Nov 15, 2018 | Updated: 03:14 AM EDT

Artificial Intelligence Helps In Astronomical Research

Mar 08, 2017 05:46 AM EST

Close

Artificial Intelligence or AI is the knowledge acquired by a system for proper completion of certain tasks. It has various uses in different spheres of science. The latest to embrace it is space science or astronomy.

According to Wired, astronomers recently are referring to leading-edge machine learning techniques of AI and use them to sharpen up blurry, noisy images from space. The trend of larger automation in the field of astronomy is fast becoming popular. The "Autodidactic" machines can identify, classify and clean up their data much more efficiently than the humans. It is being claimed that machine learning is going to be a standard digital instrument to be used vastly in astronomy.

The GAN or Generative Adversarial Network is a technique used as a part of AI in astronomy nowadays. It does "Computational Plastic Surgery" of scarred pictures received from the outer space to get pristine images. It is being claimed that these machine learning or AI techniques can turn out to be of great help in the future for astronomers.

According to Nature, AI researchers are very excited to use generative networks to train image-recognition software. It is also seen that the AI machines used by astronomers to keep noise away from large data sets, and understanding the pattern within them.

AI researchers have found new ways of generating images. An instrument known as the VAE (Variational Auto-Encoder) looks to generate new images, slightly less realistic but more diverse images than GAN. Further variants are also being generated, combining the features of GANs and VAEs for improved usage of AI.

However, the scientists are wary of the fact that the inner workings of the AI machines might be complex and can give out realistic but faulty outputs. These machines might show effective and correct results but can also give out real looking yet false videos and images and this is why it can not be completely relied upon.

©2017 ScienceTimes.com All rights reserved. Do not reproduce without permission. The window to the world of science times.
Real Time Analytics