Pattern Recognition and Matching in Ice Core Data

• Main Author: Dunn, Nathan
• Format: Text
• Language: unknown
• Published: DigitalCommons@UMaine 2015
• Subjects: ice core; pattern recognition; machine learning; Computer Sciences; Mathematics
• Online Access: https://digitalcommons.library.umaine.edu/honors/224; https://digitalcommons.library.umaine.edu/context/honors/article/1216/viewcontent/Dunn_Thesis_Final.pdf

The purpose of this research is to investigate the potential of applying concepts from ma- chine learning, such as pattern recognition and matching, to detect climatic signals in ice core data. The main components of this project are the development of a pattern language for expressing relationships between chemical signals over time, a method of tokenizing ice core chemistry data into an easily manageable form, a method of matching specific instances of climatic signals to a specific pattern string, and a method to recognize and evaluate patterns within ice core chemistry data. While there are weaknesses in each of these components, this research serves as a successful proof of concept for the feasibility of applying machine learning techniques to ice core analysis.