDF] [Code]

[CVPR'22] Are multimodal transformers robust to missing modality? [PDF] [Code]

[CVPR'22] Symmetry and uncertainty-aware object slam for 6dof object pose estimation. [PDF] [Code]

[NeurIPS'21W Best Paper Award] Deep learning for spatiotemporal modeling of Urbanization. [PDF] [Video-10m]

[ICLR'21 Spotlight] A good image generator is what you need for high-resolution video synthesis. [PDF] [Video-10m] [Code]

[CVPR'21] Uncertainty-guided Model Generalization to Unseen Domains. [PDF] [Video-5m] [Code]

[CVPR'21 Oral] Learning View-Disentangled Human Pose Representation by Contrastive Cross-View Mutual Information Maximization. [PDF] [Video-5m] [Code]

[AAAI'21] Multimodal learning with severely missing modality. [PDF] [Video-60s] [Video-15m] [Code]

[NSDI'21] Adapting Wireless Mesh Network Configuration from Simulation to Reality via Deep Learning-based Domain Adaptation. [PDF]

[IJCV'20, IF=11.5] Towards image-to-video translation: A structure-aware approach via multi-stage generative adversarial networks. [PDF]

[NeurIPS'20] Maximum-entropy adversarial data augmentation for improved generalization and robustness. [PDF] [Code]

[CVPR'20] Learning to learn single domain generalization. [PDF] [Video-60s] [Code]

[CVPR'20] Knowledge as priors: Cross-modal knowledge generalization for datasets without superior knowledge. [PDF] [Video-60s]

[TPAMI'19, IF=24.3] Towards Efficient U-Nets: A Coupled and Quantized Approach. [PDF]

[NeurIPS'19] Semantic-guided multi-attention localization for zero-shot learning. [PDF]

[NeurIPS'19] Rethinking kernel methods for node representation learning on graphs. [PDF] [Code]

[ICCV'19 Oral] AdaTransform: Adaptive Data Transformation. [PDF]

[CVPR'19] Semantic graph convolutional networks for 3d human pose regression. [PDF]

[KDD'19] Scalable Global Alignment Graph Kernel Using Random Features: From Node Embedding to Graph Embedding. [PDF]

Students

Advising PhD students:

Visiting PhD students:

Undergraduate students:

Teaching

Awards & Sponsors