New Preprint: “Unified Semantic Bridge Model for Unpaired Image-to-Image Translation” We propose the Unified Semantic Bridge Model (USBM), a fully self-supervised framework that learns diffusion bridges from a shared semantic latent space to each domain, enabling geometry-aware translation without cross-domain supervision.
New Preprint: “Merlin: A Computed Tomography Vision–Language Foundation Model and Dataset”. Extremely happy to have contributed to this major project, which presents the first vision foundation model for 3D CT imaging and is currently being revised for final publication.
Postdoc: I am extremely happy to share that I will be joining AIMI @ Stanford as a postdoctoral researcher in 2024!
New Preprint: “Prior Mismatch and Adaptation in PnP-ADMM with a Nonconvex Convergence Analysis”. We present nonconvex convergence results under prior mismatch for PnP-ADMM. This work has been accepted to ICML 2024!
New Preprint: “DOLCE: A Model-Based Probabilistic Diffusion Framework for Limited-Angle CT Reconstruction”. We present DOLCE, a new deep modelbased framework for LACT that uses a conditional diffusion model as an image prior has been accepted by ICCV 2023!
MERL Research: Extremely happy to share that I will join MERL as a summer intern in 2023!
In Press: Our paper “Online Deep Equilibrium Learning for Regularization by Denoising” has been accepted at NeurIPS 2022! We proposed a new online deep equilibrium learning framework for data-intensive imaging modalities based on implicit neural network.
LLNL Research: Extremely happy to share that I will join LLNL as a summer intern in 2022!
TCI Accepted: Our paper “CoIL: Coordinate-based Internal Learning for Imaging Inverse Problems” has been on the IEEE Transactions on Computational Imaging (TCI)!! It extends coordinate-based representation to computational imaging. Here is a short video summary.
NeurIPS Accepted: Our paper “Recovery Analysis for Plug-and-Play Priors using the Restricted Eigenvalue Condition” has been accepted at NeurIPS 2021! It establishes theoretical recovery guarantees for PnP/RED in compressive sensing.
LANL Research: Extremely happy to share that I will join LANL as a summer intern in 2021!
TCI Accepted: Our paper “SGD-Net: Efficient Model-Based Deep Learning with Theoretical Guarantees” has been accepted on the IEEE Transactions on Computational Imaging (TCI)!
ICLR Accepted: Our paper “Async-RED: A Provably Convergent Asynchronous Block Parallel Stochastic Method using Deep Denoising Priors” has been accepted to International Coference on Learning Represetations (ICLR) as spotlight.
JSTSP Accepted: Our paper “SIMBA: Scalable Inversion in Optical Tomography using Deep Denoising Priors” has been published on the IEEE Journal of Selected Topics in Signal Processing (JSTSP)!
JSTSP Accepted: Our paper “RARE: image reconstruction using deep priors learned without groundtruth” has been published on the IEEE Journal of Selected Topics in Signal Processing (JSTSP)!