Our paper FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control (Kim et al.) has been accepted to Robotics: Science and Systems (RSS) 2026 (29.7% acceptance rate) and has won the Best Paper Award!

European Conference on Computer Vision

• Denoised Variance-Based Pruning with Optimal Brain Bias CompensationGeon Tack Lee, Jaegul Choo, and Kang Eun Jeon
• InsertAnywhere: Geometrically Grounded and Optics-Aware Video Object InsertionHoiyeong Jin, Hyojin Jang, Junha Hyung, Jeongho Kim, Kinam Kim, Dongjin Kim, Huijin choi, Hyeonji Kim, and Jaegul Choo
• Infinite-Homography as Robust Conditioning for Camera-Controlled Video GenerationMin-Jung Kim, Jeongho Kim, Hoiyeong Jin, Junha Hyung, and Jaegul Choo
• ReDesign: Recovering Editable Design Structures from Raster Images via Agentic DecompositionJooyeol Yun, Jintae Park, Hyesu Lim, Junha Hyung, Hyungjin Chung, and Jaegul Choo

Annual Meeting of the Association for Computational Linguistics

• Talk to Your Slides: High-Efficiency Slide Editing via Language-Driven Structured Data ManipulationKyudan Jung, Hojun Cho, Jooyeol Yun, Soyoung Yang, Jaehyeok Jang, Jaegul Choo
• Retrieve Only Relevant Tables Whether Few or Many: Adaptive Table Retrieval Method Taehee Kim*, Seungbin Yang*, Jihwan Kim, Jaegul Choo
• Sommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language ModelsKyudan Jung, Jihwan Kim, Soyoon Kim, Jeonghoon Kim, Jaegul Choo, and Cheonbok Park

Our lab’s work on EgoX, an AI model that converts ordinary third-person videos into realistic first-person perspective views, has been featured in major Korean news outlets.

The articles highlight how EgoX can generate immersive 1st-person scenes without action cams or wearable devices, enabling new possibilities in AR/VR, robotics, sports, and immersive media.

🔗 Media Coverage

• Herald Economy: “고가 액션캠 없이도… 일반영상 → 1인칭 영상시점 전환”

• YouTube news feature on EgoX technology

IEEE International Conference on Computer Vision and Pattern Recognition

• MM-SeR: Multimodal Self-Refinement for Lightweight Image CaptioningJunha Song, Yongsik Jo, So Yeon Min, Quanting Xie, Taehwan Kim, Yonatan Bisk, Jaegul Choo
• CA-LoRA: Concept-Aware LoRA for Domain-Aligned Segmentation Dataset GenerationMinho Park, Sunghyun Park, Jungsoo Lee, Hyojin Park, Kyuwoong Hwang, Fatih Porikli, Jaegul Choo, Sungha Choi
• OPRO: Orthogonal Panel-Relative Operators for Panel-Aware In-Context Image GenerationSanghyeon Lee, Minwoo Lee, Euijin Shin, Kangyeol Kim, Seunghwan Choi, Jaegul Choo
• Memory-Efficient Fine-Tuning Diffusion Transformer via Dynamic Patch Sampling and Block SkippingSunghyun Park, Jeongho Kim, Hyoungwoo Park, Debasmit Das, Sungrack Yun, Munawar Hayat, Jaegul Choo, Fatih Porikli, Seokeon Choi
• Vector Prism: Animating Vector Graphics by Stratifying Semantic StructureJooyeol Yun, Jaegul Choo
• EgoX: Egocentric Video Generation from a Single Exocentric VideoTaewoong Kang*, Kinam Kim*, Dohyeon Kim*, Minho Park, Junha Hyung, Jaegul Choo
• AHS: Adaptive Head Synthesis via Synthetic Data AugmentationsTaewoong Kang*, Hyojin Jang*, Sohyun Jeong*, Seunggi Moon, Gihwi Kim, Hoon Jin Jung, Jaegul Choo
• Selectively Extracting and Injecting Visual Attributes into Text-to-Image ModelsSeunghwan Choi, Jooyeol Yun, Youngdo Lee, Jaegul Choo

