Projects

Bayesian Optimization with Preference Exploration by Monotonic Neural Network Ensemble

Many real-world black-box optimization problems have multiple conflicting objectives. Rather than attempting to approximate the entire set of Pareto-optimal solutions, interactive preference learning, i.e., optimization with a decision maker in the loop, allows to focus the search on the most relevant subset. However, few previous studies have exploited the fact that utility functions are usually monotonic. In this paper, we address the Bayesian Optimization with Preference Exploration (BOPE) problem and propose using a neural network ensemble as a utility surrogate model. This approach naturally integrates monotonicity and allows to learn the decision maker’s
preferences from pairwise comparisons. Our experiments demonstrate that the proposed method outperforms state-of-the-art approaches and exhibits robustness to noise in utility evaluations. An ablation study highlights the critical role of monotonicity in enhancing performance.

Link: https://arxiv.org/abs/2501.18792

Bayesian Optimization with Known Output Boundary

In Bayesian Optimization (BO), it is assumed that the objective function is a black-box function but in some
situations, extra information about the output value (such as an upper or lower bound, or even the exact optimal output) is available. Ignoring the boundary information of the objective function may hinder the efficiency of BO. In this project, we aim at designing a new BO framework that takes the boundary information into consideration and improving the performance of BO.

(published in Transactions on Machine Learning Research)

Link: https://arxiv.org/abs/2411.04744

Machine-Learning-Guided Directed Evolution for Multi Objective Protein Engineering

This is a group project collaborated with ZenithAI. The rise of machine learning and large datasets has led to replacing expensive wet lab experiments with surrogate wet lab prediction models enabling much faster and cheaper sequence screening and optimization. This project utilises variational autoencoder and Bayesian optimization to determine the protein sequence that has good performance in two properties.

Education

University of Warwick (2021-present): Mathematics of Systems CDT

University of Oxford (2020 - 2021): MSc Mathematical and Computational Finance

University of Liverpool (2018-2020): BSc Mathematics with Finance

Xi’an Jiaotong-liverpool University (2016-2018): B.Econ. In Financial Mathematics

Skills

Python (main language), C++, Julia and MATLAB

Teaching

Senior teaching assistant in Analytics in Practice

Senior teaching assistant in Business Analytics