Machine Learning Methods in Magnetospheric-Physics Time Series Data

No Thumbnail Available

Files

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Bachelor's thesis
Electronic archive copy is available locally at the Harald Herlin Learning Centre. The staff of Aalto University has access to the electronic bachelor's theses by logging into Aaltodoc with their personal Aalto user ID. Read more about the availability of the bachelor's theses.
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2024-09-06

Department

Major/Subject

Data Science

Mcode

SCI3095

Degree programme

Aalto Bachelor’s Programme in Science and Technology

Language

en

Pages

26

Series

Abstract

Today, a wealth of magnetospheric time series data from near-earth missions are utilized for various research purposes. One of the most common usage of the data is the categorization and detection of boundary crossing events to study the state of the magnetosphere throughout the duration of the mission. Due to the complexity of this task and the availability of magnetospheric data, many different machine learning algorithms have been employed to assist humans with various levels of accuracy. This thesis presents a comprehensive comparison of a 3D convolutional neural network (3D-CNN), a fully convolutional neural network (FCN), and a boosting algorithm when applied to data collected by the Magnetospheric Multiscale (MMS) mission. The 3D-CNN model has the best overall accuracy at over 98%, followed by the lighter weight boosting algorithm with approximately 96% true positive rate. Finally, the FCN underperformed compared to the other models, still with over 89% accuracy. This result could be mainly attributed to the differences in datasets and the number of labels chosen. So, these variations notably should be minimized for future analysis of machine learning models. Despite the drawbacks of this comparison, it is shown that various machine learning models with different flexibility and weights are fully capable and suitable for the task of region detection in the magnetosphere.

Description

Supervisor

Korpi-Lagg, Maarit

Thesis advisor

Weigt, Dale

Keywords

machine learning, magnetosphere, MMS, THEMIS, time series data

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202411197173

Collections

[kand] Perustieteiden korkeakoulu / SCI