AI-based Tool to find Habitable Planets

February 11, 2022

Astronomers from Indian Institute of Astrophysics have developed a new way to identify potentially habitable planets, with a high probability.

Key Points

Astronomers developed this new way in association with astronomers from BITS Pilani, Goa campus.
They used an Artificial Intelligence (AI)-based algorithm for this, where 60 potentially habitable planets have been found out of 5,000 known planets.

About the new method

The new AI-based method is named as “Multi-Stage Memetic Binary Tree Anomaly Identifier (MSMBTAI)”. This method is based on a novel multi-stage memetic algorithm (MSMA), which uses generic notion of a meme. Meme is a knowledge or idea that gets transferred from one person to another by imitation. This method was devised on the basis of a belief that Earth is an anomaly, with possibility of existence of more anomalies among thousands of data point. There are 60 potentially habitable planets out of 5000 confirmed and 8000 candidate planets proposed. The assessment was done on the basis of their close similarity to Earth. These planets can be regarded as candidates for anomalous instances in huge pool of non-habitable’ exoplanets.

What is an anomaly?

Earth being the only habitable planet among various planets is known as an anomaly.

Anomaly detection technique

The anomaly detection technique which is suitable for industrial systems applies equally for habitable planet detection. This is because in both the cases, anomaly detector is dealing with imbalanced data, where anomalies are outliers. These are less in number as compared to the normal data. However, with increasing number of discovered exoplanets, finding rare anomalous instances by characterizing them with respect to planetary parameters, populations, types, and habitability potential needs the knowledge of multiple planetary parameters from observations.

« Previous PostNext Post »