Juho Lee

Juho Lee is an Associate Professor at the Kim Jaechul Graduate School of AI, Korea Advanced Institute of Science and Technology (KAIST). Previously, he worked as a research scientist at AITRICS and completed his PhD in Computer Science & Engineering at Pohang University of Science and Technology (POSTECH) under Professor Seungjin Choi, followed by postdoctoral research at University of Oxford with Professor François Caron.

His research focuses on:

• Bayesian deep learning
• Bayesian inference
• Meta learning
• Generative models
• Uncertainty quantification
• Graph representation learning

Professor Lee's work bridges theoretical Bayesian methods with practical deep learning applications. His recent publications demonstrate significant contributions to neural processes, Bayesian optimization, and scalable inference methods. He has developed novel architectures like Set Transformer for permutation-invariant modeling and advanced techniques for uncertainty quantification in deep networks. His research shows a clear trajectory toward making Bayesian principles applicable to large-scale, real-world machine learning problems.

Notable contributions include:

• Set Transformer: A framework for attention-based permutation-invariant neural networks (ICML 2019)
• Bootstrapping neural processes (NeurIPS 2020)
• Deep amortized clustering (NeurIPS 2019 workshop)
• Learning to pool in graph neural networks for extrapolation

Professor Lee actively mentors graduate students and has advised numerous PhD candidates who co-author papers with him across NeurIPS, ICML, and ICLR. His research group SIML@KAIST develops scalable and interpretable machine learning methods with strong theoretical foundations.