 |
Wei W. Xing |
Latest Updates:
Best Paper Award ISEDA 2024!
Five papers get accepted by DAC 2024!
Two papers on physics law discovery using AI get accepted by AISTAT 2024!
Our work on multi-fidelity fusion was funded by BBSRC (UK). We will explore the use of multi-fidelity fusion in bio Biophysical modeling, which can help disentangle the contributions of different biophysical processes, such as cell-cell and cell-matrix adhesion and cortical contractility, to epithelial tissue structure.
Our work on multi-fidelity fusion was funded by Sheffield grant, AIRE. We will explore the use of multi-fidelity fusion in the automatic driving system and develop a public library for multi-fidelity fusion.
Our work on urban digital twin model gets Beijing Science and Technology Progress Award (北京市科学技术进步奖二等奖)
Best Paper Nomination ICCAD 2023: the first transfer learning method for yield estimation and optimization for SRAM circuits
Looking for Motivated Students:
I am always looking for motivated undergraduate and graduate students with background in computer science, microelectronics, electrical engineering, mathematics, statistics, or related areas, working on research fields of 1) general AI methods for engineering problems, 2) AI4EDA, and 3) AI4Science. If you are interested, please drop me an email with your CV.
Opening positions available: Ph.D., Post Doc, Engineer, Interns.
Research
We develop artificial intelligence solutions to transform computational engineering by addressing three critical challenges: making optimal decisions in complex design spaces, creating accurate models of physical systems, and performing efficient simulations. Our innovative AI approaches enhance design optimization, enable reliable surrogate modeling, and accelerate simulations while maintaining engineering reliability. Through this research in electronic design automation and energy systems, we're working toward a future where AI acts as an intelligent partner in engineering design, enhancing rather than replacing traditional methods.
Our Research: Intelligent Decision Making and Efficient Simulations
Modern engineering relies heavily on computational methods to design and optimize complex systems. From semiconductors to energy systems, engineers use mathematical modeling, numerical simulation, and optimization to develop new technologies. However, these traditional approaches face significant challenges. For instance, designing a modern integrated circuit involves optimizing thousands of parameters while ensuring reliability across different operating conditions - a task that becomes computationally intractable with conventional methods. Our work includes:
AI Copilot for Design and Optimization
Design optimization is the cornerstone of computational engineering, where engineers must find optimal solutions in complex, high-dimensional spaces. Traditional optimization methods often require numerous expensive evaluations and struggle with competing objectives. Our research develops AI methods to accelerate this process and enhance human decision-making.
We pioneer multi-fidelity Bayesian optimization approaches that intelligently combine quick approximate evaluations with selective high-fidelity simulations. For instance, in circuit design, our methods reduced optimization time from weeks to hours while maintaining reliability. Our yield optimization techniques have significantly improved manufacturing outcomes by efficiently identifying and mitigating failure modes. Through knowledge transfer across designs, we leverage past experiences to accelerate new optimization tasks, demonstrating particular success in transistor sizing and thermal design problems.
Reliable Surrogate Models and Digital Twins
Complex engineering systems require accurate models for design and analysis, but high-fidelity simulations are often too computationally expensive for practical optimization. We address this challenge by developing reliable surrogate models that maintain physical consistency while dramatically reducing computational cost.
Our research advances physics-informed neural networks and novel multi-fidelity fusion techniques (ContinuAR and GAR) that combine data from different fidelity levels to create accurate, fast-evaluating models. We've developed scalable Gaussian processes that can handle large-scale engineering data while providing crucial uncertainty quantification. These methods have enabled real-time performance prediction for complex systems while maintaining accuracy within 1% of high-fidelity simulations.
AI for Efficient and Accurate Simulations
Numerical simulation is essential for predicting system behavior, but traditional methods often require intensive computational resources. We develop AI-enhanced simulation methods that maintain accuracy while significantly reducing computational cost.
Our physics-informed neural networks have achieved up to 50x acceleration in SPICE circuit simulation and thermal analysis. Through smart model calibration techniques, we ensure these accelerated simulations remain reliable for engineering applications. Our methods adapt simulation strategies based on system behavior, enabling real-time analysis that was previously impossible.
