About

Zhengyuan Zhou joined New York University Stern School of Business as an Associate Professor of Technology, Operations, and Statistics in September 2020. His research interests lie at the intersection of machine learning, stochastic optimization, control, and game theory, focusing on leveraging tools from these fields to develop methodological frameworks for solving data-driven decision-making problems. Prior to joining NYU Stern, Professor Zhou spent the year 2019-2020 as a Goldstine research fellow at IBM research and a visiting scholar at NYU Stern. He holds a BA in Mathematics and a BS in Electrical Engineering and Computer Sciences from UC Berkeley, and has earned multiple advanced degrees from Stanford University, including a Master’s in Computer Science, a Master’s in Statistics, a Master’s in Economics, and a PhD in Electrical Engineering with minors in Mathematics and Management Science & Engineering.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Mathematical optimization
• Mathematics
• Mathematical analysis
• Geometry
• Economics
• Market economy
• Algorithm
• Combinatorics
• Applied mathematics
• Econometrics
• Public economics
• Microeconomics

Selected publications

• Marine-derived PUFAs and peptides for aging-related memory deficits: processing methods, pharmacokinetics, mechanistic insights, and clinical application

  Frontiers in Nutrition · 2026-05-14

  articleOpen access

  Cognitive decline and memory disorders are increasingly prevalent globally, especially in aging populations, imposing substantial social, emotional, and medical burdens on individuals and healthcare systems. Food-derived dietary interventions play a critical role in the prevention and management of these conditions, with marine-derived polyunsaturated fatty acids (PUFAs) and bioactive peptides emerging as promising candidates for enhancing brain health and cognitive function. This review summarizes advanced processing techniques for these bioactive substances, including physicochemical methods for the extraction and purification of PUFAs, as well as enzyme-mediated degradation of marine proteins for peptide production. It also covers their multifaceted mechanisms underlying memory enhancement, such as antioxidant, anti-apoptotic, anti-inflammatory, and cholinergic modulation, supported by preclinical animal studies and preliminary human clinical trials. Finally, existing challenges such as low bioavailability and unstandardized formulations, along with prospects including sustainable production, personalized bioactive blends, and precision nutrition, are discussed. This review first clarifies the complementary neuroprotective mechanisms of marine-derived PUFAs and peptides, and advocates for systematic exploration to translate preclinical findings into clinical applications.

  PublisherOA PDFDOI

• Integrating multivariate selection indices with weighted rank aggregation to identify drought-tolerant Populus simonii × P. nigra F1 progenies

  BMC Plant Biology · 2026-01-19

  articleOpen access1st author

  BACKGROUND: Breeding drought-tolerant poplar cultivars necessitates efficient selection strategies that can simultaneously improve multiple traits. This study evaluated the integration of multivariate selection indices with Weighted Rank Aggregation (WRA) to identify superior genotypes in hybrid poplar progenies. METHODS: progenies from three families of Populus simonii × P. nigra under controlled drought stress and well-watered conditions. Data on 16 growth, leaf, and photosynthetic traits were analyzed using four multivariate indices: the Smith-Hazel Index (SHI), FAI-BLUP, and two Multi-Trait Genotype-Ideal Genotype Distance Index (MGIDI) variants. The rankings were integrated using WRA. RESULTS: Genetic parameters revealed high heritability for key growth traits. The selection indices exhibited divergent focus, with SHI showing strong directional selection for growth but sensitivity to multicollinearity, while FAI-BLUP and MGIDI enabled more balanced multi-trait improvements. Most indices were weakly correlated (Spearman's |r| < 0.2), indicating complementary information. Venn analysis identified genotypes (e.g., C4‑246, E4‑70) performing consistently across multiple indices. The final WRA integration selected robust genotypes, including C2‑65, C4‑210, E4‑115, and E4‑70, which combine drought tolerance with desirable growth and physiological characteristics. CONCLUSIONS: Integrating multiple selection indices with WRA provides a powerful and reliable strategy for selecting drought-tolerant poplar genotypes at the seedling stage. This approach effectively balances genetic gains across traits, enhancing the efficiency of breeding programs for stress resilience.

