CSC4999 Directed readings

Machine Learning in Space Weather Forecasting

Administrative Info

Semester: Spring, 2020

Instructor: Berkay Aydin


Hours: Tuesdays 12:15 – 1:00 PM

Location: 25 Park Pl NE - Room 740


Our modern, technologically complex systems - including communications, transportation, and electrical power systems - can be disrupted and damaged by extreme space weather events. In this course, we will review predictive data analytics techniques to tackle the problem of space weather forecasting, specifically, one of the most prominent event types - solar flares. During this course, we will cover and discuss the feature selection, machine learning models, sampling methodologies, and evaluation metrics related to space weather forecasting.

Specific Learning Outcomes

Goal of this course is to provide background on the impacts of space weather events, how they are currently being predicted using advanced machine learning approaches and challenges associated with predicting space weather events. At the end of this course, students are expected to have a fundamental understanding of forecasting rare events, multivariate time series classification, and applied machine learning techniques on space weather forecasting.

Grading Criteria

In order to get a passing grade in this course, students are expected to

(1) read and understand the assigned papers and briefly present their findings (45%),

(2) participate in weekly lecture hours (10%),

(3) apply fundamental machine learning algorithms for space weather forecasting and report their findings (45%).

Academic Integrity

Students in this course are expected to abide by the Georgia State University Code of Academic Conduct. Any work submitted by a student in this course for academic credit will be the student's own work. Assignments that have been previously submitted in another course may not be submitted for this course.

Course Topic & Reading

The following overarching topics and papers will be covered in this course. Students are expected to read the assigned papers and summarize their findings in weekly lectures:

- A Brief History of Space Weather Forecasting and Impacts of Space Weather

Paper 1: Singer et al., (2001), Space Weather Forecasting: A Grand Challenge

Paper 2: Rust, Lin (1998), Solar Flares in From the Sun (pg 81 - 103)

- Solar Flare Prediction I - Challenges

Paper 3: Georgoulis, (‎2012), On Our Ability to Predict Major Solar Flares

- Solar Flare Prediction II - Machine Learning Approaches

Paper 4: Ahmad et al., (2011), Solar Flare Prediction Using Advanced Feature Extraction, Machine Learning, and Feature Selection

Paper 5: Bobra et al., (2015), Solar Flare Prediction Using SDO/HMI Vector Magnetic Field Data with a Machine Learning Algorithm

- Solar Flare Prediction III - Machine Learning Approaches

Paper 6: Florios et al. (2018), Forecasting Solar Flares Using Magnetogram-based Predictors and Machine Learning

- Solar Flare Prediction IV - Time Series Approaches

Paper 7: Ahmadzadeh et al., (2020), How to Train Your Flare Prediction Model: On Robust Sampling of Rare Events

- Space Weather Analytics Dataset for Solar Flare Prediction

Paper 8: Angryk et al. (2020), Multivariate time series dataset for space weather data analytics

- A Review on Time Series Classification

Paper 9: Bagnall et al., (2017), The great time series classification bake off: a review and experimental evaluation of recent algorithmic advances

- Building Baseline Machine Learning Models for Solar Flare Prediction

Project + Report