Imagemics is poised to enable a new understanding of life.

We use the CLIP objective (c) to train a ViT-B/16 on over 450K different class labels, all of which are taxonomic labels from the Tree of Life (a). Since the text encoder is an automatic language model, sequence representation can only depend on higher levels such as class, phylum, and kingdom (b). This naturally represents a hierarchy for the labels, helping the vision encoder learn image representations that are more closely aligned with the tree of life. Credit: BioCLIP: A Vision Foundation Model for the Tree of Life, The Ohio State University. https://imageomics.github.io/bioclip/

Imageomics, a new field of science, has made tremendous strides in the past year and is on the way to major discoveries about life on Earth, according to one of the founders of the discipline.

Tanya Berger Wolf, faculty director of the Translational Data Analytics Institute at The Ohio State University, outlined the state of imagemics in a presentation on February 17, 2024. American Association for the Advancement of Science Annual Meeting.

“The age of amigomics is coming and it’s ready for its first big discoveries,” Berger-Wolf said in an interview before the meeting.

Imageomics is a new interdisciplinary scientific field that focuses on using Tools for understanding the biology of organisms, especially biological traits, from images.

Those images can come from camera traps, satellites, drones, even vacation photos that tourists take of animals like zebras and whales, said Berger-Wolf, director of Ohio State’s Imageomics Institute.

These images contain a wealth of information that scientists have not been able to properly analyze and use and machine learning.

The field is new — the Imagemics Institute was just founded in 2021 — but big things are happening, Berger-Wolf told AAAS.

A major area of ​​study that is bearing fruit involves how phenotypes—the observable traits of animals that can be seen in photographs—are related to their genomes, the sequences of DNA that produce those traits. she does.

“We are on the cusp of understanding the direct correlation of observable phenotype to genotype,” he said.

“We couldn’t do it without imagemics. It’s advancing both artificial intelligence and biological science.”

Berger-Wolf cites new research on butterflies as an example of advances in imagemics. He and colleagues are studying replication. The appearance of which is similar to different species. One reason for mimicry is to look like a species that predators like birds avoid because they don’t have an appealing taste.

In these cases, birds—as well as humans—can’t tell the species apart by looking at them, even though the butterflies themselves know the difference. However, machine learning can analyze images and detect very subtle differences in color or other traits that distinguish butterfly species.

“We can’t tell them apart because these butterflies didn’t evolve these traits for our benefit. They evolved to signal themselves and their predators,” he said.

“The signal is there—we just can’t see it. Machine learning can allow us to figure out what those differences are.”

But more than that, we can use the imagomics approach to alter images of butterflies to see how wide the differences in mimicry need to be to fool the birds. Researchers are planning to print realistic images of butterflies with subtle differences to see how real birds respond.

It’s doing something new with AI that hasn’t been done before.

“We’re not just using AI to redefine our knowledge. We’re using AI to generate new scientific hypotheses that are actually testable. That’s exciting,” Bergerwolf said. Berger-Wolf said.

Researchers are going even further with an imageomics approach to link these subtle differences in how butterflies see the actual genes that cause these differences.

“Over the next few years we’re going to learn a lot that will push imagemics into new areas that we can only imagine right now,” he said.

A key goal is to use this new knowledge generated by imagomics to find ways to protect endangered species and the habitats where they live.

“A lot of great things will come out of Imagemics in the coming years,” Bergerwolf said.

Berger-Wolf’s AAAS presentation, titled “Imagemics: Images as a source of information about life“is part of the session”Imageomics: Powering Machine Learning to Understand Biological Traits

More information:
Imagemics: Images as a Source of Information about Life, aaas.confex.com/aaas/2024/meet… gapp.cgi/Paper/32018

Reference: Imageomics poised to enable new understanding of life (2024, February 17) Retrieved February 18, 2024, from https://phys.org/news/2024-02-imageomics-poised-enable-life.html

This document is subject to copyright. No part may be reproduced without written permission, except for any fair dealing for the purpose of private study or research. The content is provided for informational purposes only.