A newly developed machine learning tool could help scientists look for signs of life on Mars and other alien worlds. Given the very limited ability to collect samples from other planets, scientists currently have to rely on remote sensing techniques to look for signs of extraterrestrial life. life. This means that any method that can redirect or improve this search would be incredibly useful.

With this in mind, a multidisciplinary team of scientists led by Kim Warren-Rhodes of the Institute SET (The Search for Extraterrestrial Intelligence) has mapped rare life forms living in salt domes, rocks, and crystals on the Salar de Pajonales, a salt flat in California. on the border with Chile Atacama Desert and the Altiplano, or the highlands.

Warren-Rhodes then worked with Michael Phillips of the Johns Hopkins University Applied Physics Laboratory and Oxford University researcher Freddie Kalaitzis to train the model. machine learning Recognize the patterns and rules related to the distribution of life in a harsh region. This training taught the model to perceive the same patterns and conventions for a wide variety of landscapes, including those that may be located on other planets.

The team found that their system, which combined statistical ecology with artificial intelligence, was able to find and detect biosignatures up to 87.5% of the time. This is comparable to a success rate of no more than 10% achieved by random searches. In addition, the program can reduce the space required to search by up to 97%, helping scientists greatly improve their search for potential traces of chemical life or biosignatures.

“Our framework allows us to combine the power of statistical ecology with machine learning, discovering and predicting the patterns and rules that enable nature to survive and distribute itself in Earth’s most challenging landscapes,” Warren-Rhodes said in a statement..(opens in a new tab). “We hope that other groups of astrobiologists will adapt our approach to map other habitable environments and biosignatures.”

Such machine learning tools could be applied to robotic planetary missions like NASA’s Mars rover, the researchers say. Patiencecurrently searching for traces of life at the bottom of the Martian crater Jezero.

“With these models, we can develop customized roadmaps and algorithms to guide rovers to places most likely to harbor past or present life, no matter how hidden or rare,” Warren-Rhodes said.

Choosing an analogue of Mars on Earth

The team chose Salar de Pajonales as a testing phase for their machine learning model because it is a suitable analogue for the dry and arid landscape of the modern world. Anthem. The region is the dried salty bottom of a high alpine lake, bursting with a high degree of ultraviolet radiation. Although the Salar de Pajonales are considered extremely inhospitable to life, there is still a sanctuary for some creatures.

The team collected nearly 8,000 images and more than 1,000 samples from the Salar de Pajonales to identify the photosynthetic microbes living in the region’s salt domes, rocks, and alabaster crystals. Pigments secreted by these microbes represent a possible biosignature. NASA’s “Ladder of Life Discovery”. It is designed to guide scientists in the search for life beyond Earth within the practical limitations of robotic space missions.

The team also studied Salar de Pajonales using drone images similar to images of the Martian terrain taken by the High Resolution Imaging Experiment (HIRISE) camera on NASA’s Mars Reconnaissance Orbiter. These data allowed them to determine that microbial life in the Salar de Pajonales was not randomly distributed, but instead concentrated in biological hotspots closely associated with the presence of water. CNN) for recognizing and predicting major geological features in the Salar de Pajonales. Some of these features, such as patterned ground or polygonal webs, are also present on Mars. CNNs are also trained to detect and predict smaller microenvironments more likely to contain biosignatures.

For now, the researchers will continue to train their AI on the Salar de Pajonales, and then aim to test CNN’s ability to predict the location and distribution of ancient stromatolite fossils and salt-tolerant microbiomes. This will help him learn whether the rules he used in this search can also be applied to finding biosignatures in other similar natural systems.

After that, the team aims to begin mapping hot springs, frozen permafrost and rocks in dry valleys, hoping to train the AI ​​to focus on potential habitats in other extreme environments on Earth before potentially exploring other planets.