Knowledge Guided Machine Learning Accelerating Discovery using Scientific Knowledge and Data

Given their tremendous success in commercial applications, machine learning (ML) models are increasingly being considered as alternatives to science-based models in many disciplines. Yet, these "black-box" ML models have found limited success due to their inability to work well in the presence of limited training data and generalize to unseen scenarios. As a result, there is a growing interest in the scientific community on creating a new generation of methods that integrate scientific knowledge in ML frameworks. This emerging field, called scientific knowledge-guided ML (KGML), seeks a distinct departure from existing "data-only" or "scientific knowledge-only" methods to use knowledge and data at an equal footing. Indeed, KGML involves diverse scientific and ML communities, where researchers and practitioners from various backgrounds and application domains are continually adding richness to the problem formulations and research methods in this emerging field. Knowledge Guided Machine Learning: Accelerating Discovery using Scientific Knowledge and Data provides an introduction to this rapidly growing field by discussing some of the common themes of research in KGML using illustrative examples, case studies, and reviews from diverse application domains and research communities as book chapters by leading researchers. KEY FEATURES First-of-its-kind book in an emerging area of research that is gaining widespread attention in the scientific and data science fields Accessible to a broad audience in data science and scientific and engineering fields Provides a coherent organizational structure to the problem formulations and research methods in the emerging field of KGML using illustrative examples from diverse application domains Contains chapters by leading researchers, which illustrate the cutting-edge research trends, opportunities, and challenges in KGML research from multiple perspectives Enables cross-pollination of KGML problem formulations and research methods across disciplines Highlights critical gaps that require further investigation by the broader community of researchers and practitioners to realize the full potential of KGML

About the Editors List of Contributors 1 Introduction Anuj Karpatne, Ramakrishnan Kannan, and Vipin Kumar 2 Targeted Use of Deep Learning for Physics and Engineering Steven L. Brunton and J. Nathan Kutz 3 Combining Theory and Data-Driven Approaches for Epidemic Forecasts Lijing Wang, Aniruddha Adiga, Jiangzhuo Chen, Bryan Lewis, Adam Sadilek, Srinivasan Venkatramanan, and Madhav Marathe 4 Machine Learning and Projection-Based Model Reduction in Hydrology and Geosciences Mojtaba Forghani, Yizhou Qian, Jonghyun Lee, Matthew Farthing, Tyler Hesser, Peter K. Kitanidis, and Eric F. Darve 5 Applications of Physics-Informed Scientific Machine Learning in Subsurface Science: A Survey Alexander Y. Sun, Hongkyu Yoon, Chung-Yan Shih, and Zhi Zhong 6 Adaptive Training Strategies for Physics-Informed Neural Networks Sifan Wang and Paris Perdikaris 7 Modern Deep Learning for Modeling Physical Systems Nicholas Geneva and Nicholas Zabaras 8 Physics-Guided Deep Learning for Spatiotemporal Forecasting Rui Wang, Robin Walters, and Rose Yu 9 Science-Guided Design and Evaluation of Machine Learning Models: A Case-Study on Multi-Phase Flows Nikhil Muralidhar, Jie Bu, Ze Cao, Long He, Naren Ramakrishnan, Danesh Tafti, and Anuj Karpatne 10 Using the Physics of Electron Beam Interactions to Determine Optimal Sampling and Image Reconstruction Strategies for High Resolution STEM Nigel D. Browning, B. Layla Mehdi, Daniel Nicholls, and Andrew Stevens 11 FUNNL: Fast Nonlinear Nonnegative Unmixing for Alternate Energy Systems Jeffrey A. Graves, Thomas F. Blum, Piyush Sao, Miaofang Chi, and Ramakrishnan Kannan 12 Structure Prediction from Scattering Profiles: A Neutron-Scattering Use-Case Cristina Garcia-Cardona, Ramakrishnan Kannan, Travis Johnston, Thomas Proffen, and Sudip K. Seal 13 Physics-Infused Learning: A DNN and GAN Approach Zhibo Zhang, Ryan Nguyen, Souma Chowdhury, and Rahul Rai 14 Combining System Modeling and Machine Learning into Hybrid Ecosystem Modeling Markus Reichstein, Bernhard Ahrens, Basil Kraft, Gustau Camps-Valls, Nuno Carvalhais, Fabian Gans, Pierre Gentine, and Alexander J. Winkler 15 Physics-Guided Neural Networks (PGNN): An Application in Lake Temperature Modeling Arka Daw, Anuj Karpatne, William D. Watkins, Jordan S. Read, and Vipin Kumar 16 Physics-Guided Recurrent Neural Networks for Predicting Lake Water Temperature Xiaowei Jia, Jared D. Willard, Anuj Karpatne, Jordan S. Read, Jacob A. Zwart, Michael Steinbach, and Vipin Kumar 17 Physics-Guided Architecture (PGA) of LSTM Models for Uncertainty Quantification in Lake Temperature Modeling Arka Daw, R. Quinn Thomas, Cayelan C. Carey, Jordan S. Read, Alison P. Appling, and Anuj Karpatne Index

Anuj Karpatne is an Assistant Professor in the Department of Computer Science at Virginia Tech. His research focuses on pushing on the frontiers of knowledge-guided machine learning by combining scientific knowledge and data in the design and learning of machine learning methods to solve scientific and societally relevant problems. Ramakrishnan Kannan is the group leader for Discrete Algorithms at Oak Ridge National Laboratory. His research expertise is in distributed machine learning and graph algorithms on HPC platforms and their application to scientific data with a specific interest for accelerating scientific discovery. Vipin Kumar is a Regents Professor at the University of Minnesota’s Computer Science and Engineering Department. His current major research focus is on knowledge-guided machine learning and its applications to understanding the impact of human induced changes on the Earth and its environment.