Insung Kong
Insung Kong
Address
UNIST, Building 108, Room 301-2
50, UNIST-gil, Ulju-gun, Ulsan, Republic of Korea
kong[at]unist.ac.kr
Embedded Files
About me
About me
Hi! I am a tenure-track Assistant Professor in the Department of Mathematical Sciences at Ulsan National Institute of Science and Technology (UNIST).
Before joining UNIST, I was a postdoctoral researcher at University of Twente, working with Johannes Schmidt-Hieber. Before that, I completed my Ph.D. studies at the Seoul National University under the supervision of Yongdai Kim in 2024.
My research focuses on the mathematical foundations of artificial intelligence, through the lens of statistical learning theory. I am particularly interested in understanding why modern AI methods, such as deep learning and large language models, work so effectively, and in clarifying their theoretical properties, limitations, and underlying principles. Motivated by these theoretical insights, I also develop new machine learning methodologies that are statistically grounded and designed to overcome limitations of existing approaches. Another direction of my research is to use mathematical theory to provide guiding principles for future AI technologies, including biologically inspired neural networks. Through this work, I aim to deepen our theoretical understanding of artificial intelligence and contribute to principled methodologies for next-generation artificial intelligence.
Employment
Employment
2026 - UNIST, Ulsan, Republic of Korea
Assistant Professor
2024 - 2026 University of Twente, Enschede, the Netherlands
Postdoctoral Researcher
Education
Education
2018 - 2024 Seoul National University, Seoul, Republic of Korea
Doctor of Philsosphy, Statistics
Thesis: Posterior Concentration Rates of Bayesian Neural Networks [pdf]
2013-2018 Seoul National University, Seoul, Republic of Korea
Bachelor of Science, Statistics
2010-2013 Gyeonggi Science High School, Suwon, Republic of Korea
Papers
Papers
Generalization Bounds for Transformer-Based Next-Token Prediction in a Language Model [preprint]
Insung Kong, Niklas Dexheimer, Johannes Schmidt-Hieber
On the universal representation property of spiking neural networks [preprint]
Shayan Hundrieser, Philipp Tuchel, Insung Kong, Johannes Schmidt-Hieber
Bayesian Additive Regression Trees for functional ANOVA model [preprint]
Seokhun Park, Insung Kong, Yongdai Kim.
Yuha Park, Kunwoong Kim, Insung Kong, Yongdai Kim. Neural Networks. 2027.
Deep Barycentric Regression for Optimal Transport Map Estimation and its Statistical Optimality [preprint]
Kunwoong Kim, Insung Kong, Yongdai Kim. Accepted in NeurIPS 2026 as Spotlight.
Hyper Input Convex Neural Networks for Shape Constrained Learning and Optimal Transport [preprint]
Shayan Hundrieser*, Insung Kong*, Johannes Schmidt-Hieber. Accepted in NeurIPS 2026.
A Composite Activation Function for Learning Stable Binary Representations [preprint]
Seokhun Park, Choeun Kim, Kwanho Kim, Sehyun Park, Insung Kong, Yongdai Kim. Accepted in NeurIPS 2026.
Insung Kong, Juntong Chen, Sophie Langer, Johannes Schmidt-Hieber. Constructive Approximation. 2026.
A Fair Bayesian Inference through Matched Gibbs Posterior [pdf]
Jihu Lee, Kunwoong Kim, Sehyun Park, Insung Kong, Dongyoon Yang, Yongdai Kim. International Conference on Learning Representations (ICLR). 2026.
Bayesian Neural Networks for Functional ANOVA model [pdf]
Seokhun Park*, Choeun Kim*, Jihu Lee, Yunseop Shin, Insung Kong, Yongdai Kim. International Conference on Learning Representations (ICLR). 2026.
On the use of supervised anomaly detection algorithms for extremely imbalanced data [pdf]
Kyungseon Lee, Jongjin Lee, Insung Kong, Yongdai Kim. Journal of the Korean Statistical Society. 2025.
Tensor Product Neural Networks for Functional ANOVA Model [pdf]
Seokhun Park, Insung Kong, Yongchan Choi, Chanmoo Park, Yongdai Kim. International Conference on Machine Learning (ICML). 2025.
Learning deep generative models based on binomial log-likelihood [pdf]
Hwichang Jeong, Insung Kong, Yongdai Kim. Neurocomputing. 2025.
