Editor’s Note
This profile highlights the vital intersection of data science and planetary defense. Gema Parreño Piqueras’s work with NASA exemplifies how advanced algorithms and machine learning are being deployed to address global challenges, transforming vast datasets into actionable insights for safeguarding our planet.
Gema Parreño Piqueras, a data scientist from Albacete, Spain (born 1988), operates in a world of big data analysis, mathematical models, algorithms, artificial neural networks, software development, infographics, and 3D animations. Two years ago, she accepted a challenge from NASA to create an artificial intelligence system capable of classifying asteroids and predicting their potential impacts on Earth. To meet this challenge, she utilized TensorFlow, an open-source machine learning library developed by Google. Google now highlights her program as an example application of this tool for developers. BBVA also hired her to detect banking fraud using artificial intelligence. However, before all this, Parreño had a background in architecture. She studied at the Polytechnic University of Madrid, where she laid the foundation for her future in data science.
After university, the economic crisis pushed her to seek opportunities beyond architecture. This led her to co-found a startup active in the video game sector.
During this time, Parreño began taking programming courses.
She began seeking challenges to test herself.
Thus, her path intersected with that of thousands of asteroids.
In 2016, Parreño participated in the Space App Challenge, an annual competition organized by NASA. One of the main goals of the event is to challenge young scientists to propose solutions to solve various highly complex problems related to space. In that edition, one of the challenges was to create a system to analyze large volumes of information about Near-Earth Objects (NEOs), i.e., bodies like asteroids or comets orbiting the Solar System. The goal was for the program to become capable of classifying these bodies and predicting their behavior, to expand knowledge about them and anticipate possible dangerous impacts on Earth.
Parreño, along with her five teammates Julián Fernández, María del Mar Núñez, Samuel Góngora, Denis Ciccale, and José Antonio Martínez, had a weekend to focus on the problem and find a possible solution. The following week, the team technically developed the software. The data scientist explains that they started from a base of “100,000 data samples,” collected by NASA since the 1930s.
She explains that Google’s TensorFlow technology uses a set of algorithms, defined as a neural network, which allows processing vast amounts of data. Due to this characteristic, it can be applied to many different fields depending on the data fed into it, she adds.
The architecture of this artificial intelligence is based on the superposition of different layers of knowledge. In this way, the program developed by Parreño, called Deep Asteroid, is capable of simultaneously analyzing and processing information such as an asteroid’s mass, color, and age along with other data like the number of observations of that body or orbital parameters.
NASA considers there has been an impact of an NEO if the object passes within 0.01 astronomical units (the distance between Earth and the Sun), explains Parreño. And scientists are not only interested in knowing if an asteroid will literally fall and cause damage, she adds.
The scientist from Albacete states that the impact of small, non-dangerous bodies is very frequent.
What Deep Asteroid allows, she details, is to know daily if any impact is predicted in the next 24 hours and where.
Parreño says she continued perfecting the software for a year. Currently, NASA is studying its application. The time she dedicated to the program served both to gain confidence in data science and to learn new things about NEOs, she assures. She understood, for example, that asteroids are divided into several groups according to the type of orbit they have. Therefore, her program focuses especially on the two groups with the highest probability of impacting Earth.