Chao Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Chao Wang has a total of 71 co-authors. Here are the top co-authors by number of shared publications:
- Gabriele Villarini
Civil and Environmental Engineering Princeton University Princeton NJ USA
8 shared publications - James M. Done
NSF National Center for Atmospheric Research Boulder CO USA
7 shared publications - Jianguo Wu
Peking University, Beijing, China
6 shared publications Amirhossein Fallahdizcheh
Department of Industrial and Systems Engineering, University of Iowa, Iowa City, IA, USA
5 shared publications- David R. Johnson
School of Industrial Engineering Purdue University West Lafayette IN USA
5 shared publications - Zhiyong Hu
Anhui University
5 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 37 (11) , 6194-6206
Hybrid Transfer Active Learning for Multistream Processes With Within-Process and Cross-Process Correlation Modeling and Online Updating
Active learning for regression (ALR) is a prevalent tool for learning functional relationships by selectively incorporating the most informative data. However, existing ALR methods suffer from the cold-start problem and focus solely on learning single-stream functions or processes. In this work, a hybrid transfer learning framework is proposed to model within-process and cross-process functional correlations for resolving the cold-start problem in multistream ALR. In this framework, a novel multioutput Gaussian process (MGP) covariance structure is proposed to characterize both within-process and cross-process correlations, and offline learning and online updating are strategically integrated to leverage these correlations for ALR. Specifically, offline learning utilizes cross-process correlation to transfer knowledge from related source processes, providing a robust prior for the target process, while online updating incorporates the offline-learned prior with newly collected data to iteratively update the within-process functional relationship in the target process. Theoretical justifications are provided for the proposed framework, showing that the ALR error decreases monotonically as new data are acquired under the guidance of the proposed learning structure. The accuracy and superiority of the proposed framework are thoroughly verified through comparisons with benchmark methods on various numerical and real-case studies.
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 37 (11) , 6194-6206
Hybrid Transfer Active Learning for Multistream Processes With Within-Process and Cross-Process Correlation Modeling and Online Updating
Active learning for regression (ALR) is a prevalent tool for learning functional relationships by selectively incorporating the most informative data. However, existing ALR methods suffer from the cold-start problem and focus solely on learning single-stream functions or processes. In this work, a hybrid transfer learning framework is proposed to model within-process and cross-process functional correlations for resolving the cold-start problem in multistream ALR. In this framework, a novel multioutput Gaussian process (MGP) covariance structure is proposed to characterize both within-process and cross-process correlations, and offline learning and online updating are strategically integrated to leverage these correlations for ALR. Specifically, offline learning utilizes cross-process correlation to transfer knowledge from related source processes, providing a robust prior for the target process, while online updating incorporates the offline-learned prior with newly collected data to iteratively update the within-process functional relationship in the target process. Theoretical justifications are provided for the proposed framework, showing that the ALR error decreases monotonically as new data are acquired under the guidance of the proposed learning structure. The accuracy and superiority of the proposed framework are thoroughly verified through comparisons with benchmark methods on various numerical and real-case studies.
- 23 September 2026
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 278
Uncertainty-aware continual learning for class-added multimodal industrial diagnosis with time-series and textual data
- 23 September 2026
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 278
Uncertainty-aware continual learning for class-added multimodal industrial diagnosis with time-series and textual data
- Open Access
- 1 June 2026
- Published by American Geophysical Union (AGU) in Journal of Advances in Modeling Earth Systems
- Vol. 18 (6)
Physics‐Informed Bias Correction Using a Convolution‐Based Multivariate Gaussian Process
Climate models have been improving to better represent the Earth system. They have become common tools for understanding the dynamics of different physical components of the Earth system and providing insights into their future changes. However, a remaining key limitation is persistent biases in model outputs. Existing bias correction methods only rely on model outputs and do not account for the interconnected physical processes within the climate model that are responsible for the biases in simulating the climate variables (e.g., precipitation and temperature). Here we propose a novel machine learning technique using a convolution‐based multivariate Gaussian process (MGP) to reduce biases and uncertainties in temperature and precipitation from a climate model in a physically consistent manner. This approach captures the physical interactions driven by surface energy fluxes and incorporates their non‐linear relationships with precipitation and temperature, while reducing biases in climate model outputs. Our results indicate that surface energy balance processes act as key physical drivers for quantifying biases and uncertainties in seasonal and regional precipitation and temperature. Consequently, the proposed MGP approach can effectively reduce biases in both temperature and precipitation simultaneously by using their physical relationships to surface energy fluxes.