  PublisherDOI

• Weight-adjusted-waist index, estimated glucose disposal rate, C-reactive protein, and mortality risks among individuals with type 2 diabetes: 2 prospective cohort studies in China and the United Kingdom

  American Journal of Clinical Nutrition · 2025-09-13 · 2 citations

  article
  PublisherDOI

• Concurrent Learning with Aggregated States via Randomized Least Squares Value Iteration

  ArXiv.org · 2025-01-23

  preprintOpen accessSenior author

  Designing learning agents that explore efficiently in a complex environment has been widely recognized as a fundamental challenge in reinforcement learning. While a number of works have demonstrated the effectiveness of techniques based on randomized value functions on a single agent, it remains unclear, from a theoretical point of view, whether injecting randomization can help a society of agents {\it concurently} explore an environment. The theoretical results %that we established in this work tender an affirmative answer to this question. We adapt the concurrent learning framework to \textit{randomized least-squares value iteration} (RLSVI) with \textit{aggregated state representation}. We demonstrate polynomial worst-case regret bounds in both finite- and infinite-horizon environments. In both setups the per-agent regret decreases at an optimal rate of $Θ\left(\frac{1}{\sqrt{N}}\right)$, highlighting the advantage of concurent learning. Our algorithm exhibits significantly lower space complexity compared to \cite{russo2019worst} and \cite{agrawal2021improved}. We reduce the space complexity by a factor of $K$ while incurring only a $\sqrt{K}$ increase in the worst-case regret bound, compared to \citep{agrawal2021improved,russo2019worst}. Additionally, we conduct numerical experiments to demonstrate our theoretical findings.

  PublisherOA PDFDOI

• Sensitivity Analysis Under the <i>f</i> -Sensitivity Model: A Distributional Robustness Perspective

  Operations Research · 2025-12-10

  articleSenior author

  Rethinking Causal Inference Through Robust Sensitivity Models In this issue, a new study by researchers from The Wharton School and New York University introduces a breakthrough in causal inference for observational data. Traditional analyses often fail when hidden confounders distort cause-and-effect relationships, but the newly proposed f-sensitivity model tackles this challenge by measuring the “average” impact of unobserved confounding instead of its worst-case effect. This framework connects causal inference to distributionally robust optimization, providing more realistic and interpretable bounds on treatment effects. With novel estimation and debiasing techniques, the method achieves statistical validity, even under minimal assumptions. The approach offers a flexible, computationally efficient way to test how robust conclusions remain in the presence of uncertainty, marking a significant advance in data-driven decision making across economics, healthcare, and policy analysis.

  PublisherDOI

• Plasma Metabolites of Kynurenine Pathway and Risk of Stroke in the General Population: A Nested Case–control Study

  Translational Stroke Research · 2025-12-09 · 1 citations

  article
  PublisherDOI

• Delay Optimizing Based Passivity Enhancement of Converter-Side Current Controlled LCL-Type Grid Converters

  2025-05-11

  article1st authorCorresponding

  Converter-side current control has been widely used in LCL-type grid converters due to its cost-effectiveness. However, the converter suffers from instability at the grid impedance variations. To solve the problem, a delay-optimizing method is proposed with proportional capacitor voltage feedforward (CVFF) and proportional converter current feedback (CCFB). The models for the loop gain and the output impedance are analyzed, showing positive gain and increased delay in CCFB, enhancing the passive frequency range while keeping the control bandwidth. A design for the control parameters is carefully presented based on the analysis, and the effectiveness is validated by the simulation and experimental results.

  PublisherDOI

• Spatial epidemiological analysis based on township scale and analysis of influencing factors of pulmonary tuberculosis cure of Changshu city from 2015 to 2022

  PLoS ONE · 2025-01-16

  articleOpen accessCorresponding

  OBJECTIVE: This study aimed to enhance the prevention and control of pulmonary tuberculosis (PTB) and provide more effective and accurate methods in Changshu City. METHODS: The PTB patients' information came from the China Information System for Disease Control and Prevention (CISDCP). The demographic data for Changshu city and towns came from the Suzhou Statistical Yearbook and the LandScan platform. ArcGIS was used for global spatial autocorrelation analysis and local spatial autocorrelation analysis. Univariate logistic regression and multivariate logistic regression were used to analyze the influencing factors of cured PTB patients. The receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to analyze the predictive efficacy and clinical benefit of the indicators. XGBoost analysis was performed to explore the feature importance of key metrics for PTB outcome. RESULTS: A total of 3943 PTB cases were included. The annual incidence rate of new PTB in Changshu city was 27.081 per 100,000. Changshu High-tech Industrial Development Zone in Jiangsu Province and Shajiabang town were the high-high aggregation areas and hot spot areas. Diagnosis delay, TB strain types, and drug sensitivity were independent predictors of the cure of new PTB patients. CONCLUSION: The central and southern areas of Changshu were the high-high cluster areas and hot spots for PTB. Shorter diagnosis delay days and mycobacterium tuberculosis (MTB) promote the cure of tuberculosis, while drug sensitivity was a risk factor for its cure.

