TNLearn: An Open Source Python Package for Task-based Neurons
arXiv preprint arXiv:2609.27564, 2026
COMPUTER SCIENCE & ENGINEERING
Hanyu Pei
Researching machine learning
and AI security.
I am a Ph.D. student at the University of Louisville, advised by Prof. Zeyan Liu. My research interests lie in machine learning and AI security.
01 / ABOUT
A little background.
I am a Ph.D. student in the Department of Computer Science and Engineering at the University of Louisville. My interests include reinforcement learning, coding agents, AI security, machine unlearning, and neural architectures.
I obtained my M.Sc. in Mathematics from the Chinese University of Hong Kong (CUHK) in 2023, and my B.Sc. in Statistics from Wuhan University of Technology in 2022. I was honored to work as a Research Assistant with Prof. Fenglei Fan (CityU) and Prof. Ben Dai (CUHK).
02 / UPDATES
News.
A Survey on Medical Image Compression: From Traditional to Learning-Based Approaches is published online in IEEE Transactions on Radiation and Plasma Medical Sciences!
My first paper as first author since starting my Ph.D., LymphNode, has been accepted to DSN 2026!We propose a plug-and-play access-control module to enhance the security of deep neural networks (DNNs).
COLI: A Hierarchical Efficient Compressor for Large Images is published online in IEEE Transactions on Radiation and Plasma Medical Sciences!
I joined Prof. Zeyan Liu's team at the University of Louisville.
My first, first-author paper, NeuronSeek, is now available on arXiv!We introduce a novel tensor-decomposition framework for discovering task-driven neurons, with improved stability and super-super-expressivity.
7 papers
-
-
presto: Efficient, Training-free, and Open-world Object Placement via Imaginary Search
International Conference on Neural Information Processing (ICONIP), 2026 (accepted)
-
LymphNode: A Plug-and-Play Access Control Method for Deep Neural Networks
IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), 2026
-
A Survey on Medical Image Compression: From Traditional to Learning-Based Approaches
IEEE Transactions on Radiation and Plasma Medical Sciences, 2026
-
ShallowJail: Steering Jailbreaks against Large Language Models
arXiv preprint arXiv:2602.07107, 2026
-
COLI: A Hierarchical Efficient Compressor for Large Images
IEEE Transactions on Radiation and Plasma Medical Sciences, 2026
-
NeuronSeek: On Stability and Expressivity of Task-driven Neurons
arXiv preprint arXiv:2506.15715, 2025
* Denotes equal contribution. My name is shown in bold.
04 / EDUCATION
Education
University of Louisville
Ph.D. in Computer Science and Engineering
The Chinese University of Hong Kong
M.Sc. in Mathematics
Wuhan University of Technology
B.Sc. in Statistics
05 / EXPERIENCE
Experience
University of Louisville
Research Assistant · Louisville, Kentucky
City University of Hong Kong
Research Assistant · Remote
Developed symbolic regression methods for task-based neuron discovery.
The Chinese University of Hong Kong
Research Assistant · Hong Kong
Applied statistical inference to medical image analysis.
China UnionPay
Data Analysis Intern · Wuhan, China