About

Dr. Ming Zhang is a Professor and the Program Director of Community and Regional Planning in the School of Architecture at the University of Texas at Austin. His research and teaching interests include urban and regional planning with a focus on transportation, the relationship between the built environment and travel behavior, economic and equity impacts of transportation investments, and international planning issues. His recent research has concentrated on megaregional transportation issues, Transit-Oriented Development, high-speed rail, and urban spatial development. Dr. Zhang has held various academic and professional positions, including tenure-track Assistant Professor at Texas A&M University, Research Scientist at the Rockefeller Institute of Government in Albany, New York, and Lecturer and licensed Planner/Architect at Huazhong University of Science & Technology in Wuhan, China. He is also the Director of the University Transportation Center (UTC) Cooperative Mobility for Competitive Megaregions (CM2) granted by the U.S. Department of Transportation.

Research topics

• Computer Science
• Business
• Environmental planning
• Engineering
• Environmental science
• Geography
• Political Science
• Environmental resource management
• Civil engineering
• Environmental protection
• Transport engineering
• Ecology
• World Wide Web
• Public relations

Selected publications

• Attention Bootstrapping for Multi-Modal Test-Time Adaptation

  ArXiv.org · 2025-03-03

  preprintOpen accessSenior author

  Test-time adaptation aims to adapt a well-trained model to potential distribution shifts at test time using only unlabeled test data, without access to the original training data. While previous efforts mainly focus on a single modality, test-time distribution shift in the multi-modal setting is more complex and calls for new solutions. This paper tackles the problem of multi-modal test-time adaptation by proposing a novel method named Attention Bootstrapping with Principal Entropy Minimization (ABPEM). We observe that test-time distribution shift causes misalignment across modalities, leading to a large gap between intra-modality discrepancies (measured by self-attention) and inter-modality discrepancies (measured by cross-attention). We name this the attention gap. This attention gap widens with more severe distribution shifts, hindering effective modality fusion. To mitigate this attention gap and encourage better modality fusion, we propose attention bootstrapping that promotes cross-attention with the guidance of self-attention. Moreover, to reduce the gradient noise in the commonly-used entropy minimization, we adopt principal entropy minimization, a refinement of entropy minimization that reduces gradient noise by focusing on the principal parts of entropy, excluding less reliable gradient information. Extensive experiments on the benchmarks validate the effectiveness of the proposed ABPEM in comparison with competing baselines.

  PublisherOA PDF

• [Source Analysis and Ecological Risk Assessment of Heavy Metals in Soil of the Yifeng County， Jiangxi Province Based on APCS-MLR Receptor Model].

  PubMed · 2025-07-08

  article

  ， respectively， which were higher than the soil background values of the Jiangxi Province. Controlled by geological settings， the high content areas of As， Cr， Cu， and Ni were consistent with the distribution of the Qingbaikou metamorphic rock formations in the area， and the spatial distribution of Cd， Hg， Pb， and Zn showed significant differences from the distribution of rock formation types due to anthropogenic activities. The ecological risk assessment found that the risk of soil heavy metal pollution in the Yifeng County was relatively low on the whole， but there were local occurrences of slight pollution and moderate risk mainly caused by As， Cd， and Hg. The APCS-MLR receptor model identified three sources of heavy metals. Arsenic， Cr， Cu， and Ni were predominantly derived from natural sources， with contribution rates of 80.06%， 69.92%， 80.49%， and 94.27%， respectively. Lead and Zn primarily originated from transportation activities， contributing 74.73% and 66.99% to the overall contamination. Furthermore， Cd and Hg were predominantly influenced by industrial and agricultural sources， with respective contribution rates of 43.62% and 58.09%. The above research can provide data and theoretical support for soil pollution prevention and control in the Yifeng County.

  PublisherDOI

• Attention Bootstrapping for Multi-Modal Test-Time Adaptation

  Proceedings of the AAAI Conference on Artificial Intelligence · 2025-04-11 · 1 citations

  articleOpen accessSenior author

  Test-time adaptation aims to adapt a well-trained model to potential distribution shifts at test time using only unlabeled test data, without access to the original training data. While previous efforts mainly focus on a single modality, test-time distribution shift in the multi-modal setting is more complex and calls for new solutions. This paper tackles the problem of multi-modal test-time adaptation by proposing a novel method named Attention Bootstrapping with Principal Entropy Minimization (ABPEM). We observe that test-time distribution shift causes misalignment across modalities, leading to a large gap between intra-modality discrepancies (measured by self-attention) and inter-modality discrepancies (measured by cross-attention). We name this the attention gap. This attention gap widens with more severe distribution shifts, hindering effective modality fusion. To mitigate this attention gap and encourage better modality fusion, we propose attention bootstrapping that promotes cross-attention with the guidance of self-attention. Moreover, to reduce the gradient noise in the commonly-used entropy minimization, we adopt principal entropy minimization, a refinement of entropy minimization that reduces gradient noise by focusing on the principal parts of entropy, excluding less reliable gradient information. Extensive experiments on the benchmarks validate the effectiveness of the proposed ABPEM in comparison with competing baselines.

