Breakthrough in Phylogenetic Classification of Extinct Organisms Using Neural Networks: A Study in PNAS Nexus

Researchers Make Breakthrough in Using Neural Networks to Classify Extinct Organisms in Phylogenetic Trees

In a groundbreaking study published in PNAS Nexus, a team of researchers has successfully utilized neural networks to accurately classify extinct organisms within phylogenetic trees. Led by Surangi Punyasena, an associate professor of Plant Biology at the University of Illinois Urbana-Champaign, the team’s innovative approach marks a significant advancement in the field of paleontology.

The key to the team’s success lies in training the neural network model to recognize and rank organism features based on known phylogenetic information. By incorporating phylogeny into the model’s training process, the researchers were able to overcome previous challenges faced by neural networks in accurately classifying extinct organisms.

According to Marc-Élie Adaimé, a graduate student in Punyasena’s lab and first author on the study, the traditional approach of training neural networks on straightforward classification tasks limited their ability to accurately classify extinct organisms. By introducing phylogenetic context into the model, the researchers were able to create a new modeling approach that could accurately position organisms within a phylogenetic framework.

The team chose to apply their model to the classification of pollen and spores, ancient entities found throughout the fossil record. By training the model on optical superresolution images of modern and fossil pollen, the researchers were able to accurately classify extinct pollen specimens from Panama, Peru, and Columbia.

Punyasena emphasized the importance of leveraging advances in deep learning and computer vision to analyze and interpret fossil pollen. The researchers plan to expand their work to include a broader set of fossil pollen data and to apply their model to categorizing various fossil organisms beyond pollen.

The study, funded by the National Center for Supercomputing Applications and the University of Illinois, represents a significant step forward in the field of paleontology. The researchers’ innovative approach to utilizing neural networks in classifying extinct organisms within phylogenetic trees has the potential to revolutionize the way researchers decipher the evolutionary relationships of extinct organisms from fossils.

For more information, the study can be found at https://doi.org/10.1093/pnasnexus/pgad419.