I am a third-year Ph.D. student in Computer Science and Technology at the University of Electronic Science and Technology of China (UESTC), where I have been pursuing an integrated master's–Ph.D. program since September 2022, supervised by Prof. Shi Gu. I received my bachelor's degree from UESTC in June 2022.

Before that, I was a member of the UESTC ACM-ICPC team, where I enjoyed both the challenge of solving interesting algorithmic problems and the camaraderie of working with teammates.

My research interests include NeuroAI, computational neuroscience, and spiking neural networks.

Curious and enthusiastic, I strive to remain coachable and intellectually honest, with a growth mindset. As agentic AI lowers execution barriers, I place greater value on developing an independent perspective on where meaningful progress is possible, cultivating a taste for fundamental and consequential questions, and sharpening my research intuition through experimentation and reflection.

I am interested in research internship opportunities in reinforcement learning, continual learning, and embodied intelligence. Feel free to get in touch!

News

  • 2026: Our work on task-structured modularity was published in Nature Machine Intelligence.
  • 2025: Our work on temporal flexibility in spiking neural networks appeared at ICLR 2025.
  • 2024: Our work on Spiking Token Mixer appeared at NeurIPS 2024.

Publications

Figure 1: task-structured modularity in artificial neural networks
Nature Machine Intelligence · 2026

Task-structured modularity emerges in artificial networks and aligns with brain architecture

Yuhang Wu, Shikuang Deng, Kangrui Du, Marcelo G. Mattar, Yifan Wu, Dani S. Bassett, Huajin Tang, Gang Pan, Shi Gu

[Paper]Computational neuroscience
  • Why do neural networks become modular? Comparing single-task, multitask, and incremental training in recurrent networks, we show that multitask learning promotes modularity under limited capacity.
  • Growing connections during learning strengthens modularity. Combining task-derived structure with spatial constraints improves the fit to human brain networks, highlighting the role of task demands in shaping brain architecture.
Figure 1: Mixed Time-step Training workflow
ICLR 2025

Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness

Kangrui Du*, Yuhang Wu*, Shikuang Deng, Shi Gu

[Paper]Spiking neural networks
  • We identify temporal inflexibility in standard SNN training, which limits fully event-driven deployment and dynamic energy–performance trade-offs.
  • We propose Mixed Time-step Training (MTT), randomly assigning time steps to network stages during each training iteration to improve temporal flexibility.
  • We validate MTT on GPUs, neuromorphic chips, and an event-driven simulator and, to our knowledge, report the first large-model results on fully event-driven platforms (VGGSNN on CIFAR10-DVS).
Figure 1: network architecture and spike-delay issues in asynchronous hardware
NeurIPS 2024

Spiking Token Mixer: An event-driven friendly Former structure for spiking neural networks

Shikuang Deng*, Yuhang Wu*, Kangrui Du, Shi Gu

[Paper]Spiking neural networks
  • We propose STMixer, a spiking architecture built from operations supported by asynchronous hardware, addressing the deployment limitations of spike-based attention and pooling.
  • STMixer achieves performance comparable to spiking transformers at very low time steps in synchronous settings, while supporting event-driven execution.

Honors and Awards

IEEEXtreme Programming Competition

  • 2022.10 — 31st place worldwide, IEEEXtreme 16.0.

International Collegiate Programming Contest (ICPC)

  • 2021.11 — Silver Medal, Asia Jinan Regional Contest.
  • 2021.12 — Bronze Medal, Asia Nanjing Regional Contest.

China Collegiate Programming Contest (CCPC)

  • 2021.11 — Bronze Medal, Harbin Contest.
  • 2020.11 — Bronze Medal, Changchun Contest.

Education

  • 2022.09 – Present — Integrated M.S.–Ph.D., Computer Science and Technology, UESTC.
  • 2018.09 – 2022.06 — Bachelor's degree, Computer Science and Technology, UESTC.

Internships

  • 2025.06 – 2025.09 — Research Intern with the Darwin team at Zhejiang University's State Key Laboratory of Brain-Machine Intelligence, contributing to brain simulation for the Wukong (Darwin Monkey) project.
  • 2021.01 – 2021.07 — Autonomous Driving Perception Algorithm Intern at SenseTime, mentored by Hui Zhou.