Tricorder Tech: AI Matches Human Experts In Classifying Microscopic Organisms
Editor’s note: If we aspire to mount expeditions to new worlds and then embrace the task of characterizing and quantifying whatever life forms we find, the ability to map and understand whatever metabolic and genomic systems are in operation is important.
Not only do we need to know how alien biota function, but also how they evolved – what differences and similarities they may have with the origin and evolution of life on Earth. Increasing in situ capabilities like this can allow much more preliminary analysis to be done on site – or back on Earth.
As we begin to expand our search for life to other worlds we are going to need to be economical interms of the equipment we send and how we reality new knowledge back to Earth. Sample return missions are difficult even when worlds are close to one another. Doing in situ examination and documentation is going to be very important as we explore other worlds. Not only does it reduce the logistics of sending things back home but it allows data to be sent back at the speed of light. It also allows the astronaut/droid teams to engage in empirical exploration – learning from what they found so as to refine and perfect their continued searching.
As we continue to explore a world, a collection of data wil be amassed that is processed so as ot form a nascent catalog of life forms. As they are identified and named that catalog- contained within an AI wil be increasingly able to identify and differentiate successive life forms as they are encountered – and do so quickly and on the spot in situ thus amplifying the ability of human explorers and their robotic companions to understand a new world’s biota.
Foraminifera (forams) are shelled microorganisms that are abundant in the Earth’s seabed. Analyzing different species of forams provides important information about climate change, the state of the marine environment, and suitable areas for carbon capture and storage.
Past research has attempted to automate these classification tasks—a usually laborious and time-consuming manual process—with deep learning (DL) methods. Several studies show significant promise, but few have focused on the uncertainty of the methods’ classifications.
“Uncertainty estimation is crucial to avoid misclassifications that could overlook rare and ecologically significant species, says Iver Martinsen, PhD Candidate at UiT The Arctic University of Norway and SFI Visual Intelligence (VI). “It is important to develop DL methods which accurately calculate how uncertain their predictions are.”
In a recently published study in Artificial Intelligence in Geosciences, Martinsen and researchers at UiT, VI, Nofima, and NSE show how deep learning can achieve human-level performance in estimating uncertainty when classifying forams. “Using 260 images of forams and sediment grains, we trained the DL methods to detect and classify these microscopic organisms,” says Martinsen.
Evaluating the performance of such methods remains a challenging. To address this, the researchers created a human-derived set of uncertainty estimations based on classification task responses from four senior geoscientists.
“The geoscientists were given the same 260 images and were tasked to classify each of them, as well as state their confidence level. This formed a comparative baseline which allowed us to assess the models’ estimations to those of human experts,” Martinsen explains.
The study also demonstrates how human uncertainty estimations may provide a relevant and valuable baseline for comparison, he adds. Results show that the DL methods’ estimations can match—and at times be better than—expert geoscientists.
“We gain valuable insights on how these methods’ estimations compare to each other and human experts. We believe this research is a leap towards making these automated tools more reliable, trustworthy, and applicable in real-world settings,” Martinsen says.
Quantifying uncertainty in foraminifera classification: How deep learning methods compare to human experts, Artificial Intelligence in Geosciences (open access)
Astrobiology,
