About

Jun Zhu is a Professor in the Department of Statistics at the University of Wisconsin-Madison and is an Affiliate Faculty member in the Department of Entomology. His educational background includes a PhD in Statistics from Iowa State University, an MSE in Mathematical Sciences from Johns Hopkins University, and a BA in Mathematics and Computer Science from Knox College. His research interests encompass spatial statistics, spatio-temporal statistics, Markov random fields, agricultural statistics, environmental statistics, statistical ecology, environmental and population health, disease mapping, and medical imaging. His main research activities involve the development of statistical methodologies for analyzing spatially referenced data and data sampled over time, which are common in biological, physical, and social sciences. Additionally, he collaborates with research scientists to apply modern statistical methods, especially spatial and spatio-temporal statistics, to studies in agricultural, biological, ecological, environmental, health, and social systems. His research program emphasizes a close connection between methodological development and scientific collaboration, where problems identified in collaborative research motivate new statistical methods, and these methods are subsequently applied in his projects.

Research topics

• Geography
• Biology
• Ecology
• Geology
• Remote sensing
• Environmental science
• Soil science
• Meteorology
• Climatology

Selected publications

• To Return or Not to Return: Family Affluence as a Determinant of Inter-Provincial Return Migration Among University Graduates

  Chinese Education & Society · 2026-05-22

  article1st author
  PublisherDOI

• Localization of Virtual Sound Source Reproduced by the Crosstalk Cancellation System Under Different Reflective Conditions

  Archives of Acoustics · 2025-10-09

  articleOpen access

  This study explores the localization of virtual sound sources reproduced by a crosstalk cancellation system under different reflective conditions in virtual rooms, analyzing the results with binaural cues. Binaural room impulse responses were generated using the high-order image source method. By modifying the acoustic parameters of the virtual room to manipulate reflection intensity and temporal structure, psychoacoustic experiments were conducted using headphone reproduction. Results show that variations in reflection intensity within a certain range, achieved by altering the room reverberation time (RT), do not significantly affect virtual source localization. However, increasing the loudspeaker–listener distance (altering the temporal structure of reflections) deteriorates localization performance. The main difference between changes in loudspeaker–listener distance and RT lies in whether the reflection’s temporal structure changes. The study highlights the critical role of reflection temporal structure in virtual source localization. Binaural cue analysis shows that even in reverberant environments, interaural time difference (ITD) remains more consistent with localization accuracy than interaural level difference (ILD).

  PublisherDOI

• Bayesian analysis and variable selection for spatial count data with an application to Rio de Janeiro gun violence

  Spatial Statistics · 2025-02-27

  articleSenior author
  PublisherDOI

• Towards the Worst-case Robustness of Large Language Models

  ArXiv.org · 2025-01-31

  preprintOpen accessSenior author

  Recent studies have revealed the vulnerability of large language models to adversarial attacks, where adversaries craft specific input sequences to induce harmful, violent, private, or incorrect outputs. In this work, we study their worst-case robustness, i.e., whether an adversarial example exists that leads to such undesirable outputs. We upper bound the worst-case robustness using stronger white-box attacks, indicating that most current deterministic defenses achieve nearly 0\% worst-case robustness. We propose a general tight lower bound for randomized smoothing using fractional knapsack solvers or 0-1 knapsack solvers, and using them to bound the worst-case robustness of all stochastic defenses. Based on these solvers, we provide theoretical lower bounds for several previous empirical defenses. For example, we certify the robustness of a specific case, smoothing using a uniform kernel, against \textit{any possible attack} with an average $\ell_0$ perturbation of 2.02 or an average suffix length of 6.41.

