Wanglong Lu

// multimodal intelligence · generative vision · production AI

Applied multimodal intelligence for generative vision and production AI systems.

I develop multimodal and generative AI methods for high-resolution visual computing, efficient model adaptation, and reliable real-world deployment.

At Nasdaq, I build production machine learning systems for financial applications. My research spans generative image editing, restoration, super-resolution, multimodal learning, and parameter-efficient fine-tuning.

Multimodal Intelligence Generative AI High-Resolution Vision Efficient Adaptation Production ML Systems

// experience

Nasdaq

Senior Data Scientist, AI/Analytics · 2025 - Present

Building and applying machine learning systems for financial AI products in production settings.

Nasdaq Verafin

AI Research Intern · 2024 - 2025

Worked on image verification and delivered research progress presentations to cross-functional audiences.

Research Collaboration

Memorial University of Newfoundland · Wenzhou University

Collaborating with and mentoring Ph.D. and Master's students with Prof. Xianta Jiang and Prof. Hanli Zhao.

// education

Memorial University of Newfoundland

Ph.D. in Computer Science · 2025

Wenzhou University

M.Sc. in Computer Software and Theory · 2021

Communication University of Zhejiang

B.Sc. in Digital Media Technology · 2018

// research overview

My work develops applied multimodal intelligence across generative image editing, image restoration, super-resolution, efficient adaptation, assistive vision, and production machine learning. The unifying goal is to build models that preserve structure, remain controllable at high resolution, adapt efficiently, and operate reliably under real system constraints.

// featured systems

view GitHub →

UltraDiffEdit

Tuning-free real-image editing with pretrained latent diffusion models, designed to preserve unedited content while scaling generation to ultra-high resolutions.

8K EDITINGTUNING-FREETNNLS

Details Code

AdaMSS

An adaptive multi-subspace approach for parameter-efficient fine-tuning, translated from a NeurIPS contribution into the Hugging Face PEFT ecosystem.

NEURIPSPEFTOPEN SOURCE

Details Hugging Face

PRODUCTION_ML robust models / evaluation / reliable deployment financial AI systems @ Nasdaq

Production Financial AI

Building and applying machine learning systems for financial AI products, with emphasis on robustness, evaluation, maintainability, and production constraints.

APPLIED MLFINANCIAL AIRELIABILITY

Experience CV

// selected publications

view all publications →

TNNLS

Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing

Wanglong Lu, Lingming Su, Kaijie Shi, Minglun Gong, Xiaogang Jin, Hanli Zhao, Xianta Jiang

IEEE Transactions on Neural Networks and Learning Systems, 1-15, 2026

UltraDiffEdit extends off-the-shelf latent diffusion models to tuning-free real-image editing up to 8K resolution.

First Author8K EditingTuning-FreeCode Available

Details Paper Code

TVCG

FACEMUG: A Multimodal Generative and Fusion Framework for Local Facial Editing

Wanglong Lu, Jikai Wang, Xiaogang Jin, Xianta Jiang, and Hanli Zhao∗

IEEE Transactions on Visualization and Computer Graphics, 2024

A multimodal generative framework for globally consistent local facial editing with fine-grained semantic control.

First AuthorMultimodal EditingIEEE TransactionsVideo

Details Paper Video

NeurIPS

AdaMSS: Adaptive Multi-Subspace Approach for Parameter-Efficient Fine-Tuning

Jingjing Zheng, Wanglong Lu, Yiming Dong, Chaojie Ji, Yankai Cao, Zhouchen Lin

NeurIPS 2025, 2025

An adaptive multi-subspace parameter-efficient fine-tuning method for expressive incremental updates, integrated into Hugging Face PEFT.

NeurIPSHugging Face PEFTEfficient AdaptationCode Available

Details Paper Code PEFT

TCSVT

UHDRes: Ultra-High-Definition Image Restoration via Dual-Domain Decoupled Spectral Modulation

Shihao Zhao, Wanglong Lu, Binhao Wang, Tao Wang, Kaihao Zhang, Hanli Zhao

IEEE Transactions on Circuits and Systems for Video Technology, 2026

A lightweight UHD restoration framework using dual-domain decoupled spectral modulation with only 400K parameters.

IEEE Transactions400K ParametersUHD RestorationCode Available

Details Paper Code

Visual style prompt learning using diffusion models for blind face restoration

Wanglong Lu, Jikai Wang, Tao Wang, Kaihao Zhang, Xianta Jiang, Hanli Zhao*

Pattern Recognition, 2025

A diffusion-guided blind face restoration method that learns visual style prompts in pretrained generative latent space.

First AuthorDiffusion RestorationPattern RecognitionCode Available

Details Paper Code

TNSRE

Toward Biosignals-Free Autonomous Prosthetic Hand Control via Imitation Learning

Kaijie Shi, Wanglong Lu, Hanli Zhao, Vinicius Prado da Fonseca, Ting Zou, Xianta Jiang

IEEE Transactions on Neural Systems and Rehabilitation Engineering 33, 3544-3554, 2025

A biosignals-free autonomous prosthetic grasping framework that learns hand control through simulation-driven imitation learning.

