From the lab
Recent publications, research milestones, and updates from SAIL.
News
Haksoo Lim, Myeongjin Lee, Wonjoon Chang, and Jaesik Choi Unified Backbone Refinement for Diffusion Models via Internal-Latent Analysis European Conference on Computer Vision (ECCV) 2026.
News
Soyeon Kim, Kyowoon Lee, and Jaesik Choi Diffusion Integrated Gradients: Controllable Path Generation for Flexible Feature Attribution European Conference on Computer Vision (ECCV) 2026.
News
Sehyun Lee, Dahee Kwon, Damin Lee, and Jaesik Choi SPICE: Simple Polysemantic Feature Interpretation via Clustering-based Explanations European Conference on Computer Vision (ECCV) 2026.
News
Spectral Integrated Gradients for Coarse-to-Fine Feature Attribution, led by Soyeon Kim, Seongwoo Lim, Kyowoon Lee, and Jaesik Choi is accepted at KDD 2026.
News
A paper, Manifold-Aligned Guided Integrated Gradients for Reliable Feature Attribution by Soyeon Kim, Seongwoo Lim, Kyowoon Lee, and Jaesik Choi is accepted at International Conference on Machine Learning (ICML), 2026.
News
A paper, Breaking the Lock-in: Diversifying Text-to-Image Generation via Representation Modulation by Dahee Kwon, Haeun Lee, and Jaesik Choi is accepted at International Conference on Machine Learning (ICML), 2026.
News
A paper, DistMatch: Adaptive Binning via Distribution Matching for Robust Sequential Conformal Prediction by Enver Menadjiev, Jihyeon Seong, Jisu Yeo, and Jaesik Choi is accepted at International Conference on Machine Learning (ICML), 2026.
News
A paper, Incomplete Prompt Jailbreaks in Large Language Models, written by Yeonjea Kim, Bumjin Park, and Jaesik Choi is accepted at Findings of ACL 2026.
News
A paper, K-MetBench: A Multi-Dimensional Benchmark for Fine-Grained Evaluation of Expert Reasoning, Locality, and Multimodality in Meteorology, written by Soyeon, Cheongwoong, Myeongjin Lee, and Jaesik Choi is accepted at Findings of ACL 2026.
News
A paper, Towards Transparent Time Series Analysis: Exploring Methods and Enhancing Interpretability, written by Youngjin Park, Anh Tong, Sehyun Lee, Jihyeon Seong, Qin Xie, and Jaesik Choi, is accepted at ACM Computing Surveys, 2026.
News
[Paper] A paper, State-Covering Trajectory Stitching for Diffusion Planners, written by Kyowoon Lee and Jaesik Choi, is accepted at Conference on Neural Information Processing Systems (NeurIPS) 2025.
News
A review paper on XAI receives the best paper award from Information Fusion. Luca Longo, Mario Brcic, Federico Cabitza, Jaesik Choi, Roberto Confalonieri, Javier Del Ser, Riccardo Guidotti, Yoichi Hayashi, Francisco Herrera, Andreas Holzinger, Richard Jiang, Hassan Khosravi, Fred
News
A paper, On the Relationship between Populated Regions and Adversarial Robustness in Deep Neural Networks, written by Seongjin Park, Haedong Jeong, Tair Djanibekov, Giyoung Jeon, Jinseok Seol, and Jaesik Choi is accepted at ICDM 2025.
News
Human-Centric AI: From Explainability and Trustworthiness to Actionable Ethics, an ACM CIKM-2026 workshop.
News
A paper, When Format Changes Meaning: Investigating Semantic Inconsistency of Large Language Model, written by Cheongwoong Kang, Jongeun Baek, Yeonjea Kim, and Jaesik Choi is accepted at Findings of EMNLP 2025.
News
A paper, Amortized Baseline Selection via Rank-Revealing QR for Efficient Model Explanation, written by Chanwoo Lee, Youngjin Park, Hyeongeun Lee, Yeeun Yoo, Daehee Han, Junho Choi, Geonhyeong Kim, Nari Kim, and Jaesik Choi is accepted at CIKM 2025.
