Rendered and extracted features from 3D scenes that include transparent objects using TensoRF and feature field distillation.

Inspired by neural style transfer in vision, I applied these methods to speaking-to-singing style transfer.

Analyzed biases in binary VQA models by evaluating the accuracy and consistency of VQA model on modified question and image inputs.

Examined shape and color biases in co-modulated GANs.

Explored how constraints and hierarchical planning influence diversity and creativity of GPT-3 generations. Later, resulted in a publication.

Created an optimization model to optimize for the best class schedule.

Added a new layer of abstract qualities to the model and analyzed moral behaviors of participants in East Asian countries.

Implemented single scale and single scale oriented paint.