Impact and Applications
Our methods have demonstrated significant impact in electronic design automation (EDA) and energy systems. In semiconductor design, our optimization and yield enhancement techniques are helping partners improve manufacturing reliability and reduce design time. For energy systems, our surrogate modeling and simulation acceleration methods have enabled faster battery and fuel cell design optimization. We're currently working with leading companies in both sectors to deploy these technologies in real-world applications.
Group: IceLab (Intelligent Computational Engineering Lab)
Undergraduate students:
2021: Shuo Yin, Shixiang Yan, Yuxin Wang, Yuhan Bin, Guohao Dai
2022: Yuhua Zhang
2023: Weijian, Zhuohua, Longze
2024: Weijian, Zhuohua
Master students:
2021: Xiang Jin, Yinpeng Wu, Shihong Wang, Yichen Meng, Jiazheng Niu, Linchan Luo, Yuwen Deng
2022: Yanfan Liu, Mingyue Wang, Zhaoyu Shi, Fei Li
2023: Yuan Yao,
2024: Yuan Yao, Yuxua Liu
Ph.D. students:
2021: Yi Gu (coadvised with Prof. Chao Han), Shuyang Dai (coadvised with Prof. Akeel A. Shah)
2022: Wei Jiang
2024: Jin Kou, Haiyan Qin, Yanbo Wei, Junzhuo Zhou
Postdoctoral fellows:
2024: Min Tao
Teaching
Mathematical Investigation Skills (MAS1163, 2023,2024)
Machine learning (MAS369/MAS61007, 2023,2024)
Numerical method for engineering and machine learning (BH41112101, Fall 2020)
Machine learning and its application to engineering (Spring 2022)
All Publications
Conference Paper
Wei W. Xing, Rongqi Lu, Zheng Xing, zhelong wang, ning xu, Weisheng Zhao and yuanqing Chen, “TOTAL: Multi-Corners Timing Optimization Based on Transfer and Active Learning”, DAC2023
Yanfang Liu, Guohao Dai and Wei W. Xing*, Seeking the Yield Barrier: High-Dimensional SRAM Evaluation Through Optimal Manifold, DAC2023
Y. Wang, Z. Xing, and W. W. Xing*, “GAR: Generalized Autoregression for Multi-Fidelity Fusion,” NeurIPS 2022.
S. Yin, G. Dai, and W. W. Xing*, “High Dimensional Yield Estimation using Shrinkage Deep Features and Maximization of Approximated Entropy Reduction,” ASP-DAC 2022.
Zheng Wang, Wei Xing, Robert M. Kirby, and Shandian Zhe, “Physics Informed Deep Kernel Learning”, The 25th International Conference on Artificial Intelligence and Statistics (AISTATS), 2022
Shihong Wang, Xueyin Zhang, and Wei W. Xing*, “E-LMC: Extended Linear Model of Coregionalization for Predictions of Spatial Fields”, The 2022 International Joint Conference on Neural Networks (IJCNN) 2022 (to appear)
Shuo Yin, Xiang Jin, Linxu Shi, Kang Wang and Wei W. Xing*, “Efficient Bayesian Yield Analysis and Optimization with Active Learning”, The 59th Design Automation Conference (DAC), 2022
Zhou Jin, Haojie Pei, Yichao Dong, Xiang Jin, Xiao Wu, Wei W. Xing* and Dan Niu*, “Accelerating DC Circuit Simulation with Reinforcement Learning”, The 59th Design Automation Conference (DAC), 2022
Z. Wang, W. Xing, R. Kirby, and S. Zhe, “Multi-Fidelity High-Order Gaussian Processes for Physical Simulation,” in International Conference on Artificial Intelligence and Statistics (AISTAT), Mar. 2021, pp. 847–855 [PDF]
S. Li, W. Xing, R. M. Kirby, and S. Zhe, “Scalable Gaussian Process Regression Networks,” International Joint Conference on Artificial Intelligence (IJCAI), Jul. 2020, vol. 3, pp. 2456–2462 [PDF]
