Trustworthy Learning Theory

Modern AI systems introduce practical issues, including hallucination, harmful responses, and safety and security risks. This raises the following question.

How do we train AI systems that are truthful, safe, and secure with theoretical guarantees?

Keywords

  • Conformal Prediction
  • Conformal Abstention
  • Uncertainty Quantification
  • Reinforcement Learning
  • Online Learning
  • Meta-learning

17 Related Publications

Towards Functional Correctness of Large Code Models with Selective Generation
Jaewoo Jeong, Taesoo Kim, Sangdon Park
International Conference on Machine Learning (ICML), 2026
ICML NeurIPS'25 DL4C Workshop
Online Conformal Prediction with Adversarial Semi-bandit Feedback via Regret Minimization
Junyoung Yang*, Kyungmin Kim*, Sangdon Park
International Conference on Learning Representations (ICLR), 2026
ICLR ICLR'26 Reliable Autonomy Workshop
Online Conformal Abstention for Factuality Control Under Adversarial Bandit Feedback
Minjae Lee*, Yoonjae Jung*, Sangdon Park
2025
🏆 Best Paper Finalist from CKAIA ICLR'26 Reliable Autonomy Workshop
Selective Generation for Controllable Language Models
Minjae Lee*, Kyungmin Kim*, Taesoo Kim, Sangdon Park
Neural Information Processing Systems (NeurIPS), 2024
NeurIPS 🏆 Spotlight (Top 2.08%)🏆 POSTECH GSAI BK21 Best Paper Award
TRAQ: Trustworthy Retrieval Augmented Question Answering via Conformal Prediction
Shuo Li, Sangdon Park, Insup Lee, Osbert Bastani
North American Chapter of the Association for Computational Linguistics (NAACL), 2024
NAACL 🏆 ICML'23 TEACH Workshop Best Paper Award ICML'23 TEACH Workshop
PAC Prediction Sets Under Label Shift
Wenwen Si, Sangdon Park, Insup Lee, Edgar Dobriban, Osbert Bastani
International Conference on Learning Representations (ICLR), 2024
ICLR
ACon$^2$: Adaptive Conformal Consensus for Provable Blockchain Oracles
Sangdon Park, Osbert Bastani, Taesoo Kim
USENIX Security Symposium (Security), 2023
Security
CODiT: Conformal Out-of-Distribution Detection in Time-Series Data for Cyber-Physical Systems
Ramneet Kaur, Kaustubh Sridhar, Sangdon Park, Yahan Yang, Susmit Jha, Anirban Roy, Oleg Sokolsky, Insup Lee
International Conference on Cyber-Physical Systems (ICCPS), 2023
🏆 ICCPS'23 Best Paper Award Finalist
PAC Prediction Sets for Meta-Learning
Sangdon Park, Edgar Dobriban, Insup Lee, Osbert Bastani
Neural Information Processing Systems (NeurIPS), 2022
NeurIPS
Sequential Covariate Shift Detection Using Classifier Two-Sample Tests
Sooyong Jang, Sangdon Park, Insup Lee, Osbert Bastani
International Conference on Machine Learning (ICML), 2022
ICML
Towards PAC Multi-Object Detection and Tracking
Shuo Li, Sangdon Park, Xiayan Ji, Insup Lee, Osbert Bastani
2022
PAC Prediction Sets Under Covariate Shift
Sangdon Park, Edgar Dobriban, Insup Lee, Osbert Bastani
International Conference on Learning Representations (ICLR), 2022
ICLR
PAC Confidence Predictions for Deep Neural Network Classifiers
Sangdon Park, Shuo Li, Insup Lee, Osbert Bastani
International Conference on Learning Representations (ICLR), 2021
ICLR
iDECODe: In-distribution Equivariance for Conformal Out-of-distribution Detection
Ramneet Kaur, Susmit Jha, Anirban Roy, Sangdon Park, Edgar Dobriban, Oleg Sokolsky, Insup Lee
Association for the Advancement of Artificial Intelligence (AAAI), 2021
AAAI
Calibrated prediction with covariate shift via unsupervised domain adaptation
Sangdon Park, Osbert Bastani, James Weimer, Insup Lee
International Conference on Artificial Intelligence and Statistics (AISTATS), 2020
PAC Confidence Sets for Deep Neural Networks via Calibrated Prediction
Sangdon Park, Osbert Bastani, Nikolai Matni, Insup Lee
International Conference on Learning Representations (ICLR), 2020
ICLR
Resilient linear classification: an approach to deal with attacks on training data
Sangdon Park, James Weimer, Insup Lee
International Conference on Cyber-Physical Systems (ICCPS), 2017