Notebookcheck Logo
, , , , , ,
search relation.
, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,

Want to join in NASA's search for exoplanets? Google's AI can help

Schematic of a convoluted neural network trained to predict the probability of a Kepler signal caused by a planet. (Source: Google Open Source Blog)
Schematic of a convoluted neural network trained to predict the probability of a Kepler signal caused by a planet. (Source: Google Open Source Blog)
Google has released its AI and ML tools that helped NASA discover new exoplanets such as the Kepler-90 i and the Kepler-80 g. Google's model helped search 670 stars out of the 200,000 stars observed by NASA's Kepler mission and hopes to discover more such exoplanets by open sourcing the tools to anyone interested in analyzing Kepler data.

Artificial intelligence (AI) and machine learning (ML) can help in making sense of huge amounts of data. AI and ML together have helped in the development of deep learning (DL) algorithms that have been used to train artificial neural networks (ANNs) to analyze data. A recent advancement is the development of ML algorithms by Google to train ANNs to analyze data from NASA's Kepler space telescope in the hope of finding evidence of exoplanets — planets other than those in our solar system.

The Kepler and K2 missions are space observatories launched by NASA to discover the presence of Earth-sized planets — also referred to as exoplanets — orbiting other star systems with the eventual goal of finding out the possibility of extraterrestrial life on those planets. Till date, both Kepler and K2 have confirmed the existence of 2,649 exoplanet candidates. The data generated is huge and could definitely use some AI help — which is where Google stepped in.

Last year, Google's TensorFlow model combed through massive data sets to identify two new exoplanets — the Kepler-90 i and the Kepler-80 g. Now, Google is making the very same model available to all to participate in the fun of analyzing Kepler data and hopefully, discover some new worlds. In a blog post authored by Chris Shallue of the Google Brain Team, the company is now making available the source code for processing Kepler data, training the ANN, and making predictions about new exoplanet candidates. Shallue's team worked with Andrew Vanderburg at UT Austin to develop an ANN for searching exoplanetary traces. The ANN was trained using a subset of Kepler signals to distinguish between true and false positives.

Till date the model has been successful in scanning over 670 star systems out of the 200,000 identified by Kepler but Google is confident that this model is now sufficiently trained to explore other datasets. By open sourcing their model, Google hopes that the community will help in making it even more robust and enable them to play their own part in the Kepler mission.

For those who are curious, Kepler-90 i is 30% larger than our Earth with surface temperatures reaching up to 426.6 degrees Celsius. The exoplanet's discovery was made possible due to Google's above ANN model and is found to orbit the Kepler-90 star once every 14.4 days at a distance of 2,545 light years from Earth — far from being your next holiday destination.

Further information on how Kepler's detection system works and how you can use Google's model on your own can be found in the source link below.


static version load dynamic
Loading Comments
Comment on this article
Vaidyanathan Subramaniam
Vaidyanathan Subramaniam - Managing Editor - 1586 articles published on Notebookcheck since 2012
Though a cell and molecular biologist by training, I have been drawn towards computers from a very young age ever since I got my first PC in 1998. My passion for technology grew quite exponentially with the times, and it has been an incredible experience from being a much solicited source for tech advice and troubleshooting among family and friends to joining Notebookcheck in 2017 as a professional tech journalist. Now, I am a Lead Editor at Notebookcheck covering news and reviews encompassing a wide gamut of the technology landscape for Indian and global audiences. When I am not hunting for the next big story or taking complex measurements for reviews, you can find me unwinding to a nice read, listening to some soulful music, or trying out a new game.
contact me via: @Geeky_Vaidy
Please share our article, every link counts!
> Notebook / Laptop Reviews and News > News > News Archive > Newsarchive 2018 03 > Want to join in NASA's search for exoplanets? Google's AI can help
Vaidyanathan Subramaniam, 2018-03-12 (Update: 2018-03-12)