Site main picture

Jinseong Park

AI Research Fellow at KIAS

jinseong [AT] kias [dot] re [dot] kr

About

I am an AI research fellow at Korea Institute for Advanced Study (KIAS).
At KIAS, I serve as part of my alternative military service as a technical research personnel until February 2027.

Short BIO: B.S. (IME, POSTECH) → M.S./Ph.D. (SCLF Lab advised by Prof. Jaewook Lee, IE, SNU) → AI Research Fellow (CAINS, KIAS) including a visiting Postdoc (ParkLabML hosted by Prof. Mi Jung Park, UBC) → AI Fellow (CAINS, KIAS) → ??

I am always open to collaborations! Please refer to topics of interest below.

Research Interests

  • Trustworthy AI Investigating theoretical foundations and practical solutions to ensure the privacy and safety of AI systems.
    • Formal Privacy of ML: Optimizing the utility-privacy trade-off in generalization for machine learning models trained with differential privacy via theoretical analysis.
    • Emerging Privacy of Generative AI: Developing efficient machine unlearning for generative AI, including diffusion models, and preventing privacy leakage in LLMs.
    • Generalization and Robustness: Understanding generalization in weight space and enhancing deep learning safety in data space through loss landscape analysis.

  • Industrial AI Designing scalable and robust applications for industrial domains by leveraging trustworthy AI.
    • Time-Series Modeling: Investigating synthetic data generation and robust forecasting frameworks to manage noisy and scarce data in the industrial domain.
    • AI in Manufacturing: Designing privacy-preserving manufacturing frameworks and automation of the manufacturing pipeline with AI.
    • AI in Finance: Augmenting limited data samples in financial applications to build robust predictions.

Note. J: journal, C: conference, W: workshop, and P: preprint.

News

  • Feb 2026. At KIAS, I have been promoted to AI Fellow, starting March 2026.

  • Jan 2026. One paper on "diffusion unlearning" was accepted at ICLR 2026.

  • Nov 2025. One paper on "time series diffusion" was accepted at KDD 2026. See you in Jeju Island!

  • Sep 2025. One paper on "differential privacy in machine learning" was accepted at NeurIPS 2025. See you in San Diego!

  • Aug 2025. One paper on "privacy in LLMs and RAG" was accepted at EMNLP Findings 2025.

  • Jun 2025. I will visit UBC (Vancouver, Canada) as a visiting Postdoc from June to July 2025. I will also attend ICML 2025 in person :)

  • Mar 2025. I moved to KIAS as an AI Research Fellow!

  • Jul 2024. I attended ICML 2024 in person. See you in Vienna!

  • Mar 2024. One paper on "DP-SGD with diffusion models" was accepted at CVPR 2024. See you in Seattle!

  • Jan 2024. One paper on "diffusion fairness" was accepted at AAAI 2024. See you in Vancouver!

  • Dec 2023. One paper on "robust generalization" was accepted at NeurIPS 2023.

  • Jul 2023. Two papers on "generalization in DP-SGD" and "generalization of SGD" were accepted at ICML 2023. See you in Hawaii!

Publications

Recent publications and preprints are available on Google Scholar.
[J]: Journal, [C]: Conference, [W]: Workshop. †: equal contribution. *: corresponding author.

Journals:

[J6] Differentially private upsampling for enhanced anomaly detection in imbalanced data

Yujin Choi, Jinseong Park, Youngjoo Park, Jaewook Lee, Junyoung Byun

Engineering Applications of Artificial Intelligence, 2026

Paper

[J5] Modeling Asset Price Process: An Approach for Imaging Price Chart with Generative Diffusion Models

Jinseong Park, Hyungjin Ko, Jaewook Lee

Computational Economics, 2024

Paper

[J4] Fast sharpness-aware training for periodic time series classification and forecasting

Jinseong Park, Hoki Kim, Yujin Choi, Woojin Lee, Jaewook Lee

Applied Soft Computing, 2023

Paper

[J3] Generating Transferable Adversarial Examples for Speech Classification

Hoki Kim, Jinseong Park, Jaewook Lee

Pattern Recognition, 2023

Paper

[J2] Efficient differentially private kernel support vector classifier for multi-class classification

