About me
I am Zhongzheng, a fourth-year Ph.D. student at Nanyang Technological University (NTU), Singapore. Apart from NTU, I am also affiliated with I2R, A*STAR (Singapore’s national research agency) and Descartes CNRS@CREATE (a French-Singaporean collaborative research programme). I am fortunate to be supervised by Prof. Xudong Jiang and Dr. Savitha Ramasamy, and was previously co-supervised by Prof. P.N. Suganthan.
My research interests lie broadly in time series analysis, encompassing tasks such as classification and forecasting. I also explore continual learning, online learning, and multimodal learning in time series contexts. Currently, I am investigating the adaptation of time series foundation models for real-world applications.
Apart from research, I like reading 📚, music 🎵 (playing violin 🎻 sometimes) and doing sports 💪🏀. I’m also a fan of MMA 🥊 and video games 🎮 (though I’ve stopped playing them recently!). But honestly, the thing I enjoy and do the most is hanging out with my friends 🥰.
Recent News
- [Oct 2025]: Joining TikTok as a PhD Intern, working on MLLMs in LIVE applications.
- [Sep 2025]: Our paper is accepted by NeurIPS 2025🥳 See you in San Diego!
- [July 2025]: Relocated to the lab of CNRS@CREATE. Hope to meet you at UTown 🙌
- [June 2025]: Attended our programme’s yearly workshop, SINFRA 2025, in Cergy, Paris 🥐
Experience
- Oct 2025 – Jan 2026: Algorithm Engineer Intern, Global Live Team, TikTok, Singapore
- Dec 2020 – May 2021: Research Intern, I2R, A*STAR, Singapore
Education
- 2022-2025: Ph.D. in Interdisciplinary Graduate Programme, NTU.
- 2020-2021: M.Sc. in Electrical and Electronic Engineering, NTU.
- 2016-2020: B.Eng. in Automation, Northeastern University, China.
Selected publications
It’s TIME: Towards the Next Generation of Time Series Forecasting Benchmarks (arxiv, 2026) [Code] [Leaderboard]
- Multi-Scale Finetuning for Encoder-based Time Series Foundation Models (NeurIPS, 2025) [Code]
- CrisisTS: Coupling Social Media Textual Data and Meteorological Time Series for Urgency Classification (ACL, 2025)
- Class-incremental learning for time series: Benchmark and evaluation (SIGKDD, 2024) [Code]
- Class-incremental learning on multivariate time series via shape-aligned temporal distillation (ICASSP, 2023)
Awards
- NTU Premium Research Scholarship
- Outstanding Graduates in Northeastern University
- Northeastern University Outstanding Student Scholarship