News
A paper, Unsupervised Concept Discovery for Deep Weather Forecast Models with High-Resolution Radar Data, written by Soyeon Kim, Junho Choi, Subeen Lee, and Jaesik Choi is accepted at PLOS Climate 2025.
News
A paper, Deep-DFVAR: Dynamic Factor Vector Autoregression with Deep Learning for Regional House Price Index Forecasting, written by Jisu Yeo, Artyom Stitsyuk and Jaesik Choi is accepted at Neurocomputing.
News
A paper, Granular Concept Circuits: Toward a Fine-Grained Circuit Discovery for Concept Representations, written by Dahee Kwon*, Sehyun Lee* and Jaesik Choi is accepted at ICCV-2025.
News
A paper, Example-Based Concept Analysis Framework for Deep Weather Forecast Models, written by Soyeon Kim, Junho Choi, Subeen Lee and Jaesik Choi is accepted at AMS Artificial Intelligence for the Earth Systems.
News
A paper, Context-Aware Negative Sampling for Sequential Recommendation written by Jinseok Seol and Jaesik Choi, is accepted at IEEE Access.
News
A paper, Counterfactual Activation Editing for Post-hoc Prosody and Mispronunciation Correction in TTS Models, written by Kyowoon Lee, Artyom Stitsyuk, Gunu Jho, Inchul Hwang and Jaesik Choi is accepted at Interspeech 2025.
News
A paper, XAI—Explainable artificial intelligence, written by Prof. Jaesik Choi is selected to receive the 2025 Frontiers of Science Award by the International Congress of Basic Science
News
[Paper] A paper, Local Manifold Approximation and Projection for Manifold-Aware Diffusion Planning, written by Kyowoon Lee and Jaesik Choi, is accepted at International Conference on Machine Learning (ICML) 2025.
News
A paper, Deontological Keyword Bias: The Impact of Modal Verbs on Normative Judgments of Language Models, written by Bumjin Park, Jinsil Lee and Jaesik Choi, is accepted at Annual Meeting of the Association for Computational Linguistics (ACL) 2025.
News
A paper, A Perspective on Foundation Models in Chemistry, written by Junyoung Choi, Gunwook Nam, Jaesik Choi and Yousung Jung, is accepted at Journal of the American Chemical Society (JACS AU).
News
A paper, An Automated Method to Correct Artifacts in Neural Text-to-Speech Models, written by Seongyeop Jeong, June Sig Sung, Inchul Hwang and Jaesik Choi is accepted at IEEE/ACM Transactions on Audio, Speech and Language Processing (T-ASLP).
News
A paper, Enhancing Creative Generation on Stable Diffusion-based Models, written by Jiyeon Han, Dahee Kwon, Gayoung Lee, Junho Kim, Jaesik Choi is accepted at CVPR-2025.
News
Prof. Jaesik Choi is appointed as KAIST Endowed Chair Professor from March 1st, 2025 to February 29th, 2028.
News
A paper, Neural ODE Transformers: Analyzing Internal Dynamics and Adaptive Fine-tuning, written by Anh Tong and Jaesik Choi is accepted at ICLR-25.
News
A paper, Rethinking Shapley Value for Negative Interactions in Non-convex Games, written by Wonjoon Chang, Myeongjin Lee and Jaesik Choi is accepted at ICLR-25.
News
A paper, xPatch: Dual-Stream Time Series Forecasting with Exponential Seasonal-Trend Decomposition, written by Artyom Stitsyuk and Jaesik Choi is accepted at AAAI-25. xPatch domonstrates the best (STOA) performance in many contemporary time series benchmarks. Now, xPatch ranks to
News
A paper, Diverse Rare Sample Generation with Pretrained GANs, written by Subeen Lee, Jiyeon Han, Soyeon Kim and Jaesik Choi is accepted at AAAI-25.
