Papers, like this one from Google, show that Deep Learning and Machine Learning are effective in detecting exoplanets from luminosity data provided by telescopes like TESS or Kepler.

These models seem extremely quick and lightweight. Has it become somewhat of a standard practice to use these on freshly obtained TESS data? If not, why, and what is done with the data instead? Are there drawbacks to using these seemingly accurate models?

| improve this question | | | | |

Also called Astronet-K2, the deep learning A.I is efficient at spotting data that could find exo-planets, but that's were the issue is. That's all it is capable of doing, so it misses or ignores other potentially interesting data that human astronomers would classify as curious or odd, and could lead to other potentially new discoveries. It's missing the human factor, more or less. In both the source linked in the question and mine, it has only found 2 new exo-planets, so it doesn't look like is has free reign In the field. Even if the A.I is currently used automatically, it's work would be monitored and checked by humans. My source material was from 2019, so if I missed a more recent accounting of the system that has different or upgraded methods, please let me know. https://www.google.com/amp/s/www.technologyreview.com/2019/04/01/136239/deep-learning-has-found-two-exoplanets-that-human-astronomers-missed/amp/

| improve this answer | | | | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.