A team from Ukraine has won the NASA Space Apps Challenge 2023 international hackathon (a form of competition over a short period to solve real-world challenges) in the Best Use of Data category.
The all-Ukrainian Storm Prophet team consisted of six members, including computer programmers and astrophysicists. It used machine learning to compute raw data from a satellite to predict geomagnetic storms on Earth.
“In this project, our primary objective was to predict geomagnetic storms based on the DSCOVR (Deep Space Climate Observatory) data spanning the years 2016 to 2023,” the team said in their project submission.
Yevhen Tatarinov, who has a Ph.D. in mathematics and is a member of the team, also outlined the project’s scope in a video.
Their work is hosted on GitHub, which detailed how the team managed to achieve the goal using long short-term memory network, a form of recurrent neural network (RNN) architecture commonly used in natural language processing (NLP) such as ChatGPT.
“Our LSTM model demonstrated high accuracy in its results. The long-term model exhibited a 3 percent deviation from the expected data, while the short-term model showed a deviation of 1.8 percent.
“We used a Sequence to Sequence model, which takes several hours of data from DSCOVR sensors as input and provides forecasts for hours/days ahead,” the team said.
NASA Space Apps Challenge is an international hackathon that started in 2012, and it encouraged specialists from around the world to “create solutions to challenges we face on Earth and in space” using open data from NASA over the course of two days (48 hours).
Winning teams would receive an invitation to visit a NASA center or facility, and each team member could invite a guest of their choice.
In 2023, there were more than 5,500 teams participating in the hackathon, with 10 winners in total.
You can also highlight the text and press Ctrl + Enter