ADELE MYERS
Adele Myers CV
I develop interpretable AI tools to quantify complex shape changes in the human body. I am specifically interested in using these tools to study shape changes that happen in the female brain during hormonal events such as menstruation, pregnancy, and menopause. Shape analysis tools are immensely powerful in this area. While complex 3D shape changes in the brain are easily noticed by the human eye, the specific characteristics of the shape change cannot be quantified by the human eye. My work seeks to quantify subtle changes, giving doctors interpretable information about the nature of the shape change and informing diagnoses and research.
![](https://static.wixstatic.com/media/c9b0ea_1a842c035e264c8fb03db0580299bfca~mv2.jpg/v1/crop/x_110,y_1,w_281,h_471,q_80,enc_auto/c9b0ea_1a842c035e264c8fb03db0580299bfca~mv2.jpg)
Accessible Talks/Interviews
Unlocking Mysteries of the Female Brain: Santa Barbara Public Talk
Machine Learning Street Talk Interview @ NeurIPS 2022
Publications
2024
Adele Myers, Nina Miolane (2024). On Accuracy and Speed of Geodesic Regression: Do Geometric Priors
Improve Learning on Small Datasets? In Proceedings of the CVPR Conference Workshop: Learning With
Limited Labelled Data for Image and Video Understanding (L3D-IVU). Institute of Electrical and Electronics
Engineers Inc.
2023
Adele Myers, Caitlin Taylor, Emily Jacobs, Nina Miolane (2023). Geodesic Regression Characterizes 3D
Shape Changes in the Female Brain During Menstruation. In Proceedings of the ICCV Conference Workshop:
Computer Vision for Automated Medical Diagnosis. Institute of Electrical and Electronics Engineers Inc.
2022
Adele Myers, Nina Miolane (2022). Regression-based elastic metric learning on shape spaces of curves. NeurIPS workshop: Learning meaningful representations of life.
This work aims to 1) improve regression analyses for trajectories of complex shapes lying on high dimensional shape space manifolds and 2) quantify bending and stretching of cells as they change over time. More specifically, we learn the elastic metric parameters which bring a shape trajectory closest to a geodesic on the manifold of discrete curves.
Myers et. al (2022) ”ICLR 2022 Challenge for Computational Geometry Topology: Design and Results”. Proceedings of Machine Learning Research. Volume “Topological, Algebraic, and Geometric Learning”. Accepted, In press.
Provides a critical overview of results from the 2022 ICLR Geomstats hackathon, which I co-organized.
Miolane et. al. (2022). geomstats/geomstats: Geomstats v2.5.0 (2.5.0).Zenodo. https://doi.org/10.5281/zenodo.6478729
Geomstats software release. Geomstats is an open-sourced software developed by my lab, which uses differential geometry to analyze data that lie on manifolds. For authorship on this version, I wrote several differential geometry tutorial notebooks for users who need additional mathematical background before they can utilize the software.
2018
Adele Myers, Greg Spriggs (2018), Water Entrainment in Nuclear Detonations: Discovery and Investigation, LLNL Report, LLNL-TR-758735
In this technical report, I discuss the discovery I made at Lawrence Livermore National Laboratory. After inspection of historic nuclear test videos, I discovered visual evidence of water entrainment in nuclear shockwaves over water. My further work demonstrated that historic yield estimates, which neglected this phenomenon, underestimated the true yield of atmospheric detonations by ~20 percent.