CV
Academic CV and selected research experience.
Contact Information
| Name | Fengzhe Zhang |
| Professional Title | Incoming PhD Student |
| fz287@cam.ac.uk | |
| Phone | +44 7421 727 978 |
Professional Summary
Incoming PhD student at the Gatsby Computational Neuroscience Unit working on generative modelling, variational inference, diffusion-based samplers, and probabilistic machine learning.
Experience
-
2024 - 2025 Cambridge, UK
Research Assistant
University of Cambridge
Supervisor: Prof. Jose Miguel Hernandez-Lobato
- Designed diffusion-based generative samplers for molecular energy functions; early experiments halved sampling time versus baseline DDPM.
-
2024 - 2024 Cambridge, UK
MPhil Research Project: Efficient and Unbiased Sampling of Boltzmann Distributions
University of Cambridge
- Reduced number of function evaluations from 100 to 6-25 while preserving effective sample size on synthetic and equivariant n-body systems by integrating Consistency Models with importance sampling.
-
2023 - 2023 London, UK
Undergraduate Research Assistant
Imperial College London
Constrained Optimisation in Variational Autoencoders
- Eliminated hyper-parameter sweeps in beta-VAE by introducing the Constrain-KL algorithm, which enforces an exact KL target and matched NVAE accuracy on CIFAR-10.
-
2022 - 2022 London, UK
Group Project
Imperial College London
Deflation Techniques for Non-linear PDEs
- Implemented sparse Jacobian-Newton deflation in Julia, achieving a 3x speedup and uncovering five distinct solutions to a challenging 3-D nonlinear PDE.
Education
-
2023 - 2024 Cambridge, UK
MPhil
University of Cambridge
Machine Learning and Machine Intelligence
- Probabilistic ML
- Deep Learning
- Computational Statistics
- Speech Recognition
- Advanced ML
- Reinforcement Learning
-
2020 - 2023 London, UK
BSc
Imperial College London
Mathematics
- Applied Probability
- Optimisation
- Stochastic Simulation
- Computational Linear Algebra
- Intro to Statistical Learning
- Statistical Modelling
- Methods for Data Science
Teaching
Awards
-
2023 Dean's List
Imperial College London
Top 10% of Mathematics undergraduates in 2021, 2022, and 2023.
-
2022 Undergraduate Research Bursary
Department of Mathematics, Imperial College London
-
2022 Winton Prize
Imperial College London
Awarded for outstanding second-year group project.
Publications
-
2024 Efficient and Unbiased Sampling of Boltzmann Distributions via Consistency Models
ML for Physical Sciences at NeurIPS 2024
First-author workshop publication on combining consistency models with importance sampling for efficient Boltzmann sampling.
Skills
Projects
-
Efficient and Unbiased Sampling of Boltzmann Distributions
Integrated consistency models with importance sampling to reduce function evaluations from 100 to 6-25 while preserving effective sample size.
-
Constrain-KL for beta-VAE
Developed a constrained optimisation approach to set exact KL targets and avoid beta sweeps in variational autoencoders.