  PublisherOA PDFDOI

• Understanding the emerging interregional travel amid shifting societal and technological trends

  Travel Behaviour and Society · 2025-09-17

  articleSenior author
  PublisherDOI

• Declining returns of secondary vocational education on subjective class identity across generations and time periods

  Humanities and Social Sciences Communications · 2025-08-06 · 1 citations

  articleOpen access

  This study explores the association of secondary vocational education (SVE) on subjective class identification in China, focusing on comparisons with compulsory and academic high school education, and analyzing regional and historical variations. Using data from the 2013, 2015, 2017, 2018, and 2021 waves of the China General Social Survey (CGSS), this study analyzes 13,736 individuals aged 26 to 58. To address selection bias and better estimate treatment effects, the Generalized Propensity Score Matching (GPSM) method is employed to evaluate the impact of different types of secondary education on subjective class identification. The findings show that individuals with Secondary Vocational Education (SVE) report significantly higher subjective class identification compared to those with only compulsory education, especially in rural areas. While initial comparisons reveal no significant difference between SVE and academic high school graduates, Generalized Propensity Score Matching suggests that academic high school graduates hold higher class identification than SVE graduates. Regionally, education-related class identity gaps are evident in rural but not urban areas. Historically, the positive association between SVE and class identity is significant during 1986–1999, but becomes statistically insignificant during 2000–2010 and 2010-present. SVE significantly enhances subjective class identification compared to compulsory education, particularly in rural areas, with variations across different historical periods. These differences stem from the cyclical development and policy adjustments related to SVE. Therefore, future efforts should focus on strengthening support for SVE, such as expanding opportunities for further education, to maintain the positive association between SVE and subjective class identification.

  PublisherOA PDFDOI

• Next-Gen TOD: Transforming Transit Oriented Development to Embrace New Challenges and Opportunities

  Urban Rail Transit · 2025-04-22 · 10 citations

  articleOpen access1st authorCorresponding

  Abstract Transit-oriented development, as initially conceptualized (TOD 1.0), aims to re-engineer the built environment to promote transit and walking and curb car-oriented sprawl. While widely adopted and practiced, TOD 1.0 faces increasing challenges amid growing societal concerns over inequality and climate crisis. TOD-advocated development densification and diversification could adversely affect neighborhood affordability, residential stability, and microclimate conditions. The emergence of disruptive transportation technologies has also brought both threats and opportunities to transit. This article calls for a paradigm shift, transforming TOD 1.0 to Next-Gen TOD in three aspects. First, Next-Gen TOD shifts the spatial focus from TOD nodes to corridors or networks to harness the network effects of transit systems. Second, Next-Gen TOD shifts the planning/policy focus from prescribing the attributes of 3-Ds (Density, Diversity, and Design) to balancing the outcome of 3-Es (Efficiency, Equity, and Eco-Adaptivity). Third, by including additional 2-Es, Evaluative and Educative, Next-Gen TOD operates as an analytical and engagement platform to support participatory process and informed choice of planning/policy options. The article presents two examples, Hong Kong West Kowloon HSR-TOD and TOD-based affordable housing modeling in Austin, TX, to illustrate the application of Next-Gen TOD.

  PublisherOA PDFDOI

• How Does the US Affect the Lending Decision Preferences of the International Monetary Fund?

  China & World Economy · 2025-07-01

  articleOpen accessSenior authorCorresponding

  Abstract This paper examines the impact of the US on International Monetary Fund (IMF) lending decisions and explores its underlying mechanisms. The results indicate that alignment with the US diplomatic stance significantly increases the likelihood of borrowers securing IMF loans. Countries with closer economic ties to the US not only face a higher probability of obtaining IMF loans but also tend to receive larger loans. Finally, borrowers that engage in military cooperation with the US, or share military interests, are also more likely to access IMF lending. Mechanism analysis reveals that the US has exercised its influence over IMF lending preferences by appointing senior officials to key positions within the organization. However, borrowing countries can partially offset this influence by placing their own citizens in senior management roles at the IMF, thereby mitigating the adverse effects of US influence on their access to IMF loans.

