- Type Validation
- Level Advanced
- Time Weeks
- Cost Paid
Applied Machine Learning for Physical Systems
Issued by
University of Connecticut
Earners of the Applied Machine Learning for Physical Systems Badge have acquired foundational knowledge of applied aspects of machine learning (ML), including methods for handling uncertain, small, and imbalanced data, feature selection and representation learning, and model selection and assessment for materials, chemistry, and physics applications. Earners have gained exposure to advanced ML research in areas such as interpretability, stability, and meta-learning.
- Type Validation
- Level Advanced
- Time Weeks
- Cost Paid
Skills
- Complex Systems Modeling
- Critical Thinking
- Cyber-Physical Systems Engineering
- Data Curation
- Data Engineering
- Data Modeling
- Experimental Data
- Machine Learning
- Mathematical Approximation
- Mathematical Modeling
- Statistical Modeling
- Systems Analysis
- Systems Engineering
- Systems Modeling
- Uncertainty Quantification
Earning Criteria
-
Badge earners complete SE-5602 Machine Learning for Physical Sciences and Systems course at the University of Connecticut, which is a hybrid-online graduate course that can be taken from anywhere in the world. Earners can take this graduate course as a matriculated UConn graduate student or as a non-degree graduate student, which does not require admission to the UConn graduate school. Badge holders complete a course-long project and must earn a B- or better on this project to earn the badge.
-
Badge earners can apply machine learning methods and practices to the design and operation of cyber-physical systems, and can describe and characterize uncertainty, statistical inference, and estimation in models and analyze their effects on cyber-physical system behavior. See Standard [1] NAE-CPS below.
-
Badge earners can perform modeling and analysis to design and predict operating characteristics for a complex system and can develop and use approaches for deriving computationally tractable approximations to systems that are formulated in very high dimensional spaces, such as those arising in quantum mechanics. See Standard [2] DOE-SIAM below.
-
Badge earners can develop and use systematic mathematical approaches for constructing nonlinear empirical models informed by physics principles and can develop and use mathematically rigorous frameworks and efficient, robust numerical methods for data assimilation into models of complex systems that are informed by statistics-based error estimates for the assimilated data. See Standard [2] DOE-SIAM below.
-
Badge earners can incorporate observational and experimental data to model and simulate a complex system and can develop sound, computationally feasible strategies and methods for the collection, organization, statistical analysis, and use of data associated with complex systems. See Standard [2] DOE-SIAM below.
-
Badge earners can apply data engineering processes to a real-world problem using a modeling approach and can apply data engineering methods and principles to the design and operation of a cyber-physical system, can explain why probability and statistics are both relevant to engineering, with examples, and can explain why uncertainty is an important factor in engineering and explains how it might arise from many sources. See Standard [3] INCOSE ISECF below.
-
Badge earners can describe the scope and limitations of models including definition, implementation, and analysis, can describe different types of modeling and provide examples, can explain how the purpose of modeling affects the approach taken, can explain why models have a limit of valid use, and the risks of using models outside those limits, and can explain why models are developed for a specific purpose or use and provides examples. See Standard [3] INCOSE ISECF below.
-
Badge earners can use modeling tools and techniques to represent a system or system element, can interpret and use outcomes of modeling and analysis, can explain why conclusions and arguments made by others may be based upon incomplete, potentially erroneous or inadequate information with examples, and can explain why assumptions are important and why there is a need to ensure that they are based upon sound information. See Standard [3] INCOSE ISECF below.
-
Badge earners will have prepared a 15-minute paper presentation on a paper applying machine learning to a scientific domain, to be shared with other members of the class and have participated in an online class discussion forum related to their presentation. Earners also complete a course project that focuses on applying machine learning methods to address a problem in the physical sciences. Deliverables include a project proposal, midterm report, and final report.
Standards
A 21st Century Cyber-Physical Systems Education. Committee on 21st Century Cyber-Physical Systems Education; Computer Science and Telecommunications Board; Division on Engineering and Physical Sciences; National Academies of Sciences, Engineering, and Medicine. ISBN 978-0-309-45163-5 | DOI: 10.17226/23686
SIAM APPLIED MATHEMATICS AT THE U.S. DEPARTMENT OF ENERGY: Past, Present and a View to the Future. A Report by an Independent Panel from the Applied Mathematics Research Community May 2008.
INCOSE Systems Engineering Competency Framework. July 2018. INCOSE Technical Product Reference: INCOSE-TP-2018-002-01.0