Daolang Huang

"Amortize everything..."

About

Daolang is a doctoral student co-supervised by Samuel Kaski (Aalto University) and Luigi Acerbi (University of Helsinki) since July 2022, affiliated with the Probabilistic Machine Learning group at Aalto University and the Machine and Human Intelligence group at the University of Helsinki. He is fully funded by Finnish Center for Artificial Intelligence (FCAI). He received his master’s degree at Aalto University, with a major in Machine learning, data science and artificial intelligence (Macadamia). Daolang’s work focuses on amortized inference with applications to various Bayesian inference tasks. In his spare time, he is also an electronic music producer, mainly producing house music.

Click here to see my latest CV (last update: February 2025).

Research interests: amortized inference, simulation-based inference, Bayesian optimization, Bayesian experimental design, meta-learning, neural processes.

News

2025.01: Our paper “PABBO: Preferential Amortized Black-Box Optimization” has been accepted by ICLR 2025 as Spotlight (Top 5.1%)!
2025.01: Two of our papers have been accepted by AISTATS 2025!
- Amortized Probabilistic Conditioning for Optimization, Simulation and Inference
- Cost-aware Simulation-based Inference
2024.10: I have been selected as one of the top 3 nominees for the AI Researcher of the Year by AI Finland!
2024.10: Participated in the poster session and served as the photographer at Finland AI Day x Nordic AI Meet 2024.
2024.09: Our paper “Amortized Bayesian experimental design for decision-making” has been accepted by NeurIPS 2024!
2023.12: Had a great week at NeurIPS in New Orleans.
2023.11: Gave a talk about conditional neural processes at the Mathematical perspective on machine learning seminar at the University of Helsinki. [Slides]
2023.11: Gave a talk about our RCNP paper at Finland AI day 2023.
2023.09: Attended the Ellis RobustML Workshop.
2023.09: Two of our papers have been accepted by NeurIPS 2023!
- Learning robust statistics for simulation-based inference under model misspecification
- Practical Equivariances via Relational Conditional Neural Processes
2023.08: Had a fantastic week attending the Ellis Doctoral Symposium (EDS) 2023. I presented our work “Learning robust statistics for simulation-based inference under model misspecification” and also gave an introduction about the FCAI amortized inference team.
2023.07: Our paper “Augmenting Bayesian Optimization with Preference based Expert Feedback” has been accepted by ICML 2023 workshop The Many Facets of Preference-based Learning.