  PublisherOA PDFDOI

• Analysis of the Current Status and Trend of Domestic Eco-city Research based on CiteSpace

  International Journal of Social Sciences and Public Administration · 2024-03-18

  articleOpen access1st authorCorresponding

  In today's era of rapid urbanisation and increasingly prominent ecological and environmental problems, creating a good living environment and building eco-cities to implement sustainable development has become the key to urban development in China. Based on the literature statistics and Citespace data analysis, this paper analyses the research papers on eco-city in CNKI database during 2001-2021, and analyses and summarizes the current situation and trend of eco-city research in China. The study shows that: ① the research and construction of eco-city still has great value. ② The research hotspots of eco-city have five major aspects: ecological civilisation, urbanisation/ urbanisation, indicator system, eco-city construction and sustainable development. ③ The directions of ecological civilisation construction, eco-city construction and sustainable development still occupy a place in the future direction of research and development. ④ The cooperation of scholars in the field of eco-city research is not close enough. ⑤ The field of eco-city research is in the stage of multidisciplinary cross-discipline, covering a wider range of dimensions. The research perspective, research content, and research methods and technical means are all characterised by obvious multidisciplinary intersection.

  PublisherOA PDFDOI

• Evaluating equitable Transit-Oriented development (TOD) via the Node-Place-People model

  Transportation Research Part A Policy and Practice · 2024-06-01 · 35 citations

  articleSenior authorCorresponding
  PublisherDOI

• Can rural counties benefit from high-speed rail investments? The distributive economic impacts of constructing the Dallas-Houston line

  Transportation Research Interdisciplinary Perspectives · 2024-09-01 · 2 citations

  articleOpen accessSenior author

  • The construction of DAL-HOU HSR line in Texas with $15 billion investment will generate $32.27 billion economic outputs across the state. • The economic impacts have uneven distribution. Over 98 percent of the economic outputs will be distributed in the Dallas and Houston metropolitan areas. Only 0.5 percent of the outputs leaked to the rural areas within the Texas Triangle. • The suburbs have better economic connections to the metropolitan areas than rural counties through intermediate purchases for manufacturing sectors. For rural areas, those with agriculture as primary industry receive less economic impact than those with manufacturing as primary industry. • Only counties with terminal stations could have positive receivable ratios from the construction investment, while the rest of the counties along the way generates more indirect and induced effects to other counties, mostly the metropolitan counties, than those received. • To stimulate economic growth in rural counties with HSR investment, local government should provide incentives for private investment in local industries and encourage local purchases. High-speed rail (HSR) investments have demonstrated significant economic benefits in many regions around the world. However, there have been ongoing concerns over the potential unjust effects of HSR development on the rural areas where HSR routes traverse. In the United States, resistance from rural landowners is a major factor that has delayed or derailed numerous HSR projects. Understanding whether and how rural areas may or may not benefit from HSR investments helps facilitate timely project delivery and inform supplementary policymaking to achieve broader societal benefits. This paper presents a case study of the Dallas – Houston HSR line in Texas, USA, a proposed project with private investments totaling $16 billion from Texas Central Railroad, LLC. Applying Multi-Regional Input-Output (MRIO) modeling, the study estimates the economic impacts of investing in the Dallas-Houston HSR line and examines the spatial and sectoral distributions of the impacts across the rural counties along the project alignment. The privately funded Dallas – Houston HSR project stalled after a decade-long effort, largely due to opposition from the rural counties along the line. The recent formation of the Amtrak-Texas Central partnership has brought back hope to revive the HSR proposal. The findings of this study are informative for Amtrak-Texas Central to develop strategies that address the unbalanced rural–urban impacts of constructing the HSR line for ultimate project success.

  PublisherDOI

Frequent coauthors

• Tao Xu

  Beijing Institute of Technology

  13 shared

• Wenshu Li

  Shanghai Maritime University

  6 shared

• Wenjia Zhang

  6 shared

• Tao Feng

  6 shared

• Yuchen Li

  Sichuan University

  6 shared

• Zhengdong Huang

  5 shared

• Ziqi Liu

  Beijing University of Technology

  5 shared

• Liyuan Zhao

  4 shared

Labs

• Community and Regional PlanningPI

Awards & honors

• Roberta P. Crenshaw Centennial Professorship in Urban Design…
• Mike Hogg Professorship in Community and Regional Planning