In a statement released by NASA last month, the agency said that A.I. has the potential to help work on some of the biggest problems in space science. A.I. could be used to search for life on other planets or identify asteroids. NASA scientists are partnering up with leaders in the AI industry, like Intel, IBM, and Google. Together, they can apply advanced computer algorithms to solve some of those problems.
There are certain A.I. technologies that NASA is relying on, such as machine learning, to interpret data. This data will then be collected by telescopes, including the James Webb Space Telescope or the Transiting Exoplanet Survey Satellite, at some point in the future.
Giada Arney, an astrobiologist at NASA’s Goddard Space Flight Center in Greenbelt, Maryland, hopes that machine learning can help her and her team find some indication of life in data that will be collected by the telescopes and observatories.
“These technologies are very important, especially for big data sets and especially in the exoplanet field,” Arney said in the statement. “Because the data we're going to get from future observations is going to be sparse and noisy. It's going to be really hard to understand. So using these kinds of tools has so much potential to help us.”
NASA runs an eight-week program every summer that brings together leaders in the technology and space sectors, called Frontier Development (FDL).
Shawn Domagl-Goldman is a NASA Goddard astrobiologist.
“FDL feels like some really good musicians with different instruments getting together for a jam session in the garage, finding something really cool, and saying, ‘Hey we've got a band here,'” he said in the statement.
Back in 2018, an FDL team was mentored by Domagal-Goldman and Arney, and they developed a machine learning technique that relies on neural networks. They analyze images and identify the chemistry of exoplanets by using the wavelengths of light emitted or absorbed by molecules in their atmosphere.
By using this new technique, researchers could identify various molecules in the atmosphere of WASP-12b, an exoplanet. The technique did this more accurately than other methods.
According to Domagal-Goldman, the neural network can also identify when there is a lack of data. The Bayesian technique, as it is called, can also tell scientists how certain it is about its prediction.
“In places where the data weren’t good enough to give a really accurate result, this model was better at knowing that it wasn’t sure of the answer, which is really important if we are to trust these predictions,” Domagal-Goldman said.
The Bayesian technique is still being developed, but other FDL technologies are being used in the real-world. By 2017, a machine learning program was developed by FDL participants that was capable of quickly creating 3D models of nearby asteroids. It could also accurately estimate their shapes, sizes, and spin rates. This type of information is useful for NASA to detect and deflect asteroids that threaten Earth.
Astronomers traditionally use simple computer software to create 3D models, and it analyzes radar measurements of a moving asteroid. It then provides useful information to help scientists infer its physical properties based on changes in the radar signal.
Bill Diamond is SETI’s president and chief executive officer.
“An adept astronomer with standard compute resources, could shape a single asteroid in one to three months,” Diamond said. “So the question for the research team was: Can we speed it up?”
The team consisting of students from France, South Africa and the United States, along with mentors from academia and technology company Nividia, developed an algorithm capable of rendering an asteroid in as little as four days. The technique is currently used by astronomers at the Arecibo Observatory in Puerto Rico, and it does real-time shape modeling of the asteroids.
Researchers are also suggesting that A.I. technologies be built into future spacecraft, and that it would allow the spacecraft to make real-time decisions.
“A.I. methods will help us free up processing power from our own brains by doing a lot of the initial legwork on difficult tasks,” Arney said. “But these methods won't replace humans any time soon, because we'll still need to check the results.”
- The Black Box Problem in LLMs: Challenges and Emerging Solutions
- Alex Ratner, CEO & Co-Founder of Snorkel AI – Interview Series
- Circleboom Review: The Best AI-Powered Social Media Tool?
- Stable Video Diffusion: Latent Video Diffusion Models to Large Datasets
- Donny White, CEO & Co-Founder of Satisfi Labs – Interview Series