Sangdon Park (박상돈),
Postdoc Researcher (Mentor: Prof. Taesoo Kim),
School of Cybersecurity and Privacy (SCP),
College of Computing,
Georgia Tech


Meta PAC Prediction Sets is accepted to NeurIPS22.

Research Interests

Machine Learning, Trustworthy Machine Learning, Uncertainty Quantification, and Computer Security — My research interest focuses on designing safe and secure AI systems by understanding from theory to implementation and applications in computer security, computer vision, robotics, and natural language processing.


ACon²: Adaptive Conformal Consensus for Provable Blockchain Oracles
Sangdon Park, Osbert Bastani, and Taesoo Kim

Unsafe’s Betrayal: Abusing Unsafe Rust in Binary Reverse Engineering toward Finding Memory-safety Bugs via Machine Learning
Sangdon Park, Xiang Cheng, and Taesoo Kim

PAC Prediction Sets for Meta-Learning
Sangdon Park, Edgar Dobriban, Insup Lee, and Osbert Bastani
Neural Information Processing Systems (NeurIPS) 2022
[arXiv] [Paper] [Code] [Video]

Sequential Covariate Shift Detection Using Classifier Two-Sample Tests
Sooyong Jang, Sangdon Park, Insup Lee, and Osbert Bastani
International Conference on Machine Learning (ICML) 2022

Towards PAC Multi-Object Detection and Tracking
Shuo Li, Sangdon Park, Xiayan Ji, Insup Lee, Osbert Bastani

PAC Prediction Sets Under Covariate Shift
Sangdon Park, Edgar Dobriban, Insup Lee, and Osbert Bastani
International Conference on Learning Representations (ICLR) 2022
[arXiv] [Paper] [Code] [Video]

iDECODe: In-distribution Equivariance for Conformal Out-of-distribution Detection
Ramneet Kaur, Susmit Jha, Anirban Roy, Sangdon Park, Edgar Dobriban, Oleg Sokolsky, and Insup Lee
Association for the Advancement of Artificial Intelligence (AAAI) 2022

Uncertainty Estimation Toward Safe AI
PhD thesis, UPenn, Aug. 2021

PAC Confidence Predictions for Deep Neural Network Classifiers
Sangdon Park, Shuo Li, Insup Lee, and Osbert Bastani
International Conference on Learning Representations (ICLR) 2021
[arXiv] [Paper] [Code] [Video]

Calibrated Predictions with Covariate Shift via Unsupervised Domain Adaptation
Sangdon Park, Osbert Bastani, James Weimer, and Insup Lee
International Conference on Artificial Intelligence and Statistics (AISTATS) 2020
[arXiv] [Paper] [Code] [Video]

PAC Confidence Sets for Deep Neural Networks via Calibrated Prediction
Sangdon Park, Osbert Bastani, Nikolai Matni, and Insup Lee
International Conference on Learning Representations (ICLR) 2020
[arXiv] [Paper] [Code] [Video] [Short, ICML UDL20]

From Verification to Learning for Defense against Adversarial Examples in Neural Networks
Sangdon Park, Radoslav Ivanov, James Weimer, and Insup Lee
Korea Cyber-security Competition 2018
Best paper award

Resilient Linear Classification: An Approach to Deal with Attacks on Training Data
Sangdon Park, James Weimer, and Insup Lee
International Conference on Cyber-Physical Systems (ICCPS) 2017
[Paper] [arXiv] [BibTex] [DOI]

Integrated Intelligence for Human Robot Teams
Jean Oh, Thomas Howard, Matthew Walter, Daniel Barber, Menglong Zhu, Sangdon Park, Arne Suppe, Luis NavarroSerment, Felix Duvallet, Abdeslam Boularias, Oscar Romero, Jerry Vinokurov, Terence Keegan, Robert Dean, Craig Lennon, Barry Bodt, Marshal Childers, Jianbo Shi, Kostas Daniilidis, Nicholas Roy, Christian Lebiere, Martial Hebert, and Anthony Stentz
International Symposium on Experimental Robotics (ISER) 2016
[Paper] [DOI]

Abnormal Object Detection by Transformed-Canonical Scene Generation
MS thesis, Seoul National University, Aug. 2012
Distinguished Dissertation Award

Abnormal Object Detection by Canonical Scene-based Contextual Model
Sangdon Park, Wonsik Kim, and Kyoung Mu Lee
European Conference on Computer Vision (ECCV) 2012
[Project Page] [Paper] [BibTex] [Code] [Dataset]

Behavioral Intelligence for Crowd Avatar in 3D Virtual Worlds
BS thesis, Seoul National University, Feb. 2010