W. Xing, S. Elhabian, R. Kirby, R. T. Whitaker, and S. Zhe, “Infinite ShapeOdds: Nonparametric Bayesian Models for Shape Representations,” Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), vol. 34, no. 04, pp. 6462–6469, Apr. 2020 [PDF]
S. Li, W. Xing, R. Kirby, and S. Zhe, “Multi-Fidelity Bayesian Optimization via Deep Neural Networks,” Advances in Neural Information Processing Systems (NeurIPS), vol. 33, pp. 8521–8531, 2020 [PDF]
S. Zhe, W. Xing, and R. M. Kirby, “Scalable High-Order Gaussian Process Regression,” in The 22nd International Conference on Artificial Intelligence and Statistics (AISTAT), Apr. 2019, pp. 2611–2620 [PDF]
J. Wang#, W. Xing#, R. M. Kirby, and M. Zhang, “Data-Driven Model Order Reduction for Diffeomorphic Image Registration,” in Information Processing in Medical Imaging (IPMI), vol. 11492 [PDF]
W. Xing, A. A. Shah, B. Urasinska-Wojcik, and J. W. Gardner, “Prediction of impurities in hydrogen fuel supplies using a thermally-modulated CMOS gas sensor: Experiments and modelling,” in 2017 IEEE SENSORS, Glasgow, Oct. 2017, pp. 1–3, [Link]
Moore J., Xing W., Wiese J., Dailey M. Becnel T., Goffin P., J., Gaillardon, Kelly K, Butterfield T., Engaging Pre-College Students in Hypothesis Generation using a Citizen Scientist Network of Air Quality Sensors. ASEE Annual Conference and exposition
K. Kelly, W. Xing, P. Goffin, T. Sayahi, T. Becnel, P-E. Gaillardon, A.E. Butterfield, M. Meyer, R. Whitaker, Understanding how pollution episodes affect community-level air quality with a distributed sensor network, AIChE Annual Meeting, Orlando, FL, November 10-15, 2019.
J. Moore, W. Xing, M. Dailey, K. Le, T. Becnel, P. Goffin, M. Meyer, P-E Gaillardon, R.Whitaker, J. Wiese, A.E. Butterfield, K.E. Kelly, Engaging middle and high school students in hypothesis generation using a citizen-scientist network of air quality sensors, AIChE Annual Meeting, Orlando, FL, November 10-15, 2019.
T. Sayahi, P.-E. Gaillardon, R. Whitaker, M. Meyer, T. Butterfield, P. Goffin, T. Becnel, A. Biglari, D. Kaufman, T. Sayahi, W. Xing, K. Kelly, Platform: Long-term evaluation of the plantower PMS sensor. 10th International Aerosol Conference, September 2nd-7th, St. Louis, Mo, 2018.
K.E. Kelly, P.-E. Gaillardon, R. Whitaker, M. Meyer, T. Butterfield, P. Goffin, T. Becnel, A. Biglari, D. Kaufman, T. Sayahi, W. Xing, Poster: A layered framework for integrating low-cost sensor data and for engaging citizens to understand PM2.5 exposure. 10th International Aerosol Conference, September 2nd-7th, St. Louis, Mo, 2018.
Triantafyllidis, V., Xing, W., Shah, A. A., & Nair, P. B. (2016). Neural network emulation of spatio-temporal data using linear and nonlinear dimensionality reduction. In Advanced Computer and Communication Engineering Technology (pp. 1015-1029). Springer, Cham.
Journal Paper
W. Xing et al., “Data-Driven Prediction of Li-Ion Battery Degradation Using Predicted Features,” Processes, vol. 11, no. 3, Art. no. 3, Mar. 2023, doi: 10.3390/pr11030678.
W. W. Xing, S. Dai, A. A. Shah, L. Luo, Q. Xu, and P. K. Leung, “Emulating Spatial and Temporal Outputs From Fuel Cell and Battery Models: A Comparison of Deep Learning and Gaussian Process Models,” Journal of Electrochemical Energy Conversion and Storage, vol. 20, no. 1, May 2022, doi: 10.1115/1.4054195.
