Finding Exoplanets With Machine Learning: the Use of Microsoft Azure Cloud Computing Services to Process NASA Tess Telescope Data

School Name

South Carolina Governor's School for Science & Mathematics

Grade Level

12th Grade

Presentation Topic

Physics

Presentation Type

Mentored

Abstract

The increase of artificial intelligence in the world is changing how we go about our lives. The vast amount of data that is becoming available provides an opportunity to discover physical objects that were hidden before. Using NASA's TESS telescope data in combination with Microsoft Azure, this study focuses on how one can use data and artificial intelligence to go as far as discover a new planet. To do this, a resource group was created on Microsoft Azure using software as a service. From the container instances inside the resource group one can process and view the data downloaded. The data measures the brightness of nearby stars and detects reflex motion. From this information, Full Frame Images are cut into Target Pixel Files to prepare light curves. A machine learning algorithm then predicts the probability of the existence of an exoplanet orbiting the star. Thousands of planets have been discovered by tracking a star's brightness in transit. This number will increase rapidly with the advent of TESS's large amount of data and the increasing sophistication of artificial intelligence. These results are important to the study of both astrophysics and data science, especially the questions of how they can be used together in future studies. It makes a point to show how the onset of artificial intelligence is making a study previously done by a PhD available to anyone with the motivation to learn.

Location

Furman Hall 127

Start Date

3-28-2020 11:45 AM

Presentation Format

Oral Only

Group Project

No

COinS
 
Mar 28th, 11:45 AM

Finding Exoplanets With Machine Learning: the Use of Microsoft Azure Cloud Computing Services to Process NASA Tess Telescope Data

Furman Hall 127

The increase of artificial intelligence in the world is changing how we go about our lives. The vast amount of data that is becoming available provides an opportunity to discover physical objects that were hidden before. Using NASA's TESS telescope data in combination with Microsoft Azure, this study focuses on how one can use data and artificial intelligence to go as far as discover a new planet. To do this, a resource group was created on Microsoft Azure using software as a service. From the container instances inside the resource group one can process and view the data downloaded. The data measures the brightness of nearby stars and detects reflex motion. From this information, Full Frame Images are cut into Target Pixel Files to prepare light curves. A machine learning algorithm then predicts the probability of the existence of an exoplanet orbiting the star. Thousands of planets have been discovered by tracking a star's brightness in transit. This number will increase rapidly with the advent of TESS's large amount of data and the increasing sophistication of artificial intelligence. These results are important to the study of both astrophysics and data science, especially the questions of how they can be used together in future studies. It makes a point to show how the onset of artificial intelligence is making a study previously done by a PhD available to anyone with the motivation to learn.