Dmitry Kangin
I am a Research Fellow at the Centre for AI Fundamentals, working with Professor Samuel Kaski.
My research focuses on machine learning for science and human‑in‑the‑loop decision making, with particular emphasis on aligning generative models, such as diffusion models, with human goals.
I am working on extending human-in-the-loop generative machine learning methods beyond standard domains such as image generation to the domains such as robotics, material discovery, and protein design. I also work on the problems of transfer learning for biomedical data and experimental design. By integrating theoretical insights with empirical validation, my goal is to develop new algorithms that address the challenges of co‑design in machine learning for science.
To tackle these problems, I draw on methods from control theory, reinforcement learning, stochastic differential equations, and Bayesian experimental design. Ultimately, I aim to build approaches that support meaningful collaboration between machine‑learning models and human experts—empowering, rather than replacing, scientists in the pursuit of new discoveries.
