Research

Dissertation

Robust Modeling through Causal Priors and Data Purification in Machine Learning

Ph.D., Electrical and Computer Engineering, UCLA, 2024. Advised by Gregory Pottie.

PureGen: Universal Data Purification for Train-Time Poison Defense via Generative Model Dynamics

PureGen purifies poisoned training data using iterative Langevin dynamics of Energy-Based Models and Denoising Diffusion Probabilistic Models, achieving state-of-the-art poison defense with minimal impact on classifier generalization.

Towards Composable Distributions of Latent Space Augmentations

A composable latent space augmentation framework using VAEs that allows augmentations to be combined through linear transformations, preserving specific augmentation variances and improving geometric interpretability over standard and Conditional VAEs.

Causal Structural Hypothesis Testing and Data Generation Models

CSHTEST and CSVHTEST use non-parametric structural causal knowledge and deep neural networks to perform hypothesis testing on causal models, validated on simulated DAGs, a synthetic pendulum dataset, and real-world medical trauma data.

De-Biasing Generative Models via Counterfactual Methods

De-Biasing Generative Models using Counterfactual Methods

CCGM combines a causal latent space VAE with modifications for causal fidelity to generate de-biased datasets from biased training data, offering fine-grained control over causal structure in both image and tabular data generation.