An algorithm that helps discover more than 300 exoplanets in deep space


NASA said that the existence of 301 exoplanets has been confirmed thanks to a new deep learning algorithm, as scientists made this discovery thanks to the deep neural network. ExoMiner, which was created using data from the spacecraft Kepler affiliated with NASA K2.

According to the British newspaper, “Daily Mail”, the space agency’s supercomputer uses Pleiades”, which is able to decipher the difference between real and false exoplanets.

Also, with the addition of the newly confirmed planets, which orbit distant stars in the universe, this brings the total number of confirmed exoplanets to 4,870.

Unlike other machine learning programs for detecting exoplanets, ExoMiner It is not a black box, there is no ambiguity about why it decides that something is a planet or not, it can easily explain the features in the data that drive ExoMiner To reject or confirm a planet.

NASA reveals that exoplanets are “confirmed” when various observational techniques highlight features only visible by planets; And they are validated when using statistics.

Newly published research shows that a neural network is more consistent and accurate when it eliminates the wrong possibilities that scientists have spotted, and it gives researchers additional details about why they take ExoMiner for the decision you made.

Added Hamid Valizadegan, Head of Project ExoMiner And the machine learning manager: “When he says ExoMiner If something is a planet, you can be sure it’s a planet.”

Prepare ExoMiner Extremely accurate and in some ways more reliable than both current instrument classifiers and human experts, NASA notes that of the 301 exoplanets added to the list, none are thought to be Earth-like or in the habitable zone of their parent stars, but some It shares certain characteristics of other exoplanets close to Earth.


Please enter your comment!
Please enter your name here