publications

2026

  1. QFlash: Bridging Quantization and Memory Efficiency in Vision Transformer Attention
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    In International Joint Conference on Artificial Intelligence (IJCAI), Aug 2026
    To appear

2025

  1. Luthier: Bridging Auto-Tuning and Vendor Libraries for Efficient Deep Learning Inference
    Yongin Kwon, JooHyoung Cha, Sehyeon Oh, and 3 more authors
    ACM Transactions on Embedded Computing Systems, Sep 2025
    Presented at ESWEEK CASES 2025 (journal track)
  2. Work-in-Progress: I-FlashAttention: Fully Integer Fused Attention for Efficient Vision Transformers
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    In International Conference on Compilers, Architecture, and Synthesis for Embedded Systems (CASES), Sep 2025
  3. TriPlanNet: Triangle Path Planning Network for A Variable Truss Robot with Deep Learning
    Choonghan Lee, Leah Harris, Sehyun Oh, and 4 more authors
    In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, Oct 2025
    13th International Workshop on Assistive Computer Vision and Robotics (ACVR)
  4. Data Transmission Optimization for NPU System Scalability
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    IEMEK Journal of Embedded Systems and Applications, Jun 2025
  5. Optimizing Real-Time Object Detection in a Multi-Neural Processing Unit System
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    Sensors, Mar 2025
    Invited extension of our AICompS 2024 Distinguished Paper

2024

  1. AICompS
    Optimizing Real-Time Object Detection in a Multi NPU System with Double Buffering and Queue-Based Processing
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    In The 1st International Conference on Artificial Intelligence Computing and Systems (AICompS), Dec 2024
  2. PCIe-Based Multi-NPU Data Transmission Optimization
    Sehyeon Oh, Yongin Kwon, and Jemin Lee
    In IEMEK Autumn Conference, Nov 2024