- 1 June 2026
- Published by American Geophysical Union (AGU) in Journal of Advances in Modeling Earth Systems
- Vol. 18 (6)
Open AccessPhysics‐Informed Bias Correction Using a Convolution‐Based Multivariate Gaussian Process
Climate models have been improving to better represent the Earth system. They have become common tools for understanding the dynamics of different physical components of the Earth system and providing insights into their future changes. However, a remaining key limitation is persistent biases in model outputs. Existing bias correction methods only rely on model outputs and do not account for the interconnected physical processes within the climate model that are responsible for the biases in simulating the climate variables (e.g., precipitation and temperature). Here we propose a novel machine learning technique using a convolution‐based multivariate Gaussian process (MGP) to reduce biases and uncertainties in temperature and precipitation from a climate model in a physically consistent manner. This approach captures the physical interactions driven by surface energy fluxes and incorporates their non‐linear relationships with precipitation and temperature, while reducing biases in climate model outputs. Our results indicate that surface energy balance processes act as key physical drivers for quantifying biases and uncertainties in seasonal and regional precipitation and temperature. Consequently, the proposed MGP approach can effectively reduce biases in both temperature and precipitation simultaneously by using their physical relationships to surface energy fluxes.
- Open Access
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. PP (01628828) , 1-12
A Bayesian Hierarchical Framework for Non-linear and Iterative Transfer Learning of Gaussian Process
Classical multi-output GP (MGP) models often suffer from limitations such as the assumption of linear inter-process correlations, sensitivity to data imbalance, and lack of support for online updates. This paper proposes a novel Bayesian hierarchical framework to address these challenges. Specifically, we introduce a three-layer architecture that models shared hyper-parameters as random variables and updates them asymmetrically through a hierarchical structure. The proposed framework treats source data as historical information to form informative priors and uses target data for posterior update, which facilitates a flexible and process-specific knowledge transfer. We theoretically establish that this treatment induces nonlinear inference for the target process and allows efficient, iterative updates without re-optimizing the entire model. Comprehensive evaluations on synthetic data and real-world case studies in battery diagnostics and nano-sensor design demonstrate that our method significantly outperforms existing single-output and hierarchical MGP baselines in both accuracy and computational efficiency. The results suggest that our approach offers a principled and scalable solution for transfer learning under data-scarce and heterogeneous conditions.
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. PP (01628828) , 1-12
Open AccessA Bayesian Hierarchical Framework for Non-linear and Iterative Transfer Learning of Gaussian Process
Classical multi-output GP (MGP) models often suffer from limitations such as the assumption of linear inter-process correlations, sensitivity to data imbalance, and lack of support for online updates. This paper proposes a novel Bayesian hierarchical framework to address these challenges. Specifically, we introduce a three-layer architecture that models shared hyper-parameters as random variables and updates them asymmetrically through a hierarchical structure. The proposed framework treats source data as historical information to form informative priors and uses target data for posterior update, which facilitates a flexible and process-specific knowledge transfer. We theoretically establish that this treatment induces nonlinear inference for the target process and allows efficient, iterative updates without re-optimizing the entire model. Comprehensive evaluations on synthetic data and real-world case studies in battery diagnostics and nano-sensor design demonstrate that our method significantly outperforms existing single-output and hierarchical MGP baselines in both accuracy and computational efficiency. The results suggest that our approach offers a principled and scalable solution for transfer learning under data-scarce and heterogeneous conditions.
