Investigating Evolutionary Radiation of Birds through Deep Learning Morphological Exploration

Peter Bentley (primary)
Computer Science
Ryan Felice (secondary)
Cell and Developmental Biology


Birds- the only living dinosaurs- are incredibly diverse with over 10,000 species. Yet extinct dinosaurs were even more diverse. What are the constraints that limit anatomical evolution in birds? While we can’t turn back time and run the experiment of avian evolution over again, we can use artificial intelligence to investigate the underlying factors that influence the evolution of novelty. Using 3D shape analysis and deep learning approaches, we will explore the patterns of correlation between the brain and skull across Dinosauria. By using AI to generate a landscape of novel morphologies, we will uncover hidden patterns of evolutionary constraint.


1. Felice, R.N., and Goswami, A. 2018. Developmental Origins of Mosaic Evolution in the Avian Cranium. Proceedings of the National Academy of Sciences, 115(3), 555-560. doi:10.1073/pnas.1716437115
Felice, R.N., Watanabe, A., Cuff, A.R., Hanson, M., Bhullar, B-A., Rayfield, E.R., Witmer, L. M., Norell, M.A., Goswami, A. 2020. 2. Decelerated dinosaur evolution at the origin of birds. PLOS Biology, 18(8), e3000801.
3. Bardua, C., Felice, R.N., Watanabe, A., Fabre, A-C., and Goswami, A. 2019. A practical guide to surface sliding semi-landmarks in morphometric analyses. Integrative Organismal Biology, obz016
4. Zhijie Wu, Xiang Wang, Di Lin, Dani Lischinski, Daniel Cohen-Or, Hui Huang. 2019. SAGNet: structure-aware generative network for 3D-shape modeling. ACM Transactions on Graphics. July 2019.
5. Pugh Justin K., Soros Lisa B., Stanley Kenneth O. 2016. Quality Diversity: A New Frontier for Evolutionary Computation. Frontiers in Robotics and AI v3. doi: 10.3389/frobt.2016.00040

Genes, development and STEM* approaches to biology
Area of Biology
Techniques & Approaches
BioinformaticsEngineeringImage ProcessingMathematics / StatisticsSimulation / Modelling