Jiajun He

Hello! I’m Jiajun He (何佳峻), a second-year PhD student in the Machine Learning Group at the University of Cambridge, supervised by Prof. José Miguel Hernández-Lobato. My PhD is fully funded by the Harding Distinguished Postgraduate Scholarship.

Research

My research sits at the intersection of probabilistic inference and generative modelling, with a particular emphasis on stochastic dynamics and path-space methods. I develop algorithms that use generative models to make probabilistic inference and sampling more efficient, while drawing on probabilistic inference principles to make generative models more controllable and efficient.

In recent work, I have developed:

• Non-equilibrium (path-space) approaches for free energy estimation
• Non-equilibrium (path-space) parallel tempering — replica exchange — for accelerated sampling
• Path-space inference for efficient and flexible control of generative models

I am also interested in information-theoretic approaches to compact representations and compression, as well as generative methods for scientific discovery and design.

Background

Previously, I completed an MPhil in Machine Learning and Machine Intelligence at the University of Cambridge, graduating with distinction as first in the cohort. For my MPhil dissertation, I worked with Prof. José Miguel Hernández-Lobato and Gergely Flamich on data compression. Before that, I obtained an MSc in Bioinformatics at the University of Copenhagen, working with Prof. Thomas Wim Hamelryck, and a BSc in Biological Science at Tsinghua University, supervised by Prof. Xuerui Yang.

Beyond research

I am also a landscape photographer — you can explore my work on Instagram.