International Conference on Learning Representations

• RL makes MLLMs see better than SFTJunha Song, Sangdoo Yun, Dongyoon Han, Jaegul Choo, Byeongho Heo
• ExpGuard: LLM Content Moderation in Specialized DomainsMinseok Choi*, Dongjin Kim*, Seungbin Yang*, Subin Kim, Youngjun Kwak, Juyoung Oh, Jaegul Choo, Jungmin Son
• LiveWeb-IE: A Benchmark For Online Web Information ExtractionSeungbin Yang*, Jihwan Kim*, Jaemin Choi, Dongjin Kim, Soyoung Yang, ChaeHun Park, Jaegul Choo
• Robust and Interpretable Adaptation of Equivariant Materials Foundation Models via Sparsity-promoting Fine-tuningYoungwoo Cho*, Seunghoon Yi*, Wooil Yang, Sungmo Kang, Young-woo Son, Jaegul Choo, Joonseok Lee, Soo Kyung Kim, Hongkee Yoon
• More Than What Was Chosen: LLM-based Explainable Recommendation Beyond Noisy User PreferencesChung Park, Hyeongjun Yun, Taesan Kim, Junui Hong, Dongjoon Hong, Mira Myong, Jihoon Oh, MinCheol Cho, Kijung Park, Min sung Choi, Jihwan Seok, Jaegul Choo

AAAI Conference on Artificial Intelligence

• Think Wise, Collaborate Effectively: A Rationale-Aware LLM-Based Recommender with Reinforcement Learning from Collaborative SignalsChung Park, Taesan Kim, Hyeongjun Yun, Dongjoon Hong, Junui Hong, Kijung Park, MinCheol Cho, Min sung Choi, Jihwan Seok, and Jaegul Choo
• SphereDiff: Tuning-free 360° Static and Live Wallpaper Generation via Spherical Latent RepresentationMinho Park, Taewoong Kang, Jooyeol Yun, Sungwon Hwang, and Jaegul Choo

International Conference on 3D Vision

• TexAvatars : Hybrid Texel-3D Representations for Stable Rigging of Photorealistic Gaussian Head AvatarsJaeseong Lee*, Junyeong Ahn*, Taewoong Kang, Jaegul Choo

IEEE International Conference on Acoustics, Speech and Signal Processing

• Expression Domain Translation Network for Cross-domain Head ReenactmentKang et al.

AAAI Conference on Artificial Intelligence

• MagiCapture: High-Resolution Multi-Concept Portrait CustomizationHyung et al.
• iDet3D: Towards Efficient Interactive Object Detection for LiDAR Point CloudsChoi et al.
• When Model Meets New Normals: Test-time Adaptation for Unsupervised Time-series Anomaly DetectionKim et al.

Winter Conference on Applications of Computer Vision

• LensNeRF: Rethinking Volume Rendering based on Thin-Lens Camera ModelKim et al.
• Slice and Conquer: A Planar-to-3D Framework for Efficient Interactive Segmentation of Volumetric ImagesCho et al.

Conference on Empirical Methods in Natural Language Processing

• SimCKP: Simple Contrastive Learning of Keyphrase RepresentationsChoi et al.
• AniEE: A Dataset of Animal Experimental Literature for Event ExtractionKim et al.
• Towards Formality-Aware Neural Machine Translation by Leveraging Context InformationKim et al.

Conference on Neural Information Processing Systems

• Learning to Discover Skills through GuidanceKim et al.
• PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement LearningLee et al.

ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems

• FedGeo: Privacy-Preserving User Next Location Prediction with Federated Learning

International Joint Conference on Natural Language Processing and the Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics

• PRiSM: Enhancing Low-Resource Document-Level Relation Extraction with Relation-Aware Score CalibrationChoi et al.
• Learning to Diversify Neural Text Generation via Degenerative ModelHong et al.

ACM International Conference on Information and Knowledge Management

• ST-RAP: A Spatio-Temporal Framework for Real Estate AppraisalLee et al.
• Cracking the Code of Negative Transfer: A Cooperative Game Theoretic Approach for Cross-Domain Sequential RecommendationPark et al.

