Dasom Ahn (Tommy Ahn)

I am a Ph.D. student in Computer Science and Engineering at Keimyung University, working on explainable visuomotor policy learning, with research spanning concept bottleneck models, diffusion policies, and multimodal representation learning.

Education

• Ph.D., School of Computer Engineering, Keimyung University (2023.09 ~ ) - Advisor: Prof. Byoung Chul Ko
• M.S., School of Computer Engineering, Keimyung University (2021.09 ~ 2023.08) - Advisor: Prof. Byoung Chul Ko
• B.S., School of Computer Engineering, Keimyung University (2013.03 ~ 2018.02)

Research Interest

• Deep Learning, Computer Vision
• Video understanding (Action recognition, Temporal action detection, etc.)
• Energy-based Models
• eXplainable AI (Concept-based, Transformer-based, etc.)
• Visuomotor policy (Diffusion policies, etc.)

News

• [2026/06] A paper is accepted by ICRA 2026.
• [2025/11] A paper is published to Neurocomputing.
• [2025/10] A paper is accepted by ICCVW 2025.
• [2024/12] A paper is accepted by ACCV 2024.
• [2024/07] A paper is accepted by ICML 2024.
• [2023/10] A paper is published to IEEE Access 2024.
• [2023/10] A paper is accepted by ICML 2024.
• [2023/10] A paper is published to Springer.
• [2023/10] A paper is accepted by ICCV 2023.
• [2023/08] A paper is published to Springer Applied Intelligence.
• [2023/01] STAR-Transformer is accepted to WACV 2023.
• [2022/07] Shift-ViT is accepted to ITC-CSCC 2022.

Professional Service

• Reviewer: CVPR, ICLR, ECCV, ICML

Publications

(* indicates equal contribution)

Conference Papers

• Shifted Flow Policy: Uncertainty-aware Time Reparameterization for Visuomotor Learning
  Dasom Ahn, Chanhyuk Jung, Joonki Baek, Sungkeun Yoo, Byoung Chul Ko*
  IEEE International Conference on Robotics and Automation, 2026.
  [CVF]

• Flow-Guided Policies: Overcoming Diffusion Limitations for Robust Robot Imitation Learning
  Chanhyuk Jung, Sangwon Kim, Kwang-Ju Kim, Dasom Ahn, Joonki Baek, Sungkeun Yoo, Byoung Chul Ko
  International Conference on Computer Vision Workshops, 2025.
  CVF

• EQ-CBM: A Probabilistic Concept Bottleneck with Energy-based Models and Quantized Vectors
  Sangwon Kim, Dasom Ahn, Byoung Chul Ko, In-su Jang, Kwang-Ju Kim
  Asian Conference on Computer Vision, 2024.
  CVF

• Scene Graph Generation Strategy with Co-occurrence Knowledge and Learnable Term Frequency
  Hyeongjin Kim, Sangwon Kim, Dasom Ahn, Jong Taek Lee, Byoung Chul Ko
  International Conference on Machine Learning, 2024.
  arXiv

• Cross-Modal Learning with 3D Deformable Attention for Action Recognition
  Sangwon Kim, Dasom Ahn, Byoung Chul Ko
  IEEE/CVF International Conference on Computer Vision, 2023.
  arXiv

• STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition
  Dasom Ahn, Sangwon Kim, Hyunsu Hong, Byoung Chul Ko
  IEEE/CVF Winter Conference on Applications of Computer Vision, 2023.
  arXiv · CVF

• Shift-ViT: Siamese Vision Transformer using Shifted Branches
  Dasom Ahn, Hyeongjin Kim, Sangwon Kim, Hyunsu Hong, Byoung Chul Ko
  International Technical Conference on Circuits/Systems, Computers and Communications, 2022.
  IEEE Xplore

Journal Papers

• LAttE: A Label-Free and Multimodal Framework for Context-Aware Person Re-Identification
  Dasom Ahn, Sangwon Kim, Kwang-Ju Kim, Byoung Chul Ko
  Neurocomputing (Elsevier), 2025.
  ScienceDirect

• Interpretable Information Visualization for Enhanced Temporal Action Detection in Videos
  Dasom Ahn, Jong-Ha Lee, Byoung Chul Ko
  IEEE Access, 2024.
  IEEE Xplore

• BTD-RF: 3D Scene Reconstruction Using Block-Term Tensor Decomposition
  Seon Bin Kim, Sangwon Kim, Dasom Ahn, Byoung Chul Ko
  Applied Intelligence (Springer), 2024.

• STAR++: Rethinking Spatio-Temporal Cross Attention Transformer for Video Action Recognition
  Dasom Ahn, Sangwon Kim, Byoung Chul Ko
  Applied Intelligence (Springer), 2023.
  Springer

Domestic Papers

• Dasom Ahn, Sung Jin Kim, Byoung Chul Ko,
  “시간 기반 개념 표현을 활용한 설명가능한 로봇 정책 확산 제어 모델,”
  제38회 영상처리 및 이해에 관한 워크샵 (IPIU 2026), Feb. 2026.

• Dasom Ahn, Sangwon Kim, Byoung Chul Ko,
  “설명 가능한 의사결정을 위한 에너지 기반 개념 병목 모델 설계,”
  제36회 영상처리 및 이해에 관한 워크샵 (IPIU 2024), Feb. 2024.

• Dasom Ahn, Byoung Chul Ko,
  “향상된 시간 특징 학습을 사용한 새로운 시간적 행동 인식 및 지역화 시스템,”
  제33회 인공지능신호처리합동학술대회 (AISP 2023), Sep. 2023.

• Dasom Ahn, Sangwon Kim, Byoung Chul Ko,
  “Vision Transformer 기반의 동적 얼굴 표정 인식 시스템,”
  2021 한국소프트웨어종합학술대회 (KSC 2021), Dec. 2021.

• Dasom Ahn, Sangwon Kim, Byoung Chul Ko,
  “교차 주의집중 트랜스포머를 이용한 인간 행동 인식 시스템,”
  2022 대한전자공학회 하계종합학술대회 (IEIE), pp. 839–840, Jun. 2022.
  Best Paper Award

Patents

• 고병철, 안다솜, 김상원, 변이 브랜치를 사용한 샴 비전 변환기에 의한 이미지 분류 방법 및 장치 (IMAGE CLASSIFICATION METHOD AND APPARATUS BY SIAMESE VISION TRANSFORMER USING SHIFTED BRANCHES), 등록번호(일자) 1028922120000 (2025.11.24) (연구재단, 지역대학우수,2025 종료과제)
• 고병철, 안다솜, 비전 트랜스포머 기반의 얼굴 표정 인식장치 및 방법 (VISION TRANSFORMER-BASED FACIAL EXPRESSION RECOGNITION APPARATUS AND METHOD), 등록번호/일자 10-2780363 (2025.03.07) (연구재단, 지역대학우수,2019 종료과제)
• 고병철, 안다솜, 시공간 교차 주의집중 트랜스포머를 이용한 인간 행동 인식 시스템 및 방법(SPATIO-TEMPORAL CROSS ATTENTION TRANSFORMER SYSTEM AND METHOD FOR HUMAN ACTION RECOGNITION), 등록번호/일자 1027263850000 (2024.10.31) (연구재단, 지역대학우수,2025 종료과제)

Honor & Awards

• 2023 Best Paper: AISP, S.Korea
• 2022 Best Paper: IEIE, S.Korea
• 2016 Best Paper: WISET, S.Korea