  PublisherOA PDFDOI

• A Multi-Sensor Fusion Approach for the Assessment of Water Stress in Woody Plants

  Forests · 2025-11-27

  articleOpen access1st author

  Climate change poses significant threats to forest ecosystems, with drought stress being a major factor affecting tree growth and survival. The accurate and early diagnosis of plant water status is, therefore, critical for advancing climate-smart forestry. However, traditional monitoring approaches often rely on single-sensor data or manual field surveys, limiting their capacity to comprehensively capture the complex physiological and structural dynamics of plants under water deficit. To address this gap, this study developed an indoor multi-sensor phenotyping platform, based on a three-axis mobile truss system, which integrates a hyperspectral camera, a thermal infrared imager, and a LiDAR scanner for coordinated high-throughput data acquisition. We further propose a novel hybrid model, the Whale Optimization Algorithm-based Multi-Kernel Extreme Learning Machine (WOA-MK-ELM), which enhances classification robustness by adaptively fusing hyperspectral and thermal features within a dual Gaussian kernel space. We use Perilla frutescens as a model species, achieving an accuracy of 93.03%, an average precision of 93.11%, an average recall of 94.04%, and an F1-score of 0.94 in water stress degree classification. The results demonstrate that the proposed framework not only achieves high prediction accuracy but also provides a powerful prototype and a robust analytical approach for smart forestry and early warning systems.

  PublisherDOI

• Spatial and spatio-temporal cluster detection using stacking

  Spatial Statistics · 2025-09-24

  article
  PublisherDOI

• Animal Trajectory Imputation and Uncertainty Quantification via Deep Learning

  Environmetrics · 2025-07-23

  articleOpen access

  ABSTRACT Imputing missing data in animal trajectories is crucial for understanding animal movements during unobserved periods. However, the traditional methods, such as linear interpolation and the continuous‐time correlated random walk model, are often inadequate to capture the complexity of animal movements. Here, we develop a deep learning approach to animal trajectory imputation by a conditional diffusion model. Unlike the traditional methods, our deep learning method uses observed data and external covariates to impute missing positions along an animal trajectory, capturing periodic patterns and the influence of covariates, which leads to more accurate imputations. In a case study of imputing deer trajectories, our method not only provides more accurate deterministic imputations than existing approaches but also achieves uncertainty quantification through probabilistic imputation.

  PublisherDOI

• A sequential importance sampling method for efficiently estimating posterior augmented mean of structural performance

  Structures · 2025-02-27

  article1st author
  PublisherDOI

• A Mixture Scan Statistic for Semi-continuous Data with Covariates and Spatial Correlation

  Journal of Agricultural Biological and Environmental Statistics · 2024-11-04

  article
  PublisherDOI

• Accurate and Reliable Predictions with Mutual-Transport Ensemble

  arXiv (Cornell University) · 2024-05-30

  preprintOpen access

  Deep Neural Networks (DNNs) have achieved remarkable success in a variety of tasks, especially when it comes to prediction accuracy. However, in complex real-world scenarios, particularly in safety-critical applications, high accuracy alone is not enough. Reliable uncertainty estimates are crucial. Modern DNNs, often trained with cross-entropy loss, tend to be overconfident, especially with ambiguous samples. To improve uncertainty calibration, many techniques have been developed, but they often compromise prediction accuracy. To tackle this challenge, we propose the ``mutual-transport ensemble'' (MTE). This approach introduces a co-trained auxiliary model and adaptively regularizes the cross-entropy loss using Kullback-Leibler (KL) divergence between the prediction distributions of the primary and auxiliary models. We conducted extensive studies on various benchmarks to validate the effectiveness of our method. The results show that MTE can simultaneously enhance both accuracy and uncertainty calibration. For example, on the CIFAR-100 dataset, our MTE method on ResNet34/50 achieved significant improvements compared to previous state-of-the-art method, with absolute accuracy increases of 2.4%/3.7%, relative reductions in ECE of $42.3%/29.4%, and relative reductions in classwise-ECE of 11.6%/15.3%.

  PublisherOA PDFDOI

Frequent coauthors

• Brian H. Aukema

  36 shared

• Kenneth F. Raffa

  University of Wisconsin–Madison

  31 shared

• Jakob Gulddahl Rasmussen

  Aalborg University

  27 shared

• Jesper Møller

  27 shared

• Tingjin Chu

  University of Melbourne

  18 shared

• Y. A. Chang

  16 shared

• Haonan Wang

  National University of Singapore

  14 shared

• Pei‐Sheng Lin

  National Health Research Institutes

  11 shared

Education

• Ph.D., Entomology

  University of California, Berkeley

  1993

• M.S., Entomology

  University of California, Berkeley

  1989

• B.S., Entomology

  Nanjing Agricultural University

  1985