News
Prof. Jaesik Choi is elected as an associate member of the National Academy of Engineering of Korea. The National Academy of Engineering of Korea was established to foster further innovation in engineering and technology by recognizing engineers who have made exceptional achievem
News
Dr. Haedong Jeong joins the School of Art & Technology at Korea University as an assistant professor. Congratulate Prof. Jeong! https://creative.sogang.ac.kr/a6/
News
KAIST's Center for Explainable AI has released an open-source plug-and-play framework that simplifies AI explainability without requiring expert knowledge. The framework automatically detects model structures, recommends compatible explanation algorithms, and provides visual and
News
Prof. Nari Kim received the prestigious award from the Chairperson of the Personal Information Protection Commission with the contribution to the technical contributions of the Korean privacy law. https://news.kaist.ac.kr/news/html/news/?mode=V&mng_no=40090
News
Our paper, Probing Network Decisions: Capturing Uncertainties and Unveiling Vulnerabilities without Label Information, written by Youngju Joung, Sehyun Lee and Jaesik Choi, is received the best paper honorable mention at International Conference on Pattern Recognition and Artific
News
Jiyeon Han (PhD student) is invited to present at Ai4 Summit 2024 AI Research Summit https://app.swapcard.com/widget/event/ai4-2024/person/RXZlbnRQZW9wbGVfMzE1NzkwMjg=
News
Kyowoon Lee successfully defended his PhD thesis, Learning to Achieve Goals via Curriculum and Hierarchical Reinforcement Learning. Congratulations Dr. Lee!
News
A paper, Memorizing Documents with Guidance in Large Language Models, written by Bumjin Park and Jaesik Choi is accepted at IJCAI-2024. 1. Problem This work tackles the problem of storing document-wise memories in LLMs. 2. Proposed Method This work proposes a document representat
News
A paper, Towards Dynamic Trend Filtering through Trend Points Detection with Reinforcement Learning, written by Jihyeon Seong, Sekwang Oh and Jaesik Choi is accepted at IJCAI-2024.
News
Dr. Anh Tong (the first PhD graduate of SAIL) joins the Graduate School of AI at Korea University as an assistant professor. Congratulate Prof. Tong! http://xai.korea.ac.kr/teacher/teacher
News
A paper, Explainable Artificial Intelligence (XAI) 2.0: A manifesto of open challenges and interdisciplinary research directions, is accepted at Information Fusion.
News
A paper, Pathwise Explanation of ReLU Neural Networks, written by Seongwoo Lim, Won Jo, Joohyung Lee and Jaesik Choi, is accepted at AISTATS-24.
News
A paper, Towards Diverse Perspective Learning with Select over Multiple Temporal Poolings, written by Jihyeon Seong, Jungmin Kim and Jaesik Choi is accepted at AAAI-24.
News
A paper, Understanding Distributed Representations of Concepts in Deep Neural Networks without Supervision, written by Wonjoon Chang, Dahee Kwon and Jaesik Choi is accepted at AAAI-2024.
News
Explainable AI-Based Interface System for Weather Forecasting Model written by Soyeon Kim, Junho Choi, Yeji Choi, Subeen Lee, Artyom Stitsyuk, Minkyoung Park, Seongyeop Jeong, You-Hyun Baek and Jaesik Choi is presented at International Conference on Human-Computer Interaction Int
News
- Ye Eun Chun, Sunjae Kwon, Kyunghwan Sohn, Nakwon Sung, Junyoup Lee, Byoung Ki Seo, Kevin Compher, Seung-won Hwang and Jaesik Choi, CR-COPEC: Causal Rationale of Corporate Performance Changes to learn from Financial Reports, Findings of EMNLP, 2023. - Cheongwoong Kang and Jaesik
News
A paper, SigFormer: Signature Transformers for Deep Hedging, written by Anh Tong, Thanh Nguyen-Tang, Dongeun Lee, Toan Tran and Jaesik Choi is accepted at ACM ICAIF 2023.
News
A paper, Refining Diffusion Planner for Reliable Behavior Synthesis by Automatic Detection of Infeasible Plans, written by Kyowoon Lee, Seongun Kim and Jaesik Choi is accepted at NeurIPS-2023.
News
Dear potential candidate of the lab (SAIL@KAIST) and the center (XAI@KAIST), Recently we have received many applications for the graduate studentship positions at KAIST Kim Jaechul Graduate School of AI and the researcher positions at Explainable Artificial Intelligence Center at
News
A paper, Beyond Single Path Integrated Gradients for Reliable Input Attribution via Randomized Path Sampling, written by Giyoung Jeon, Haedong Jeong and Jaesik Choi is accepted at ICCV-2023.
