Paul E. Chang

Machine Learning Engineer

Efficient AI

Paul E. Chang

About

I am a Machine Learning Engineer at DataCrunch, a cloud company where I focus on making AI models more efficient. My work centers on optimizing large language models and foundation models for improved performance and cost-effectiveness.

I also hold a visiting position at the Machine and Human Intelligence group at the University of Helsinki, Finland. Previously, I completed my PhD in Arno Solin's research group at Aalto University, Finland, and collaborated with the Approximate Bayesian Inference Team @ RIKEN AIP.

I'm particularly interested in the intersection of probabilistic modeling and foundation models (GenAI), with a focus on developing efficient AI solutions. I regularly share my findings and insights here.

Publications

2025

Amortized Probabilistic Conditioning for Optimization, Simulation and Inference

Chang, Paul E, Loka, Nasrulloh, Huang, Daolang, Remes, Ulpu, Kaski, Samuel, Acerbi, Luigi.

28th Int. Conf. on Artificial Intelligence & Statistics (AISTATS 2025)

arXiv
2024

Function-space Parameterization of Neural Networks for Sequential Learning

Scannel, Aidan*, Mereu, Riccardo*, Chang, Paul E, Tamir, Ella, Solin, Arno.

International Conference on Learning Representations (ICLR), 2024

arXiv
2023

Memory-based dual Gaussian processes for sequential learning

Chang, Paul E*, Verma, Prakhar*, John, ST, Solin, Arno, and Khan, Mohammad Emtiyaz.

International Conference on Machine Learning (ICML), 2023 (Oral Presentation)

arXiv
2022

Sequential Learning in GPs with Memory and Bayesian Leverage Score

Verma, Prakhar, Chang, Paul E, Solin, Arno, Khan, Mohammad Emtiyaz.

Asian Conference in Machine Learning (ACML) workshop "Continual Lifelong Learning", 2022

OpenReview
2022

Fantasizing with Dual GPs in Bayesian Optimization and Active Learning

Chang, Paul E, Verma, Prakhar, John, ST, Picheny, Victor, Moss, Henry, Solin, Arno.

NeurIPS workshop "Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems", 2022

arXiv
2021

Dual Parameterization of Sparse Variational Gaussian Processes

Adam, Vincent*, Chang, Paul*, Khan, Emtiyaz, Solin, Arno.

NeurIPS, 2021

arXiv