Biological Shape-Space: Applying Machine Learning to the Fossil Pollen Record

At the Ideas Lab, Surangi Punyasena - an early-career paleocologist with limited background in computer science or image analysis - teams up with computer scientists and applied mathematicians from four different universities. Together they find a new way to approach fossil pollen identifications in the geological record, potentially speeding up climate change studies. The new machine-learning algorithms may also have use in other shape classification challenges, well beyond paleontology.