News
Seongun Kim, a PhD student of our lab, presented at Explainable Robotics Workshop at ICRA 2023 as an invited speaker.
News
Our paper, Variational Curriculum Reinforcement Learning for Unsupervised Discovery of Skills, written by Seongun Kim, Kyowoon Lee and Jaesik Choi is accepted at ICML-23.
News
Our paper, Rarity Score: A New Metric to Evaluate the Uncommonness of Synthesized Images, written by Jiyeon Han, Hwanil Choi, Yunjey Choi, Junho Kim, Jung-Woo Ha and Jaesik Choi is accepted at ICLR-23.
News
A paper, Adaptive and Explainable Deployment of Navigation Skills via Hierarchical Deep Reinforcement Learning, written by Kyowoon Lee, Seongun Kim and Jaesik Choi, is accepted at ICRA-23.
News
Prof. Choi presented the work at IPAM XAI Workshop held in UCLA. https://www.ipam.ucla.edu/programs/workshops/explainable-ai-for-the-sciences-towards-novel-insights/ https://www.youtube.com/watch?v=9k692NjtcAU
News
Prof. Choi's article, South Korea’s Response to Surging AI Use in the US and China, is published in Global Asia, a international relational magazine published by the East Asia Foundation.
News
Two papers are accepted at NeurIPS-2022. Anh Tong, Thanh Nguyen-Tang, Toan Tran and Jaesik Choi, Learning White Noises in Neural Stochastic Differential Equations Giyoung Jeon, Haedong Jeong and Jaesik Choi, Distilled Gradient Aggregation: Purify Features for Input Attribution in
News
We have open positions for shot/long-terms AI researchers/SW Engineers. If you are interested in, please send your CV to jaesik.choi (at) kaist.ac.kr.
News
A collaborative work on explainable artificial intelligence received recognition by Samsung.
News
Haedong Jeong successfully defended his PhD thesis, Example-based Methods to Explain the Internal Generative Mechanism of Deep Generative Neural Networks. Congratulations Dr. Jeong!
News
Our paper, Can We Find Neurons that Cause Unrealistic Images in Deep Generative Networks?, written by Hwanil Choi, Wonjoon Chang, Jaesik Choi is accepted at IJCAI-2022.
News
The KAIST Explainable Artificial Intelligence Center is selected by the Ministry of Science and ICT to host a new XAI project (13 Billion KRW, 130억원/5년), Development of Plug and Play Explainable Artificial Intelligence Platform, for the next 57 months.
News
Our paper, An Unsupervised Way to Understand Artifact Generating Internal Units in Generative Neural Networks, written by Haedong Jeong, Jiyeon Han and Jaesik Choi is accepted at AAAI-2022.
News
The KAIST Faculty Personnel Committee and the KAIST Board of Trustees approved the tenure application of Prof. Jaesik Choi.
News
Prof. Choi received a prestigious POSCO Open and Collaboration (O&C) Award at the 33rd POSCO Annual Technical Conference for his contribution of explainable AI research in the POSCO smart factory.
News
KAIST XAI center, led by Prof. Jaesik Choi, made an MOU with IBK Bank for the research and education of explainable AI.
News
A new book, An Introduction to Lifted Probabilistic Inference edited by Guy Van den Broeck, Kristian Kersting and Sriraam Natarajan, on probabilistic inference is available at MIT press. Prof. Jaesik Choi contributed to write two book chapters "6 Lifted Aggregation and Skolemizat
News
Our joint work with Prof. Yuseong Jung's lab at KAIST is accepted at Chemical Science [IF 9.825] Juhwan Kim, Geun Ho Gu, Juhwan Noh, Seongun Kim, Suji Gim, Jaesik Choi* and Yousung Jung*, Predicting Potentially Hazardous Chemical Reactions Using Explainable Neural Network, Chemic
News
Our paper “Explaining the Decisions of Deep Policy Networks for Robotic Manipulations” written by Seongun Kim and Jaesik Choi is accepted to be presented at 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)!
