Open to Collaboration

I am an M.S. student at Stanford University, working on reliable generative models and data-efficient learning techniques for vision.

I am always happy to discuss research collaboration and internship opportunities. Feel free to reach out via email.

Hi, I’m Ning Zhu, an M.S. student in Electrical Engineering at Stanford University. I received my Bachelor’s degree (with Honours of the First Class) from the University of Electronic Science and Technology of China (UESTC) in a joint program with the University of Glasgow. I have been fortunate to be advised by Prof. Guotai Wang at UESTC, and to collaborate with Prof. Han Liu and Jerry Yao-Chieh Hu at Northwestern University. My research interests span two complementary directions. The first is generative models, covering image and video generation, training-free generation, post-training techniques, and the theoretical foundations of generative modeling. The second is data-efficient learning, encompassing active learning, data curation, and unsupervised anomaly detection. Within this direction, I focus on selecting the most informative and highest-quality subsets from massive data to reduce annotation costs while improving downstream performance, and on learning strong representations from unlabeled data. My overarching goal is to build reliable and label-efficient models that can be trusted in real-world deployment across industry, healthcare, and robotics.

🔥 News

  • Jun 2026: Our paper “Dual-Space Cold-Start Active Learning Guided by SAM3 for Medical Image Segmentation” was accepted in [MICCAI’26].
  • Jun 2026: 🏆 Awarded the IET Prize by the Institution of Engineering and Technology (IET) — one of only ~93 students worldwide selected in 2025 from over 100 IET-accredited universities, nominated by the Glasgow College UESTC department in recognition of academic distinction across the bachelor’s degree.
  • May 2026: Our paper “Stabilizing Temporal Inference Dynamics for Online Surgical Phase Recognition” was accepted in [MICCAI’26].
  • Apr 2026: 🐧 Started a research internship at Tencent IEG (Game AI Engine Department), focusing on video generation post-training.
  • Sep 2025: Our paper “High-Order Flow Matching: Unified Framework and Sharp Statistical Rates” was accepted in [NeurIPS’25].
  • Sep 2025: Our paper “Deep Adaptive Wavelet Autoencoder with Mutually Independent Empirical Cumulative Distribution for Unsupervised Motor Anomaly Detection” was accepted in [EAAI’25].
  • Aug 2025: Our paper “MedCAL-Bench: A Comprehensive Benchmark on Cold-Start Active Learning with Foundation Models for Medical Image Analysis” was released on [arXiv].
  • Jul 2025: Our paper “Adversarial Frequency Component Reconstruction Constraint for Helicopter Vibration Signal Anomaly Detection: An Unsupervised Dual-Domain Approach” was published in [IEEE TIM’25].
  • May 2025: Our paper “CSAL-3D: Cold-Start Active Learning for 3D Medical Image Segmentation via SSL-Driven Uncertainty-Reinforced Diversity Sampling” was accepted in [MICCAI’25] (Best Paper & Young Scientist Awards Shortlist).
  • May 2025: Our paper “SUGFW: A SAM-Based Uncertainty-Guided Feature Weighting Framework for Cold Start Active Learning” was accepted in [MICCAI’25].
  • May 2025: Our paper “Unsupervised Anomaly Detection for Aircraft PRSOV with Random Projection-Based Inner Product Prediction” was published in [IEEE TIM’25].
  • Sep 2024: Our paper “An Adversarial Training Framework Based on Unsupervised Feature Reconstruction Constraints for Crystalline Silicon Solar Cells Anomaly Detection” was published in [IEEE TIM’24].

🔬 Research Interests

  • Generative Models — image and video generation, training-free generation, post-training techniques, and the theoretical foundations of generative modeling.
  • Data-Efficient Learning — active learning, data curation, and unsupervised anomaly detection, with applications across industry, healthcare, and robotics.

📑 Selected Publications

(*: indicates equal contribution; #: indicates corresponding author)

2026

MICCAI 2026 (Early Accept)
Stabilizing teaser
Stabilizing Temporal Inference Dynamics for Online Surgical Phase Recognition
Yang Liu*, Ning Zhu*, Jingjing Peng, Xiwu Chen, Alejandro Granados, Guotai Wang, Sebastien Ourselin
MICCAI 2026
Online surgical phase recognition via stabilized temporal inference dynamics for low-latency operating-room assistance.
arXiv 2026
Rectify Then Diffuse teaser
Rectify Then Diffuse: Disentangling Concepts Before Denoising Trajectory Unfolds
Ning Zhu, An Chen, Mengfei Zhao, Juntao Xu, Jingze Liang, Boyuan Gu, Liang-Jian Deng
arXiv 2026
A training-free framework that rectifies the initial concept allocation once before denoising, disentangling multiple concepts for faithful text-to-image compositional generation.
arXiv 2026
One Knob to Rule Them All teaser
One Knob to Rule Them All: A Unified Optimal Transport View of Cold-Start Active Learning
Ning Zhu, Xiaochuan Ma, Juntao Xu, Jingze Liang, Mengfei Zhao, An Chen, Liang-Jian Deng
arXiv 2026
A unified optimal-transport view of cold-start active learning with a data-adaptive Sinkhorn selection rule that automatically adapts to the data and task.