W. W. Xing, X. Jin, T. Feng, D. Niu, W. Zhao, and Z. Jin, “BoA-PTA: A Bayesian Optimization Accelerated PTA Solver for SPICE Simulation,” ACM Trans. Des. Autom. Electron. Syst., vol. 28, no. 2, p. 27:1-27:26, Dec. 2022, doi: 10.1145/3555805.
Shah, A. A., Yu, F., Xing, W. W., & Leung, P. K. (2022). Machine learning for predicting fuel cell and battery polarisation and charge–discharge curves. Energy Reports, 8, 4811-4821.
Xing, W. W., et al. “Emulating Spatial and Temporal Outputs From Fuel Cell and Battery Models: A Comparison of Deep Learning and Gaussian Process Models.” Journal of Electrochemical Energy Conversion and Storage 20.1 (2023): 011007.
Y. Gu, C. Han, Y. Chen, and W. W. Xing, “Mission Replanning for Multiple Agile Earth Observation Satellites Based on Cloud Coverage Forecasting,” IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing, vol. 15, pp. 594–608, 2022, [Link]
W. W. Xing, M. Cheng, K. Cheng, W. Zhang, and P. Wang, “InfPolyn, a Nonparametric Bayesian Characterization for Composition-Dependent Interdiffusion Coefficients,” Materials, vol. 14, no. 13, p. 3635, Jun. 2021 [Link]
H. Wang, C. Li, W. Xing, Y. Ye, and P. Wang, “A machine learning approach to quantify dissolution kinetics of porous media,” Journal of Machine Learning for Modeling and Computing, 2.2, 2021 [Link]
W. I. Ibrahim, M. R. Mohamed, R. M. T. R. Ismail, P. K. Leung, W. W. Xing, and A. A. Shah, “Hydrokinetic energy harnessing technologies: A review,” Energy Reports, vol. 7, pp. 2021–2042, Nov. 2021 [Link]
W. W. Xing, A. A. Shah, P. Wang, S. Zhe, Q. Fu, and R. M. Kirby, “Residual Gaussian process: A tractable nonparametric Bayesian emulator for multi-fidelity simulations,” Applied Mathematical Modelling, vol. 97, pp. 36–56, Sep. 2021
W. W. Xing, R. M. Kirby, and S. Zhe, “Deep coregionalization for the emulation of simulation-based spatial-temporal fields,” Journal of Computational Physics, vol. 428, p. 109984, Mar. 2021
Kerry E. Kelly#, Wei W. Xing#, Tofigh Sayahi, Logan Mitchell, Tom Becnel, Pierre-Emmanuel Gaillardon, Miriah Meyer, and Ross T. Whitaker., “Community-Based Measurements Reveal Unseen Differences during Air Pollution Episodes,” Environmental Science & Technology, vol. 55, no. 1, pp. 120–128, Jan. 2021
W. W. Xing, F. Yu, P. K. Leung, X. Li, P. Wang, and A. A. Shah, “A new multi-task learning framework for fuel cell model outputs in high-dimensional spaces,” Journal of Power Sources, vol. 482, p. 228930, Jan. 2021
W. Xing, S. Y. Elhabian, V. Keshavarzzadeh, and R. M. Kirby, “Shared-Gaussian Process: Learning Interpretable Shared Hidden Structure Across Data Spaces for Design Space Analysis and Exploration,” Journal of Mechanical Design, vol. 142, no. 8, Aug. 2020
W. Xing, M. Razi, R. M. Kirby, K. Sun, and A. A. Shah, “Greedy nonlinear autoregression for multifidelity computer models at different scales,” Energy and AI, vol. 1, p. 100012, Aug. 2020
C. Mullen, T. Collins, W. Xing, R. Whitaker, T. Sayahi, T. Becnel, P. Goffin, P. Gaillardon, M. Meyer, K. Kelly, “Patterns of distributive environmental inequity under different PM2.5 air pollution scenarios for Salt Lake County public schools,” Environmental Research, vol. 186, p. 109543, Jul. 2020
D.V. Mallia, A.K. Kochanski, K.E. Kelly, R. Whitaker, W. Xing, L.E. Mitchell, A. Jacques, A. Farguell, J. Mandel, P.E. Gaillardon, and , T. Becnel. “Evaluating Wildfire Smoke Transport Within a Coupled Fire-Atmosphere Model Using a High-Density Observation Network for an Episodic Smoke Event Along Utah’s Wasatch Front,” Journal of Geophysical Research: Atmospheres, vol. 125, no. 20, p. e2020JD032712, 2020