Posterior concentrations of fully-connected Bayesian neural networks with general priors on the weights [pdf]
Insung Kong, Yongdai Kim. Journal of Machine Learning Research (JMLR). 2025.
Fair Representation Learning for Continuous Sensitive Attributes using Expectation of Integral Probability Metrics [pdf] [arxiv]
Insung Kong*, Kunwoong Kim*, Yongdai Kim. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). 2025.
Fairness Through Matching [pdf]
Kunwoong Kim, Insung Kong, Jongjin Lee, Minwoo Chae, Sangchul Park, Yongdai Kim. Transaction on Machine Learning Research (TMLR). 2024.
On Measuring the Quality of Group Fairness [pdf]
Kunwoong Kim, Insung Kong, Jongjin Lee, Minwoo Chae, Sangchul Park, Yongdai Kim. Journal of Artificial Intelligence Research and Applications. 2024.
Enhancing Adversarial Robustness in Low-Label Regime via Adaptively Weighted Regularization and Knowledge Distillation [pdf]
Dongyoon Yang, Insung Kong, Yongdai Kim. International Conference on Computer Vision (ICCV). 2023.
Covariate balancing using the integral probability metric for causal inference [pdf]
Insung Kong, Yuha Park, Joonhyuk Jung, Kwonsang Lee, Yongdai Kim. International Conference on Machine Learning (ICML). 2023.
Masked Bayesian Neural Networks: Theoretical Guarantee and its Posterior Inference [pdf]
Insung Kong, Dongyoon Yang, Jongjin Lee, Ilsang Ohn, Gyuseung Baek, Yongdai Kim. International Conference on Machine Learning (ICML). 2023.
Improving Adversarial Robustness by Putting More Regularizations on Less Robust Samples [pdf]
Dongyoon Yang, Insung Kong, Yongdai Kim. International Conference on Machine Learning (ICML). 2023.
Learning fair representation with a parametric integral probability metric [pdf]
Dongha Kim, Kunwoong Kim, Insung Kong, Ilsang Ohn, Yongdai Kim. International Conference on Machine Learning (ICML). 2022.
Teaching
Teaching
Fall 2026 Linear Algebra
UNIST, Republic of Korea
Spring 2024 Statistical Machine Learning
Sookmyung Women’s University, Republic of Korea
Talks
Talks
2026-02-16 ANOVA-TPNN: Toward an Interpretable and Identifiable Neural Network
Data Science in Finance seminar, Purdue University, USA
2025-12-19 Statistical Properties of Deep Heaviside Networks
The winter meeting of The Korean Statistical Scoiety, Seoul National University, Republic of Korea
2025-09-01 Study on the Prior of Bayesian Neural Networks for Nonparametric Regression.
The European Conference on Numerical Mathematics and Advanced Applications (ENUMATH), Heidelberg University, Germany
2025-06-13 Deep Heaviside Networks and their Augmentations
The 4th workshop on Mathematics and AI, University of Tilburg, the Netherlands
2025-06-05 On the Optimality of Bayesian Neural Networks with Gaussian Priors
Academic Seminar, Ruhr University Bochum, Germany
2025-05-13 Deep Heaviside networks: from artificial neural networks to biological neural networks
Academic Seminar, Inha University, Republic of Korea
2025-05-07 On the Expressivity of Deep Heaviside Networks
Workshop on the Statistical Theory of Neural Networks, Egmond aan Zee, the Netherlands
2024-12-27 Function approximations using DNNs and applications for statistical learning
Academic Seminar, Seoul National University, Republic of Korea
2024-06-13 A Study on Bayesian Neural Networks and Their Variations
Academic Seminar, Sookmyung Women’s University, Republic of Korea
2024-06-07 Posterior concentrations of fully-connected Bayesian neural networks with general priors on the weights
Bayesian Statistical Research Society, University of Seoul, Republic of Korea
Referee Service
Referee Service
Annals of Statistics
Bernoulli
Communications in Statistics - Theory and Methods
Computational Statistics
Electronic Journal of Statistics
IEEE Transactions on Information Theory
IEEE Transactions on Pattern Analysis and Machine Intelligence
Journal of Applied Probability
Journal of Statistical Planning and Inference
Neurips
Statistical Papers
Google Sites
Report abuse