Joel Paulson receives NSF CAREER Award
Assistant Professor Joel Paulson, who joined the William G. Lowrie Department of Chemical and Biomolecular Engineering at The Ohio State University in 2019, has received the prestigious National Science Foundation CAREER Award. The award, much coveted by junior faculty, recognizes faculty who exemplify the role of teacher-scholars by delivering outstanding research and excellent education. Recipients are deemed to have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organizations.
Paulson's project, titled "Advancing efficient global optimization of extremely expensive functions under uncertainty using structure-exploiting Bayesian methods" begins a funding cycle on February 1, 2023 until January 31, 2028 with support totalling $517,535.
Paulson's research focuses on control theory and algorithms, optimization of complex systems under uncertainty, and sustainable and smart manufacturing. Essentially, he uses computational methods to ascertain the optimal quality, efficiency, and sustainability of engineered products and processes through the development of advanced decision-making strategies in the presence of uncertainty. These strategies are formulated in terms of stochastic (random) mathematical optimization problems that can be applied to a broad range of applications.
"Optimization problems arise in virtually all human endeavors and can be used to model decisions as simple as choosing what food to eat for dinner, or as complex as designing and manufacturing life-saving medicines," Paulson explained.
"My group is developing new efficient computational methods/algorithms for solving 'very hard' optimization problems that are not solvable with existing approaches," he continued. "The expression 'very hard' in this case can refer to many different things, including lack of knowledge about the structure of the model equations, limited amounts of data from heterogenous sources, or large-scale problems with many degrees of freedom. By translating the 'optimization task' into an 'iterative learning' task, we are able to explicitly design new methods that identify/discover optimal solutions in a very resource-efficient manner, which opens the door to a new class of problem-solving that has been previously thought to be too hard to solve directly," he said.
Paulson's research is important because the ability to take advantage of our resources in the most efficient manner is the key to ensuring that society at large can continue to prosper and move towards a sustainable future. His work addresses many of today's important and socially-relevant problems--problems that fall into this class of "very hard." For example, the design of drugs and sustainable materials requires expensive experiments to characterize quality; auto-tuning of of advanced controllers in robotics and smart manufacturing necessitates collecting data from real systems to evaluate performance; and inverse optimization problems include learning unknown decision-making models from data. In all of these cases, it is very difficult to build a complete and accurate “digital twin” (computational model) of the full system.
"How do we compare or define 'good' versus 'bad' decisions, especially when there are large amounts of uncertainty in the impact of these decisions?" Joel Paulson
The new methods Paulson plans to develop as a part of the CAREER-funded research will enable practitioners to much more efficiently design their experiments and/or simulations so that they can quickly discover the conditions that lead to optimal behavior—using any metric they want to use to quantify performance.
"I have always been interested in decision-making. How do we compare or define 'good' versus 'bad' decisions, especially when there are large amounts of uncertainty in the impact of these decisions?" Paulson mused. "This interest naturally led me to learn about two fields: numerical optimization—how to deal with characterizing optimal solutions using computers, and feedback control—how to take advantage of newly-observed data to improve decisions over time," he said.
Paulson's background includes advanced studies in both areas. Most of his PhD was focused on numerical optimization, while his postdoctoral research concentrated on feedback control. His goal is to unify the two.
"I always saw connections / relationships between these two areas, but they have been traditionally treated separately. To get practically useful results, however, you really need to consider both areas. Optimization gives you the ability to formally define the best set of options, given some current predictions from a model; and feedback gives you the ability to correct your inaccurate predictions with measured observations so that they get closer and closer to some ground truth/reality as new data becomes available," he added.
Professor Paulson has published three book chapters and 19 peer-reviewed articles in journals such as ACS Nano, Journal of Physical Chemistry Letters, Organic Process Research & Development, Journal of Process Control, and International Journal of Robust and Nonlinear Control.
He received his PhD in 2016 from the Massachusetts Institute of Technology (MIT), where he won an NSF Graduate Research Fellowship and multiple awards for research and outstanding teaching and mentoring. His advisors were Professors Richard Braatz and Michael Strano. Paulson also obtained his M.S. CEP in chemical engineering at MIT in 2013.
Prior to joining Ohio State, Paulson was a postdoctoral researcher at the University of California, Berkeley. He was also a visiting research scientist in the Department of Electrical, Computer, and Biomedical Engineering at the University of Pavia, Italy.
Since 2020 he has been co-Principal Investigator on two other NSF projects. He and co-PIs L.-S. Fan and Research Assistant Professor Andrew Tong received a highly-competitive NSF Emerging Frontiers in Research and Innovation (EFRI) grant (funding period: 2020-2023; $2million). Also in 2020, he and two other professors won an NSF Idea Machine grant to research the ecological capacity to provide goods and services in the face of demands imposed by a technological society (January 1, 2021-23; $300,000).
Advancing efficient global optimization of extremely expensive functions under uncertainty using structure-exploiting Bayesian methods
Mathematical optimization is the process of maximizing a performance or quality indicator by identifying the best possible value among the set of all feasible options. Optimization problems arise in virtually all human endeavors related to decision making including engineering, economics, sustainability, healthcare, and manufacturing. Instances of such optimization problems are particularly challenging to solve whenever evaluating performance and/or testing for feasibility requires an expensive simulation or experiment whose results may be corrupted by random errors. Bayesian optimization (BO) is a class of machine learning-based optimization algorithms that has recently been shown to achieve state-of-the-art performance in several important applications from this problem class such as in deep machine learning, validation of expensive simulators, and material and drug design. However, traditional BO methods treat the mathematical functions that model performance and feasibility as black boxes with unknown structure, which sets a fundamental limit on their computational efficiency. This observation is the key motivation for this research project, which looks to overcome these efficiency barriers via the development of new algorithms that exploit known problem structures within the Bayesian framework. These novel capabilities will be applied to three unsolved problems currently impacting society: (1) identifying unknown mechanisms in cellular decision-making processes for biomanufacturing; (2) discovery of new sustainable and economical lithium-ion battery electrode materials; and (3) real-time energy minimization in industrial heating, ventilation, and air conditioning (HVAC) systems. In addition, the project looks to tightly integrate research and educational activities through the development of interactive workshops and games related to decision science, which will be made accessible to the public. Through collaboration with local educators, planned outreach activities also will provide K-12 students from underrepresented communities with opportunities to learn about decision science.
The proposed optimization methodology is inspired by the principle of grey-box modeling, which states that one should avoid learning what is already known when applying machine learning methods. The investigator conjectures a significant reduction in experimental and/or computational effort can be obtained in practice over traditional Bayesian optimization (BO) methods by properly leveraging prior (or domain) knowledge, which is almost always available in practice. Since prior knowledge can come in many diverse forms, the proposed research will focus on some of the most common and important examples. The three specific research aims are: (1) optimizing with hybrid physics-based and data-driven models given noisy and incomplete datasets; (2) optimizing with constrained multi-fidelity models that fuse data from a collection of heterogeneous sources of variable accuracy and cost; and (3) scaling to high-dimensional and sparse data problems through the incorporation of non-myopic and graph-structured formulations. The proposed research aims to promote convergence of statistics, machine learning, optimization, and process systems engineering. More broadly, the improved methods developed as a part of this research project will allow practitioners to solve a wide range of grey-box optimization problems with greater speed and accuracy. Planned outreach activities include educating K-12 students about decision-making under uncertainty via interactive workshops and games, incorporating new data-driven optimization material into the chemical engineering curriculum, and organizing cross-disciplinary professional workshops on the potential significance and impacts of cutting-edge BO technology.