- Open Access
- 19 May 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Local drivers in accelerating North American heat stress
Climate change increases heat extremes, threatening human health and economies. Using reanalysis data and climate simulations, we show that since the 1940s, population exposure to extreme heat (wet-bulb globe temperature > 32 °C) has increased by 21% in the U.S. At 2 °C of global warming, exposure increases by 273% because heat-stress frequency increases exponentially with warming. Additionally, 2 °C warming leads to increased nighttime heat stress and decreased work capacity, indicating severe health and economic impacts. Heat stress rises fastest in high-latitude areas, while humid regions experience the greatest exposure increases. In northern regions, heatwave frequency increases with warming, whereas in southern regions, events merge into month-long heatwaves. Increasing temperatures and humidity, along with decreasing wind speed, influence regional heat stress, underscoring the need for tailored adaptation strategies. Overall, heat stress exposure is projected to escalate with additional warming, underscoring the need for mitigation and adaptation to protect vulnerable populations.
- 19 May 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessLocal drivers in accelerating North American heat stress
Climate change increases heat extremes, threatening human health and economies. Using reanalysis data and climate simulations, we show that since the 1940s, population exposure to extreme heat (wet-bulb globe temperature > 32 °C) has increased by 21% in the U.S. At 2 °C of global warming, exposure increases by 273% because heat-stress frequency increases exponentially with warming. Additionally, 2 °C warming leads to increased nighttime heat stress and decreased work capacity, indicating severe health and economic impacts. Heat stress rises fastest in high-latitude areas, while humid regions experience the greatest exposure increases. In northern regions, heatwave frequency increases with warming, whereas in southern regions, events merge into month-long heatwaves. Increasing temperatures and humidity, along with decreasing wind speed, influence regional heat stress, underscoring the need for tailored adaptation strategies. Overall, heat stress exposure is projected to escalate with additional warming, underscoring the need for mitigation and adaptation to protect vulnerable populations.
- 9 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (1) , 527-541
Transfer Learning of Stochastic Kriging for Individualized Prediction
Stochastic Kriging (SK) is a generalized variant of Gaussian process regression, and it is developed for dealing with non-i.i.d. noise in functional responses. Although SK has achieved substantial success in various engineering applications, its intrinsic modeling strategy by focusing on the sample mean limits its flexibility and capability of predicting individual functional samples. Moreover, the performance of SK can be impaired under scarce data scenarios, which are commonly encountered in engineering applications, especially for start-up or just deployed systems. In this paper, we propose a novel transfer learning framework to address the challenges of individualization and data scarcity in traditional SK. The proposed framework features a within-process model to facilitate individualized prediction and a between-process model to leverage information from related processes for resolving the issue of data scarcity. The within- and between-process models are integrated through a tailored convolution process, which quantifies interactions within and between processes using a specially designed covariance matrix and corresponding kernel parameters. Statistical properties are investigated on the parameter estimation of the proposed framework, which provide theoretical guarantees for the performance of transfer learning. The proposed method is compared with benchmark methods through various numerical and real case studies, and the results demonstrate the superiority of the proposed method in dealing with individualized prediction of functional responses, especially when limited data are available in the process of interest. The reproducibility code is available in the supplementary materials.