International Conference on Computer Vision

• Shortcut-V2V: Compression Framework for Video-to-Video Translation based on Temporal Redundancy ReductionChung et al.
• Learning to Generate Semantic Layouts for Higher Text-Image Correspondence in Text-to-Image SynthesisPark et al.
• Towards Open-Set Test-Time Adaptation Utilizing the Wisdom of Crowds in Entropy MinimizationLee et al.
• Label Shift Adapter for Test-Time Adaptation under Covariate and Label ShiftsPark et al.
• FaceCLIPNeRF: Text-driven 3D Face Manipulation using Deformable Neural Radiance FieldsHwang et al.
• CAFA: Class-aware Feature Alignment for Test-time AdaptationJung et al.
• 3D-Aware Generative Model for Improved Side-View Image SynthesisJo et al.

European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases

• Pre-training Contextual Location Embeddings in Personal Trajectories via Efficient Hierarchical Location RepresentationsPark et al.
• Deep Imbalanced Time-series Forecasting via Local Discrepancy DensityPark et al.

Conference on Uncertainty in Artificial Intelligence

• Towards Physically Reliable Molecular Representation LearningYi et al.

• HistRED: A Historical Document-Level Relation Extraction DatasetYang et al.
• Towards Accurate Translation via Semantically Appropriate Application of Lexical ConstraintsBaek et al.
• DEnsity: Open-domain Dialogue Evaluation Metric using Density EstimationPark et al.

International Conference on Machine Learning

• On the Importance of Feature Decorrelation for Unsupervised Representation Learning in Reinforcement LearningLee et al.

IEEE International Conference on Computer Vision and Pattern Recognition

• Local 3D Editing via 3D Distillation of CLIP KnowledgeHyung et al.

IEEE International Conference on Acoustics, Speech and Signal Processing

• Perspective Projection-Based 3D CT Reconstruction from Biplanar X-raysKyung et al.
• Best Student Paper Award

International Conference on Learning Representations

• TTN: A Domain-Shift Aware Batch Normalization in Test-Time AdaptationLim et al.

Conference of the European Chapter of the Association for Computational Linguistics

• PePe: Personalized Post-editing Model Utilizing User-generated Post-editsLee et al.

AAAI Conference on Artificial Intelligence

• Revisiting the Importance of Amplifying Bias for DebiasingLee et al.

Winter Conference on Applications of Computer Vision

• iColoriT: Towards Propagating Local Hint to the Right Region in Interactive Colorization by Leveraging Vision TransformerLee et al.
• CG-NeRF: Conditional Generative Neural Radiance Fields for 3D-aware Image SynthesisJo et al.
• Training Auxiliary Prototypical Classifiers for Explainable Anomaly Detection in Medical Image SegmentationCho et al.
• Guiding Users to Where to Give Color Hints for Efficient Interactive Sketch Colorization via Unsupervised Region PrioritizationCho et al.

Conference on Empirical Methods in Natural Language Processing

• PNEG: Prompt-based Negative Response Generation for Dialogue Response Selection TaskLee et al.
• Rethinking Style Transformer by Energy-based Interpretation: Adversarial Unsupervised Style Transfer using Pretrained ModelCho et al.

• Mining Multi-Label Samples from Single Positive LabelsCho et al.
• WaveBound: Dynamic Error Bounds for Stable Time Series ForecastingCho et al.
• CEDe: A collection of expert-curated datasets with atom-level entity annotations for Optical Chemical Structure RecognitionHormazabal et al.

International Conference on Computational Linguistics

• Reweighting Strategy based on Synthetic Data Identification for Sentence SimilarityKim et al.

Sunghyun wins the Best Paper Award for his paper AnimeCeleb: Large-Scale Animation CelebHeads Dataset for Head Reenactment (Kim et al.) at KAIA Summer Conference 2022.

ACM International Conference on Information and Knowledge Management

• Residual Correction in Real-Time Traffic ForecastingKim et al.

Hojoon wins the Best Short Paper Honorable Mention Award for his paper Towards Validating Long-Term User Feedbacks in Interactive Recommendation Systems (Lee et al.) at SIGIR 2022.

