Kaicheng Zhang (Kevin)

Machine Learning PhD Student

University of Edinburgh

About

I am an engaging and highly motivated PhD student specialising in Machine Learning and enrolled under CDT in Mathematical Modelling, Analysis, and Computation (MAC-MIGS CDT), jointly supervised by Desmond Higham and Francesco Tudisco at University of Edinburgh.

My current research focuses on agentic AI for quantitative finance, physics-inspired oscillatory approaches for graph and time-series modelling.

My broader interests also include discrete diffusion language models and AI security, with previous work on the robustness of image classifiers, LLMs, and machine-unlearned diffusion models. During my Bachelor's degree, I also worked on optical neural networks.

News

• Jan 2026 Two papers accepted!
  • Stuart-Landau Oscillatory Graph Neural Network — WWW 2026
  • Are We Measuring Oversmoothing in Graph Neural Networks Correctly? — ICLR 2026
• Dec 2025 The website goes live!
• Dec 2025 Started a research scientist intern position at [stealth startup] on building scalable multi-agent systems, supervised by Michal Valko.
• Nov 2025 Our team 8k3 won the First Place at ICAIF 2025 Crypto Forecasting Competition with Stuart Landau Oscillatory Recursive Neural Network (SLRNN), an extension of our work on graph neural networks (SLGNN). [Paper] [LinkedIn]