News
Our paper "Semi-Supervised Training of Deep Generative Models for High-Dimensional Anomaly Detection" written by Qin Xie, Peng Zhang, Boseon Yu, Jaesik Choi is accepted to be published at IEEE Transactions on Neural Networks and Learning Systems!
News
Our work, Automatic Correction of Internal Units in Generative Neural Networks written by Ali Tousi, Haedong Jeong, Jiyeon Han, Hwanil Choi and Jaesik Choi, presented in CVPR-21 is appeared in the media. Korea: KAIST ‘AI 실수’ 바로잡는 수리기술 개발 http://www.aitimes.kr/news/articleView.htm
News
Our lab is hosting with the PHM Asia Pacific Data Challenge (Chair: Prof. Jaesik Choi, Supporting Staff: Seijun Chung and Seongyeop Jeong) with colleagues. Please consider participating the carefully prepared competition! http://phmap.org/data-challenge/
News
Our paper, Interpreting Internal Activation Patterns in Deep Temporal Neural Networks by Finding Prototypes, written by Sohee Cho, Wonjoon Chang, Ginkyeng Lee and Jaesik Choi is accepted at KDD-2021.
News
Our paper, Learning Basis Functions of Deep Predictive Neural Networks with Global Spatio-Temporal Covariance Loss, written by Boseon Yoo, Jiwoo Lee, Janghoon Ju, Seijun Chung, Soyeon Kim and Jaesik Choi is accepted at ICML-2021.
News
A paper authored by our lab members is accepted at CVPR-21. – Ali Tousi, Haedong Jeong, Jiyeon Han, Hwanil Choi and Jaesik Choi, Automatic Correction of Internal Units in Generative Neural Networks, IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021.
News
A paper, Prof. Jaesik Choi worked as a co-corresponding author, is accepted at ACS Analytical Chemistry [IF 6.785] Hyunbin Kim, Hae Nim Lee, Jaesik Choi*, Jihye Seong*, Spatiotemporal characterization of GPCR activity and function during endosomal trafficking pathway, ACS Analyti
News
Anh Tong successfully defended his PhD thesis, first in the lab! Congratulations Dr. Tong! With SAIL members, Anh has led to publish 5 papers in top AI conferences (ICML, IJCAI and AAAI)!
News
Our team won the AI competition (the 1st place in the general (non-commercial) track) hosted by the Korea East West Power. https://youtu.be/euAc2vxryM4
News
SAIL members, Kyunghwan, Cheongwoong and Soonjae, won the second award at the 2020 ETRI Open API Use Case Contest Competition! Congratulations!! https://aiopen.etri.re.kr/intro_notice_view.php?wr_id=90
News
Three papers authored by our lab members are accepted at AAAI-21. - Anh Tong, Toan Tran, Hung Bui and Jaesik Choi, Learning Compositional Sparse Gaussian Processes with a Shrinkage Prior, AAAI Conference on Artificial Intelligence (AAAI), 2021. - Anh Tong and Jaesik Choi, Charact
News
안녕하십니까? KAIST 설명가능인공지능 연구센터는 과학기술정보통신부의 후원으로 설립된 연구센터로, 지난 2020년 1월 한국과학기술원에 새 터전을 마련하고 고려대학교, 서울대학교, 연세대학교, UNIST와 AITRICS 소속 250여명의 연구원들이 세계적 수준의 연구를 진행하고 있으며, 미국 DARPA, 독일 베를린 빅데이터 센터등 해외 기관과도 활발하게 교류하고 있습니다. 본 센터는 인공지능 모델의 결정을 인간이 이해하고 해석할 수 있는 수준의 설명을 제공하는 설명가능 인공지능(XAI: Exp
News
Prof. Choi's online lecture will be broadcasted at 8pm (KST) in the following link, Youtube - https://www.youtube.com/watch?v=MMQjmeTmoxQ NAVER - https://tv.naver.com/kaos https://ikaos.org/kaos/apply/view.php?kc_idx=79 Intro Video: https://www.youtube.com/watch?v=7Czt2xTQNmY
News
KAIST AI대학원 성남연구센터 제1회 기술설명회가 2020년 10월 16일 금요일 13:00~18:00에 개최됩니다. 본 설명회는 Zoom을 이용해 온라인으로 개최됩니다. 사전신청을 통해 참가신청을 받고 있으며, 신청자에 한하여 개별상담도 진행됩니다. 포스터의 QR코드 또는 아래의 링크를 통해 사전신청이 가능합니다. - 사전 신청 링크: http://gg.gg/KAISTGSAI