European Conference on Computer Vision

• Towards Accurate Open-Set Recognition via Background-Class RegularizationCho & Choo
• High-Resolution Virtual Try-On with Misalignment and Occlusion-Handled ConditionsLee et al.
• Style Your Hair: Latent Optimization for Pose-Invariant Hairstyle Transfer via Local-Style-Aware Hair AlignmentKim et al.
• AnimeCeleb: Large-Scale Animation CelebHeads Dataset for Head ReenactmentKim et al.

ACM International Conference on Multimedia

• ReFu: Refine and Fuse the Unobserved View for Detail-Preserving Single-Image 3D Human ReconstructionShim et al.

International Conference on Medical Image Computing and Computer Assisted Intervention

• Morphology-Aware Interactive Keypoint EstimationKim et al.

IEEE Conference on Games

• Enemy Spotted: In-game Gun Sound Dataset for Gunshot Classification and LocalizationPark et al.

International ACM SIGIR Conference on Research and Development in Information Retrieval

• Towards Validating Long-Term User Feedbacks in Interactive Recommendation SystemsLee et al.

EG/VGTC Conference on Visualization

• DASH: Visual Analytics for Debiasing Image Classification via User-Driven Synthetic Data AugmentationKwon et al.

Congratulations to Prof. Jaegul Choo for becoming a tenured professor at KAIST! We got together to celebrate him and listened to his heartfelt speech :)

Letsur, a startup led by Gyuhyeon, wins one of the top 20 spots for Samsung's C-Lab Outside program and receives various business supports from Samsung Research. Check out the news article.

International Conference on Learning Representations

• Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution ShiftKim et al.

Pacific-Asia Conference on Knowledge Discovery and Data Mining

• PASTA: Parallel Spatio-Temporal Attention with Spatial Auto-Correlation Gating for Fine-Grained Crowd Flows PredictionPark et al.

International World Wide Web Conference

• DraftRec: Personalized Draft Recommendation for Victory in Multi-Player Online Battle Arena GamesLee et al.

Hyesu, Jeonghoon, and Sanghyeon win the Best Poster Award for their papers AVoCaDo: Strategy for Adapting Vocabulary to Downstream Domain (Hong et al.) and Deep Edge-Aware Interactive Colorization against Color-Bleeding Effects (Kim et al.) at the 2022 KAIST AI Workshop!

British Machine Vision Conference

• Continuous-Time Video Generation via Learning Motion Dynamics with Neural ODEKim et al.
• Improving Face Recognition with Large Age Gaps by Learning to Distinguish ChildrenLee et al.
• HairFIT: Pose-Invariant Hairstyle Transfer via Flow-based Hair Alignment and Semantic-Region-Aware InpaintingChung et al.

Conference on Neural Information Processing Systems

• Learning Debiased Representation via Disentangled Feature AugmentationLee et al.

Asian Conference on Machine Learning

• Augmenting Imbalanced Time-series Data via Adversarial Perturbation in Latent SpaceKim et al.

• AVoCaDo: Strategy for Adapting Vocabulary to Downstream DomainHong et al.
• Novel Natural Language Summarization of Program Code via Leveraging Multiple Input RepresentationsChen et al.

International Conference on Computer Vision

• BiaSwap: Removing Dataset Bias with Bias-Tailored Swapping AugmentationKim et al.
• Deep Edge-Aware Interactive Colorization against Color-Bleeding EffectsKim et al.
• Standardized Max Logit: A Simple yet Effective Approach for Identifying Unexpected Road Obstacles in Urban-Scene SegmentationJung et al.

Machine Learning for Healthcare

• Knowledge Graph-based Question Answering with Electronic Health RecordsPark et al.
• intro talk

IEEE International Conference on Image Processing

• K-Hairstyle: A Large-Scale Korean Hairstyle Dataset for Virtual Hair Editing and Hairstyle ClassificationKim et al.
• project page

• Unsupervised Neural Machine Translation for Low-Resource Domains via Meta-LearningPark et al.
• Constructing Multi-Modal Dialogue Dataset with Text-to-Image ReplacementLee et al.
• Learning to Generate Questions by Learning to Recover Answer-containing SentencesBack et al.