Jinseong Park, Yujin Choi, Junyoung Byun, Jaewook Lee, Saerom Park

Information Sciences, 2023

Paper Code

[J1] Exploring Diverse Feature Extractions for Adversarial Audio Detection

Yujin Choi, Jinseong Park, Jaewook Lee, Hoki Kim

IEEE Access, 2023

Paper

Conferences:

[C11] TimeBridge: Leveraging Priors via Diffusion Bridge for Time Series Generation

Jinseong Park†, Seungyun Lee†, Woojin Jeong, Yujin Choi, Jaewook Lee

ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2026

Paper

[C10] Co-occurring Associated REtained concepts in Diffusion Unlearning

Miso Kim, Georu Lee, Yunji Kim, Hoki Kim*, Jinseong Park*, Woojin Lee*

International Conference on Learning Representations (ICLR), 2026

Paper

[C9] Multi-Class Support Vector Machine with Differential Privacy

Jinseong Park, Yujin Choi, Jaewook Lee

Advances in Neural Information Processing Systems (NeurIPS), 2025

Paper

[C8] Safeguarding Privacy of Retrieval Data against Membership Inference Attacks: Is This Query Too Close to Home?

Yujin Choi, Youngjoo Park, Junyoung Byun, Jaewook Lee, Jinseong Park*

EMNLP Findings, 2025

Paper

[C7] Are Self-Attentions Effective for Time Series Forecasting?

Dongbin Kim, Jinseong Park, Jaewook Lee, Hoki Kim

Advances in Neural Information Processing Systems (NeurIPS), 2024

Paper Code

[C6] In-distribution Public Data Synthesis with Diffusion Models for Differentially Private Image Classification

Jinseong Park†, Yujin Choi†, Jaewook Lee

The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

Paper Code

[C5] Fair Sampling in Diffusion Models through Switching Mechanism

Yujin Choi†, Jinseong Park†, Hoki Kim, Jaewook Lee, Saerom Park

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024

Paper Code

[C4] Fantastic Robustness Measures: The Secrets of Robust Generalization

Hoki Kim, Jinseong Park, Yujin Choi, Jaewook Lee

Advances in Neural Information Processing Systems (NeurIPS), 2023

Paper Code

[C3] Differentially Private Sharpness-Aware Training

Jinseong Park, Hoki Kim, Yujin Choi, Jaewook Lee

International Conference on Machine Learning (ICML), 2023

Paper Code

[C2] Implicit Jacobian regularization weighted with impurity of probability output

Sungyoon Lee, Jinseong Park, Jaewook Lee

International Conference on Machine Learning (ICML), 2023

Paper

[W1] Enhancing non-linear asset volatility forecasting models with investor sentiment and Explainable AI

Seungju Lee, Jinseong Park, Jaewook Lee

Korea Computer Congress (KCC) XAI Workshop, 2023, Best Paper Award (2/37)

Paper

[C1] Towards Better Understanding of Training Certifiably Robust Models against Adversarial Examples

Sungyoon Lee, Woojin Lee, Jinseong Park, Jaewook Lee

Advances in Neural Information Processing Systems (NeurIPS), 2021

Paper Code

Vitæ

Experience

[Project] COSMAX Dec. 2023 - Sep. 2024

Deep learning prediction and generative models for color cosmetics

[Project] LNG industry research project with POSCO EnergySep. 2019 - Dec. 2019

Researcher, with Prof. Ribin Seo at POSTECH

[Lecturer] Soongsil UniversitySep. 2024 - Feb. 2025

Analysis of Financial Time Series, Fall Semester 2024, School of Finance

[Invited Talks] DesiloJul. 2024

Deep Learning with Differential Privacy

[Attendee] Graduate Summer School on Algorithmic FairnessJul. 2022

Take the summer school organized by Cynthia Dwork, Guy Rothblum, and Noa Dagan, Institute for Pure & Applied Mathematics (IPAM) at UCLA

[Intern] Industrial Data Engineering & Analytics (IDEA) Lab.Sep. 2019 - Dec. 2019

Research Intern, hosted by Prof. Hyunbo Cho at POSTECH

[Intern] SK TelecomJun. 2018 - Aug. 2018

Intern, R&D Digital Transformation Team

Jinseong Park

based on Miguel Méndez templates
(Martin Saveski and Marina Aisa)