- 9 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (1) , 527-541
Transfer Learning of Stochastic Kriging for Individualized Prediction
Stochastic Kriging (SK) is a generalized variant of Gaussian process regression, and it is developed for dealing with non-i.i.d. noise in functional responses. Although SK has achieved substantial success in various engineering applications, its intrinsic modeling strategy by focusing on the sample mean limits its flexibility and capability of predicting individual functional samples. Moreover, the performance of SK can be impaired under scarce data scenarios, which are commonly encountered in engineering applications, especially for start-up or just deployed systems. In this paper, we propose a novel transfer learning framework to address the challenges of individualization and data scarcity in traditional SK. The proposed framework features a within-process model to facilitate individualized prediction and a between-process model to leverage information from related processes for resolving the issue of data scarcity. The within- and between-process models are integrated through a tailored convolution process, which quantifies interactions within and between processes using a specially designed covariance matrix and corresponding kernel parameters. Statistical properties are investigated on the parameter estimation of the proposed framework, which provide theoretical guarantees for the performance of transfer learning. The proposed method is compared with benchmark methods through various numerical and real case studies, and the results demonstrate the superiority of the proposed method in dealing with individualized prediction of functional responses, especially when limited data are available in the process of interest. The reproducibility code is available in the supplementary materials.
- Open Access
- 1 July 2025
- Published by Springer Nature in Scientific Data
- Vol. 12 (1) , 1-12
Ensemble downscaled climate dataset for Alaska and Hawaii under historical and future conditions
While high-resolution future climate data are increasingly available for the contiguous United States, there has been limited focus on Alaska and Hawaii. Our study provides high resolution daily climate data —precipitation and temperature —at 10 km for Alaska and 1 km for Hawaii, based on 23 climate models from the Coupled Model Intercomparison Project Phase 6 and four scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5). The dataset includes outputs from different bias correction techniques (seven for precipitation, six for temperature). While one method (i.e., empirical quantile mapping) generally outperforms the others, its performance can vary by climate quantity, region, or climate model, motivating our aim to provide an ensemble dataset of bias-corrected outputs. Our results show that, on average, Alaska and Hawaii are projected to become warmer (6.7 °C and 5.8 °C) and wetter (46% and 31%) by the end of the 21st century for high-emissions scenarios compared to the historical period, despite large regional variability. Our dataset aims to support climate research, mitigation and adaptation strategies, and risk assessments for Alaska and Hawaii.
- 1 July 2025
- Published by Springer Nature in Scientific Data
- Vol. 12 (1) , 1-12
Open AccessEnsemble downscaled climate dataset for Alaska and Hawaii under historical and future conditions
While high-resolution future climate data are increasingly available for the contiguous United States, there has been limited focus on Alaska and Hawaii. Our study provides high resolution daily climate data —precipitation and temperature —at 10 km for Alaska and 1 km for Hawaii, based on 23 climate models from the Coupled Model Intercomparison Project Phase 6 and four scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5). The dataset includes outputs from different bias correction techniques (seven for precipitation, six for temperature). While one method (i.e., empirical quantile mapping) generally outperforms the others, its performance can vary by climate quantity, region, or climate model, motivating our aim to provide an ensemble dataset of bias-corrected outputs. Our results show that, on average, Alaska and Hawaii are projected to become warmer (6.7 °C and 5.8 °C) and wetter (46% and 31%) by the end of the 21st century for high-emissions scenarios compared to the historical period, despite large regional variability. Our dataset aims to support climate research, mitigation and adaptation strategies, and risk assessments for Alaska and Hawaii.
- Open Access
- 25 April 2025
- Published by Springer Nature in Research Square
Local Drivers in Accelerating North American Heat Stress
Climate change increases the frequency and intensity of heat extremes, threatening human health and economies1–4. We use reanalysis data and regional climate simulations with a 4- km grid5 to calculate the wet-bulb globe temperature (WBGT)6, a key index for assessing heat stress7, 8. Our analysis shows that historical climate change has increased population exposure to extreme heat (WBGT >32,◦ 17 C) by 21% across the Contiguous United States. However, as global warming reaches 2,◦C, exposure surges by 273% due to the exponential rise in heat stress frequency with temperature. Annual maximum WBGT increases fastest in high-latitude areas, while BFH frequency increases most in humid subtropical regions like 1the Gulf Coast and the Caribbean. In northern regions, heatwave frequency increases with warming while in southern regions individual events merge into mega-heatwaves lasting over a month under 2 ◦C warming. Drivers of WBGT changes differ by region, with temperature and longwave radiation dominating maintain and high-latitude land areas. Humidity plays a crucial role in humid subtropical regions, while decreasing wind speeds impact the Midwest and Northeast. These regional differences highlight the need for tailored adaptation strate gies. Our findings stress the urgent need for mitigation and adaptation strategies to protect vulnerable populations from rising heat-related risks9–12.