EG/VGTC Conference on Visualization

• VATUN: Visual Analytics for Testing and Understanding Convolutional Neural NetworksPark et al.
• demo
• talk

ACM Conference on Computer-Supported Cooperative Work and Social Computing

• Understanding Human-side Impact of Sampling Image Batches in Subjective Attribute LabelingChung et al.

Annual Conference of the North American Chapter of the Association for Computational Linguistics

• Restoring and Mining the Records of the Joseon Dynasty via Neural Language Modeling and Machine TranslationKang et al.

IEEE International Conference on Computer Vision and Pattern Recognition

• Not Just Compete, But Collaborate: Local Image-to-Image Translation via Cooperative Mask PredictionKim et al.
• VITON-HD: High-Resolution Virtual Try-On via Misalignment-Aware NormalizationChoi et al.
• RobustNet: Improving Domain Generalization in Urban-Scene Segmentation via Instance Selective WhiteningChoi et al.
• code
• demo
• code
• talk

IEEE International Conference on Acoustics, Speech and Signal Processing

• Efficient Adversarial Audio Synthesis via Progressive UpsamplingCho et al.

AAAI Conference on Artificial Intelligence

• Vid-ODE: Continuous-Time Video Generation with Neural Ordinary Differential EquationPark et al.
• code

IEEE International Conference on Data Engineering

• An Empirical Experiment on Deep Learning Models for Predicting Traffic DataLee et al.

British Machine Vision Conference

• Multimodal Image Translation with Stochastic Style Representations and Mutual Information LossNa et al.

ACM International Conference on Information and Knowledge Management

• ST-GRAT: Spatio-Temporal Graph Attention Network for Traffic ForecastingPark et al.

International Conference on Machine Learning

• Learning De-biased Representations with Biased RepresentationsBahng et al.
• code
• talk

International Conference on Pattern Recognition

• End-To-End Multi-Task Learning of Missing Value Imputation and Forecasting in Time-Series DataKim et al.

IEEE International Conference on Computer Vision and Pattern Recognition

• Exploring Unlabeled Faces for Novel Attribute DiscoveryBahng et al.
• Reference-based Sketch Image Colorization using Augmented-Self Reference and Dense Semantic CorrespondenceLee et al.
• Cars Can’t Fly Up in the Sky: Improving Urban-Scene Segmentation via Height-driven Attention NetworksChoi et al.
• intro
• code

International Conference on Learning Representations

• NeurQuRI: Neural Question Requirement Inspector for Answerability Prediction in Machine Reading ComprehensionBack et al.

Conference on Empirical Methods in Natural Language Processing

• NL2pSQL: Generating Pseudo-SQL Queries from Under-specified Natural Language QuestionChen et al.

ACM International Conference on Information and Knowledge Management

• Recommender System Using Sequential and Global Preference via Attention Mechanism and Topic ModelingKang et al.

• SANVis: Visual Analytics for Understanding Self-Attention NetworksPark et al.
• demo

British Machine Vision Conference

• Learning to Focus and Track Extreme Climate EventsKim et al.

IEEE International Conference on Computer Vision and Pattern Recognition

• Coloring With Limited Data: Few-Shot Colorization via Memory-Augmented NetworksYoo et al.
• Image-to-Image Translation via Group-wise Deep Whitening and ColoringCho et al.
• project page
• talk

ACM CHI Conference on Human Factors in Computing Systems

• AILA: Attentive Interactive Labeling Assistant for Document Classification through Attention-Based Deep Neural NetworksChoi et al.
• intro
• full demo

Conference on Empirical Methods in Natural Language Processing

• MemoReader: Large-Scale Reading Comprehension through Neural Memory ControllerBack et al.

European Conference on Computer Vision

• Coloring with Words: Guiding Image Colorization Through Text-based Palette GenerationBahng et al.
• code

IEEE International Conference on Computer Vision and Pattern Recognition

• StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image TranslationChoi et al.
• code
• video

ACM CHI Conference on Human Factors in Computing Systems

• TopicOnTiles: Tile-Based Spatio-Temporal Event Analytics via Exclusive Topic Modeling on Social MediaChoi et al.
• intro
• full demo