2025

arXiv 2025
MedCAL-Bench teaser
MedCAL-Bench: A Comprehensive Benchmark on Cold-Start Active Learning with Foundation Models for Medical Image Analysis
Ning Zhu*, Xiaochuan Ma*, Shaoting Zhang, Guotai Wang
arXiv 2025
A comprehensive benchmark studying cold-start active learning with foundation models for medical image analysis.
MICCAI 2025
CSAL-3D teaser
CSAL-3D: Cold-Start Active Learning for 3D Medical Image Segmentation via SSL-Driven Uncertainty-Reinforced Diversity Sampling
Ning Zhu, Ping Ye, Lanfeng Zhong, Qiang Yue, Shaoting Zhang, Guotai Wang
MICCAI 2025 Best Paper & Young Scientist Awards Shortlist
Cold-start active learning for 3D medical image segmentation using self-supervised uncertainty-reinforced diversity sampling.
MICCAI 2025
SUGFW teaser
SUGFW: A SAM-Based Uncertainty-Guided Feature Weighting Framework for Cold Start Active Learning
Xiaochuan Ma, Jia Fu, Lanfeng Zhong, Ning Zhu, Guotai Wang
MICCAI 2025
A SAM-based uncertainty-guided feature weighting framework for effective cold-start active learning.
NeurIPS 2025
High-Order Flow Matching: Unified Framework and Sharp Statistical Rates
Maojiang Su, Jerry Yao-Chieh Hu, Yi-Chen Lee, Ning Zhu, Jui-Hui Chung, Shang Wu, Zhao Song, Minshuo Chen, Han Liu
NeurIPS 2025
A unified high-order flow matching framework with sharp statistical convergence rates.
EAAI 2025
Wavelet motor teaser
Deep Adaptive Wavelet Autoencoder with Mutually Independent Empirical Cumulative Distribution for Unsupervised Motor Anomaly Detection
Pinze Ren*, Ning Zhu*, Dandan Peng, Liyuan Ren, Huan Wang
EAAI 2025
An unsupervised motor anomaly detector using a deep adaptive wavelet autoencoder with mutually independent ECD constraints.
IEEE TIM 2025
Helicopter teaser
Adversarial Frequency Component Reconstruction Constraint for Helicopter Vibration Signal Anomaly Detection: An Unsupervised Dual-Domain Approach
Ning Zhu, Tianzhi Xu Dong, Dandan Peng
IEEE TIM 2025
An unsupervised dual-domain method with adversarial frequency-component reconstruction for helicopter vibration anomaly detection.
IEEE TIM 2025
PRSOV teaser
Unsupervised Anomaly Detection for Aircraft PRSOV with Random Projection-Based Inner Product Prediction
Dandan Peng*, Ning Zhu*, Te Han, Zhuyun Chen, Chenyu Liu
IEEE TIM 2025
Unsupervised anomaly detection for aircraft pressure-regulating valves via random-projection inner-product prediction.

2024

IEEE TIM 2024
Solar cell teaser
An Adversarial Training Framework Based on Unsupervised Feature Reconstruction Constraints for Crystalline Silicon Solar Cells Anomaly Detection
Ning Zhu, Jing Wang, Ying Zhang, Huan Wang, Te Han
IEEE TIM 2024
An adversarial training framework with unsupervised feature-reconstruction constraints for solar-cell defect detection.

🎓 Education

Stanford University

Sep 2026 – Present (expected) · M.S. in Electrical Engineering

Graduate student in the Department of Electrical Engineering.

University of Electronic Science and Technology of China (UESTC), Glasgow College (Joint Program with University of Glasgow)

Sep 2022 – Jun 2026 · B.Eng. in Electronic Information Engineering

GPA: 3.99 · Rank: 1/247

Awarded the IET Prize by the Institution of Engineering and Technology (IET, est. 1871) — Europe's largest engineering professional society. One of only ~93 students selected globally in 2025 from over 100 IET-accredited universities, nominated by the Glasgow College UESTC department in recognition of distinction across the bachelor's degree.

Graduated with Honours of the First Class; Advisor: Prof. Guotai Wang.

🤝 Academic Services

  • Reviewer: MICCAI 2026, Pattern Recognition, IEEE Transactions on Sustainable Energy.

📚 Teaching

  • Teaching Assistant, Machine Learning and Artificial Intelligence (UoG3036), Semester 2024–2025.