- 25 April 2025
- Published by Springer Nature in Research Square
Open AccessLocal Drivers in Accelerating North American Heat Stress
Climate change increases the frequency and intensity of heat extremes, threatening human health and economies1–4. We use reanalysis data and regional climate simulations with a 4- km grid5 to calculate the wet-bulb globe temperature (WBGT)6, a key index for assessing heat stress7, 8. Our analysis shows that historical climate change has increased population exposure to extreme heat (WBGT >32,◦ 17 C) by 21% across the Contiguous United States. However, as global warming reaches 2,◦C, exposure surges by 273% due to the exponential rise in heat stress frequency with temperature. Annual maximum WBGT increases fastest in high-latitude areas, while BFH frequency increases most in humid subtropical regions like 1the Gulf Coast and the Caribbean. In northern regions, heatwave frequency increases with warming while in southern regions individual events merge into mega-heatwaves lasting over a month under 2 ◦C warming. Drivers of WBGT changes differ by region, with temperature and longwave radiation dominating maintain and high-latitude land areas. Humidity plays a crucial role in humid subtropical regions, while decreasing wind speeds impact the Midwest and Northeast. These regional differences highlight the need for tailored adaptation strate gies. Our findings stress the urgent need for mitigation and adaptation strategies to protect vulnerable populations from rising heat-related risks9–12.
- Open Access
- 11 March 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1)
Climate change reduces the wind chill hazard across Alaska
- 11 March 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1)
Open AccessClimate change reduces the wind chill hazard across Alaska
- Open Access
- 9 February 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (3)
Precipitation‐ and Temperature‐Driven Future Changes to Flooding in Alaska
Examining large‐scale projected changes in streamflow and flood extent (e.g., inundation) for Alaska is essential for raising awareness of flood hazards under a changing climate and supporting broad‐scale adaptation planning. Therefore, we examine projected changes in peak streamflow timing and magnitude using a physically based hydrologic model. For model inputs, we utilize climate simulations conducted at 4‐km horizontal grid spacing over Alaska from 2005 to 2016, providing a historical and future pseudo‐global warming scenario. Analysis of hydrographs reveals the peak timing shifts slightly earlier in the year for most of Alaska's streams. The change in peak magnitude is more heterogeneous across the state, with the northernmost region showing the highest projected increases. The changes in timing are driven by temperature, while precipitation and temperature drive the changes in magnitude. These changes are then transformed into inundation maps, showing a similar albeit more muted pattern compared to the changes in magnitude.
- 9 February 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (3)
Open AccessPrecipitation‐ and Temperature‐Driven Future Changes to Flooding in Alaska
Examining large‐scale projected changes in streamflow and flood extent (e.g., inundation) for Alaska is essential for raising awareness of flood hazards under a changing climate and supporting broad‐scale adaptation planning. Therefore, we examine projected changes in peak streamflow timing and magnitude using a physically based hydrologic model. For model inputs, we utilize climate simulations conducted at 4‐km horizontal grid spacing over Alaska from 2005 to 2016, providing a historical and future pseudo‐global warming scenario. Analysis of hydrographs reveals the peak timing shifts slightly earlier in the year for most of Alaska's streams. The change in peak magnitude is more heterogeneous across the state, with the northernmost region showing the highest projected increases. The changes in timing are driven by temperature, while precipitation and temperature drive the changes in magnitude. These changes are then transformed into inundation maps, showing a similar albeit more muted pattern compared to the changes in magnitude.