News
A paper is accepted at ICAIF-2020. Youngjin Park, Deokjun Eom, and Jaesik Choi, Improved Predictive Deep Temporal Neural Networks with Trend Filtering, ICAIF-2020
News
KAIST AI대학원 확률형 인공지는 연구실에서 수행하는 연구(시계열 데이터 분석 및 설명가능 인공지능)에 관심있는 지원자는 이력서(CV)를 jaesik.choi (at) kaist.ac.kr로 지원 부탁드립니다. 관심있는 지원자를 위해서 다음과 같은 FAQ를 준비했습니다. https://docs.google.com/document/d/1Y7nZnkSMRI4uB9WJTTsmCyR-mVM3Z4plQ9_wJlYMadY/edit?usp=sharing If you are interested in research
News
A tutorial proposed to KDD 2020, Interpreting and Explaining Deep Neural Networks: A Perspective on Time Series Data, is accepted.
News
A paper written by Nguyen T. Nguyen and Prof. Jaesik Choi in collaboration with Prof. Yong-ri Choi's group and Prof. Sam H. Noh's one is accepted at USENIX ATC 20. Jay H. Park, Gyeongchan Yun, Chang M. Yi, Nguyen T. Nguyen, Seungmin Lee, Jaesik Choi, Sam H. Noh and Young-ri Choi,
News
A papers is accepted at ICRA-2020. YongHyeok Seo, Dongju Shin, Jaesik Choi, and Se Young Chun, A Single Multi-Task Deep Neural Network with Post-Processing for Object Detection with Reasoning and Robotic Grasp Detection, ICRA-2020 (pdf, video).
News
A perspective, review article, XAI - Explainable artificial intelligence, co-authored by Prof. Jaesik Choi with David Gunning (DARPA, currently at Facebook) and other researchers is published at Science Robotics.
News
Anh Tong, a PhD student of UNIST, received NAVER PhD Fellowship award.
News
Two papers are accepted at AAAI-2020. Giyoung Jeon*, Haedong Jeong* and Jaesik Choi, An Efficient Explorative Sampling Considering the Generative Boundaries of Deep Generative Neural Networks, AAAI-2020 (Accepted, Oral, * contributed equally). Woojeong Nam, Shir Gur, Jaesik Choi,
News
Our research team (http://iaic.unist.ac.kr) successfully has finished a research project with Korea East-West Power Company to build time series deep learning models which diagnose tubes in Power Plant Boilers of the Dangjin power station. Our AI module and software are now runni
News
Our paper, Markov Information Bottleneck to Improve Information Flow in Stochastic Neural Networks, written by Thanh and Jaesik is accepted at Special Issue "Information–Theoretic Approaches to Computational Intelligence" of Entropy. https://www.mdpi.com/journal/entropy/special_i
News
Our XAI (Explainable Artificial Intelligence) Center has openings for interns and international exchange students. If you are interested in collaborating with our lab, please send your cv and a cover letter to jaesik.choi at kaist.ac.kr.
News
Our lab has multiple positions for PhD students and researchers. If you are interested in the positions, please apply by sending a cover letter and CV to jaesik.choi at kaist.ac.kr. If you are serious to join our lab, it would be good to read at least one of papers published by o
News
A patent, "Data reduction methods, systems, and devices", is issued by USPTO. https://patents.google.com/patent/US10366078B2
News
In September 2019, Prof. Choi and the Statistical Artificial Intelligence Lab at UNIST moved to the KAIST Graduate School of AI. If you are interested in joining our lab as a graduate student, please apply to the graduate program at the KAIST AI Graduate School.