IEEE TransactionsAssistive VisionImitation LearningProject Page

Details Paper Project

// open source, impact & talks

Research to Open Source

AdaMSS · Hugging Face PEFT

A parameter-efficient fine-tuning contribution integrated into a widely used open-source machine learning ecosystem.

Reproducible Vision Systems

UltraDiffEdit · UHDRes · EchoSR

Public implementations and project resources that connect research contributions with usable technical systems.

Technical Communication

IEEE invited talk · Guest lecture · CBC coverage

Communicating generative AI and assistive vision research across academic, industry, and public audiences.

// current news

Now I am currently working at Nasdaq as a Senior Data Scientist.
Jul. 2026 Our paper "Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing" has been published in IEEE Transactions on Neural Networks and Learning Systems, 1-15. Source code is available on GitHub.
Jul. 2026 Our paper "UHDRes: Ultra-High-Definition Image Restoration via Dual-Domain Decoupled Spectral Modulation" has been published in IEEE Transactions on Circuits and Systems for Video Technology. Congratulations to Shihao. Source code is available on GitHub.
Jun. 2026 Our paper "Simulation-Driven Imitation Learning for Biosignals-Free Shared-Autonomy Prosthetic Grasping" was released as an arXiv preprint, arXiv:2606.07389. Congratulations to Kaijie.
May 2026 Our paper "EchoSR: Efficient Context Harnessing for Lightweight Image Super-Resolution" has been published in Information Fusion, 104471. Cheers to Binhao. Source code is available on GitHub.
Apr. 2026 AdaMSS has been integrated into the Hugging Face PEFT package. GitHub
Mar. 2026 I was invited as a guest speaker at the IEEE Newfoundland and Labrador Section's 2025-2026 AGM and gave a talk titled "Generative Models for Semantic Image Editing: Multimodal Approaches" at Memorial University.

More News (2025)

On Dec. 4, 2025, our paper "AdaMSS: Adaptive Multi-Subspace Approach for Parameter-Efficient Fine-Tuning" has been presented at NeurIPS 2025. GitHub
Our paper "Toward Biosignals-Free Autonomous Prosthetic Hand Control via Imitation Learning" has been published in IEEE Transactions on Neural Systems and Rehabilitation Engineering, 33:3544-3554. Congratulations to Kaijie. This project was also interviewed by the CBC. Check the main page for more information.
On April 26, 2025, our paper "Degradation-Aware Frequency-Separated Transformer for Blind Super-Resolution" was successfully published at the Computational Visual Media Conference. Congratulations to Binhao Wang.
On March 6, 2025, our paper "TextDoctor: Unified Document Image Inpainting via Patch Pyramid Diffusion Models" was released.
On Feb. 15, 2025, our paper "Real-time dual-eye collaborative eyeblink detection with contrastive learning" has been successfully published in Pattern Recognition. Congratulations to Yu Wang.
On Feb. 05, 2025, I served as a guest lecturer for Bio-Inspired Robotics (ENGI 9986), delivering a 150-minute course on "Deep Generative Models and Applications" to 10 graduate students. I am grateful to Dr. Ting Zou for the invitation.
On Jan. 31, 2025, I completed my internship at Nasdaq Verafin and delivered a presentation on our recent progress in image verification to an audience of 39 attendees.

More News (2024)

On Dec. 29, 2024, our paper Visual Style Prompt Learning Using Diffusion Models for Blind Face Restoration was published at Pattern Recognition. Page
On November 28, 2024, our paper has been published in Neurocomputing after a three-year review process. It was the starting point of my research and was finally accepted on the morning of my thesis defense.
On November 21, 2024, I successfully defended my Ph.D. thesis. I am deeply grateful to my supervisors, Dr. Xianta Jiang, Dr. Hanli Zhao, and Dr. Yuanzhu Chen, and to my committee members, collaborators, and peers at Memorial University of Newfoundland, Wenzhou University, and Nasdaq.
On Nov. 14, 2024, I gave an oral presentation at the 33rd NECEC conference for our recent document image restoration algorithm, with an audience of 30 attendees.
On Sep. 23, 2024, I gave a talk to share my experience in university and postgraduate study life at Digital Media Technology, Communication University of Zhejiang, with an audience of 120 attendees.
On August 30, 2024, I gave a talk on our recent work in text document restoration at Nasdaq Verafin, with an audience of 45 attendees.
On July 11, 2024, our paper titled "Handling The Non-Smooth Challenge in Tensor SVD: A Multi-Objective Tensor Recovery Framework" was accepted at ECCV.

// academic service

Invited Reviewer

2026

Invited Reviewer