Benchmarks Physics-Informed Neural ODEs vs. Fine-Tuned Foundation Models (Chronos) on Cenozoic fossil data. Includes a Random Forest analysis of the Lilliput Effect to test climate vs. taphonomy drivers. Results show Neural ODEs outperform generic Transformers (MSE 13.05) in reconstructing deep-time evolutionary history.
bioinformatics deep-learning computational paleoclimate paleontology macroevolution scientific-machine-learning neural-odes chronos-t5 computational-paleobiology
-
Updated
Dec 14, 2025 - Jupyter Notebook