Latent Variable Gaussian Process (LVGP) Source Code for Mixed Variable Machine Learning

NU 2021-247

Inventors

Wei Chen, Daniel Apley, Suraj Yerramilli, Ramin Baghgar Bostanabad (University of California, Irvine)

 

Short Description

LVGP-PyTorch is a cross-platform Python implementation of the Latent Variable Gaussian Process (LVGP) methodology for creating machine learning models with either qualitative or quantitative inputs or mixture of both.  The code provides prediction as well as uncertainty quantification that can be easily integrated with other adaptive learning methods, such as Bayesian Optimization (BO) for decision making or design optimization. The library is a cross-platform library and works on Windows, Mac, and Linux. There are no specific hardware requirements.

 

Purchase Options

Following purchase code will be delivered through FTP transfer. Further instructions regarding the receipt of code will be sent within one to two weeks. Please ensure that the correct contact email is provided during checkout.

Price: $27,500.00 USD

 

Citations

Zhang, Y., Tao, S., Chen, W., and Apley, D., “A Latent Variable Approach to Gaussian Process Modeling with Qualitative and Quantitative Factors”, Technometrics, DOI: 10.1080/00401706.2019.1638834, 2019.

 

Zhang, Y., Apley, D., and Chen, W., “Bayesian Optimization for Materials Design with Mixed Quantitative and Qualitative Variables”, Scientific Report, 10, Article number: 4924, 2020.


For technical questions, please contact:

Professor Wei Chen: weichen@northwestern.edu

Professor Daniel Apley: apley@northwestern.edu

 

Please note that software code is provided "as is" and all uses must comply with the grant, obligations and limitations included in the click license agreement embedded within the checkout process. All click license terms must be accepted to compete this purchase.

Patent Information: