Doctoral Researcher
ELLIS Institute Finland · Department of Computer Science, Aalto University
I'm a second-year PhD student at Aalto University and the ELLIS Institute Finland, supervised by Prof. Samuel Kaski. My research focuses on meta-learning, amortized inference, and Bayesian optimization.
Before the PhD, I completed my Master's in Computer Science at Aalto University, and Bachelor's in Computer Science and Engineering at Harbin Institute of Technology, Shenzhen. Outside research I enjoy bouldering 🧗 and doodling 🖊️.
A dimension-agnostic transformer that replaces the surrogate-plus-acquisition stack of (multi-objective) Bayesian optimization with a single forward pass; 50–1000× faster proposal time at matched Pareto quality.
Enables constrained Bayesian experimental design by combining offline pre-training of an amortized policy and a posterior network with online multi-step lookahead planning using scenario trees.
A transformer neural process trained with reinforcement learning that fully amortizes preferential Bayesian optimization: meta-learning both the surrogate and the acquisition function. Several orders of magnitude faster than Gaussian-process baselines, and often more accurate.
Workshops & Preprints
Generalized additive models — including their recent neural-network variants — exhibit multiple forms of nonidentifiability that complicate interpretation. We argue for restraint when claiming such models as interpretable or suitable for safety-critical applications.