  PublisherOA PDFDOI

• From whales to waves: Social media sentiment, volatility, and whales in cryptocurrency markets

  The British Accounting Review · 2025-05-01 · 4 citations

  articleOpen access

  This paper examines the relationship between cryptocurrency market dynamics and investor sentiment, employing advanced techniques like time-variant Granger causality and asymmetric time-varying parameter vector autoregression (TVP-VAR) frequency connectivity. We create unique sentiment analysis tools, including a custom cryptocurrency sentiment lexicon, to deeply analyze content in the cryptocurrency domain, particularly focusing on investor discussions and viewpoints. Our findings demonstrate a significant, evolving link between market sentiment and cryptocurrency movements. A key observation is that the volatility of shock transmission is tightly connected to major market events, often influenced by large-scale investors, or “whales”. Our study indicates that market sentiment consistently affects both short- and long-term cryptocurrency volatility, underlining the crucial influence of investor sentiment in driving the dynamics of the cryptocurrency market. This underscores the importance of understanding investor sentiment for predicting and navigating the cryptocurrency market.

  PublisherDOI

• Calcium Supplement Combined with Dietary Supplement Kidtal Can Promote Longitudinal Growth of Long Bone in Calcium-Deficient Adolescent Rats

  Nutrients · 2025-06-10

  articleOpen access

  Objective: Growth retardation in adolescents caused by nutritional deficiency requires effective intervention. A novel dietary supplement containing bamboo shoot extract, amino acids, and calcium citrate (Kidtal + Ca, KDTCa) was evaluated for its growth-promoting effects. Methods: After acclimatization, sixty-three 3-week-old male Sprague-Dawley (SD) rats were randomly divided into a normal control group and model groups. Growth retardation was induced in the modeling groups through calcium-deficient feeding, followed by administration of KDTCa, bamboo shoot extract and amino acids (Kidtal), or calcium citrate (CC). After 6 weeks of intragastric administration, the mechanical properties, microstructure, and growth plate development of bone were evaluated using three-point bending, micro-CT, and H&amp;E staining, respectively. Bone calcium/phosphorus distribution and fecal calcium apparent absorption rate were measured by ICP-MS. Results: All inter-group differences were analyzed using one-way analysis of variance and checked using the Tuckey test. KDTCa treatment dose-dependently enhanced bone development in calcium-deficient rats. Compared to the model group, H-KDTCa significantly restored naso-anal length (p &lt; 0.05) and body weight (p &lt; 0.01). KDTCa supplementation significantly restored calcium and phosphorus levels in blood and bone. Three-point bending experiments showed that the stiffness and bending energy were increased by 142.58% and 384.7%. In bone microarchitecture, both bone mineral density (BMD) and microstructural parameters were significantly improved. These findings were consistent with the increased long bone length (p &lt; 0.05) and decreased serum BALP/TRACP levels (p &lt; 0.001). Dose-dependent IGF-1 elevation (p &lt; 0.01) potentially mediated growth plate elongation by 35.34%. Notably, KDTCa increased calcium apparent absorption by 6.1% versus calcium-only supplementation at equal intake. Conclusions: KDTCa improves bone microstructure and strength, restores bone metabolism, and enhances growth plate height via promoting IGF-1 secretion to facilitate bone development. Further studies are needed to determine whether the components and calcium in Kidtal have a synergistic effect.

  PublisherOA PDFDOI

Frequent coauthors

• Nicholas Bambos

  Stanford University

  56 shared

• Panayotis Mertikopoulos

  Laboratoire d'Informatique de Grenoble

  45 shared

• Zhirong Guo

  East China Jiaotong University

  37 shared

• Ming Wu

  Nanjing Medical University

  32 shared

• Wenshu Luo

  University of Zurich

  25 shared

• Xiao‐shu Hu

  25 shared

• Peter W. Glynn

  24 shared

• José Blanchet

  18 shared