Research
Computational Neuroimaging Science Lab
Graduate researcher under Dr. Kilian Pohl
Stanford University
Spring 2022 - present
Designed a training technique for deep learning models to extract unobserved features from their own latent-space representations
Applied method to train a model to learn a neurocognitive disorder classification task while learning to track brain age evolution
Succeeded in representing brain age through self-learned latent-space curves without dropping classification performance
Obtained 0.96 R 2 -score on extracting patients’ ages from their MRI scans while retaining the model’s original task performance
Stanford Machine Learning Group
Graduate researcher under Dr. Andrew Ng
Stanford University
Spring 2022 - present
Designing a pre-training automated method to remove mislabeled training samples on supervised learning image classification / detection tasks through embedding-space properties and gradient analysis
Achieved SOTA mislabel detection results on CIFAR10 and CIFAR100
Intended application will attempt to remove corrupted auto-labeled satellite images identifying wind farms and oil platforms
GT-LIGO Research Group
Undergraduate researcher under Dr. Laura Cadonati
Georgia Tech
Fall 2019 - Spring 2021
Researched computational models for glitch detection and classification on gravitational wave data
Work as a member of the LSC - the Laser-Interferometer Gravitational Waves Observatory Scientific Collaboration - on the computational efforts to improve data quality
Developed a statistical study on the efficiency of data quality channels in identifying glitches in gravitational wave data
Developed a Machine Learning model to identify and classify glitches with comparable results to quality channels