FrameCite Video Knowledge Agent
A video knowledge agent that searches spoken explanations and visual descriptions, inspects original frames, and returns claims with playable timestamp citations.
Read more
Portfolio
SPREAD: Sampling-based Pareto front Refinement via Efficient Adaptive Diffusion
SPREAD learns a conditional diffusion model over decision variables and then iteratively refines sampled candidates during reverse diffusion to move them toward Pareto-optimal regions.
Read more
MOREL: Enhancing Adversarial Robustness through Multi-Objective Representation Learning
MOREL is a method to improve deep neural network robustness against adversarial attacks using a multi-objective optimization approach.
Read more
Pareto-Splat: A Detailed Guide to Quality-Efficiency Trade-offs in 3D Gaussian Splatting
A reproducible research pipeline for studying quality, speed, and model-size trade-offs in 3D Gaussian Splatting.
Read more