C. Gadd, W. Xing, M. M. Nezhad, and A. A. Shah, “A Surrogate Modelling Approach Based on Nonlinear Dimension Reduction for Uncertainty Quantification in Groundwater Flow Models,” Transport in Porous Media, vol. 126, no. 1, pp. 39–77, Jan. 2019
V. Triantafyllidiis, W. W. Xing, P. K. Leung, A. Rodchanarowan, and A. A. Shah, “Probabilistic sensitivity analysis for multivariate model outputs with applications to Li-ion batteries,” Journal of Physics: Conference Series, vol. 1039, p. 012020, Jun. 2018
A. A. Shah, W. W. Xing, and V. Triantafyllidis, “Reduced-order modelling of parameter-dependent, linear and nonlinear dynamic partial differential equation models,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 473, no. 2200, p. 20160809, Apr. 2017
W. W. Xing, V. Triantafyllidis, A. A. Shah, P. B. Nair, and N. Zabaras, “Manifold learning for the emulation of spatial fields from computational models,” Journal of Computational Physics, vol. 326, pp. 666–690, Dec. 2016
W. Xing, A. A. Shah, and P. B. Nair, “Reduced dimensional Gaussian process emulators of parametrized partial differential equations based on Isomap,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 471, no. 2174, p. 20140697, Feb. 2015
Award
ICCAD2023 Best paper nomination (2023)
Beijing Science and Technology Progress Award (北京市科学技术进步奖二等奖) (2023)
EDA elite championship competition second prize (2022)
Excellent Young Scholars of Beihang University (2021)
National university innovation and entrepreneurship competition Third prize (2021)
National university innovation and entrepreneurship competition first prize of North China (2021)
EDA elite championship competition first prize (2020)
EDA elite championship competition second prize (2020)
Best poster award in IPMI2019 (2019)
PhD scholarship University of Warwick (2012)
Outstanding graduate award of Shenzhen University (2012)
Scholarship of academician Yan Shuzi (Top academic award in the University) (2012)
Top award, first prize, third prize and special prize in the 2011 China Electronic Innovation Design Competition (2011)
First prize in the innovation projects of 2011 China Educational Robots Competition (2011)
Commercial scholarship “Haorizi Scholarship” (first class award)(2011)
Commercial scholarship “Metro Scholarship” (first class award)(2011)
Academic Innovation Award(2011)
Academic Scholarship of First Class of Shenzhen University (2009,2008)
Excellent Academic Students of Shenzhen University award (2008,2009,2010,2011)
Codes
FidelityFusion: Tractiable Multi-Fidelity Fusion based on Gaussian process all in one package
Mini-GP: Pytorch-based, pytorch NN like, super easy to use Gaussian process package, designed for beginners and developers (with easy modification)!
FD solver for Richard equation with porou media random field
GreedyNAR: Greedy Nonlinear Autoregression
SharedGP: Shared Gaussian Process Learning Interpretable Shared Hidden Structure Across Data Spaces for Design Space Analysis
InfPolyn: Interdiffusion Coefficient Extraction using physics informed nonparametric Bayesian method
Tutorials
Services
Journal Reviewer
Computer Method in Applied Mechanics and Engineering
Journal of Computational Science
Journal of Machine learning for modeling and computing
Journal of Computing and Information Science in Engineering
Computer and Graphics
Computer Physics Communication
Transport in Porous Media
Fluid
Electronics
Mathematics
Biology
Remotesensing
Sensors
Conference Reviewer
AISTAT (2021,2022,2023)
ICMLIP (2020) program committee