PERSONAL DETAILS
CAREER HISTORY
| Jan. 2025 – Current |
Senior researcher in Electronics and Telecommunications Research Institute (ETRI), Artificial Intelligence Computing Research Laboratory / AI SoC Research Division / PIM AI SoC Research Section |
EDUCATION HISTORY
| Mar. 2019 – Feb. 2025 | Ph.D. in School of Computing, KAIST
Advisor: Soontae Kim [LINK] |
| --- | --- |
| Mar. 2017 – Feb. 2019 | Master in School of Computing, KAIST
Advisor: Soontae Kim [LINK] |
| Mar. 2011 – Feb. 2017 | Bachelor in Computer Engineering, SKKU
Advisor: Yungyung Cheong [LINK] |
PUBLICATIONS AND PATENTS
- 장명재, 이상호, 김지은, 최재웅, 박건희, 권현정, 천민규, 조용철, 최민석, 한진호. (2026, June). NPU 내 LLM 추론 가속을 위한 LLM 파라미터 최적화 방안. 대한전자공학회 하계학술대회 (pp. ). (Korean) [LINK]
- (Patent) MX format 양자화를 고려한 3-Stage LLM 데이터 최적화 기법, MX Format Quantization-based 3-Stage LLM Data Optimization Method, 2026-03-26, 2026-0054873, (KR, US).
한국전자통신연구원, 장명재, 최재웅, 한진호
- (Patent) Smart DMA 기반 Multi-Cast 지원 Multi-Cast Network-on-Chip(NoC) 구조, Smart DMA-based Multi-Cast Support Multi-Cast Network-on-Chip (NoC) Architecture, 2025-12-09, 2025-0194439, (KR, JP, US).
한국전자통신연구원, 한진호, 김주엽, 권현정, 박건희, 장명재, 정재훈, 최재웅, 황근영
- (Patent) 시스톨릭 어레이 기반 뉴럴 네트워크 하드웨어 가속 방법 및 장치, A method and apparatus for acceleration of neural networks based on systolic array, 2025-12-08, 2025-0192904, (KR, US).
한국전자통신연구원, 김진규, 권현정, 김주엽, 장명재, 한진호
- 장명재, 권현정, 최재웅, 천민규, 한진호. (2025, Oct.). 칩렛 이종집적 첨단패키기 기반 LLM 가속기 설계 동향. Electronics and Telecommunications Trends (pp. 13-21). ETRI. (Korean) [LINK]
- Jang, M., Kim, J., Nam, H., Kim, S., & Kim, S. (2025, March). C2C: A Framework for Critical Token Classification in Transformer-based Inference Systems. In 2025 Design, Automation & Test in Europe Conference & Exhibition (DATE) (pp. 1-2). IEEE. [LINK]
- (Ph.D. Thesis) Jang, Myeongjae. "An Efficient and Secure Deep Neural Networks System Driven by Data Characteristics." (2025). [LINK (KAIST Library)]
- Jang, M., Kim, J., Nam, H., & Kim, S. (2023). Zero and Narrow-Width Value-Aware Compression for Quantized Convolutional Neural Networks. IEEE Transactions on Computers (TC). [LINK]
- Kim, J., Jang, M., Nam, H., & Kim, S. (2023, October). HARP: Hardware-Based Pseudo-Tiling for Sparse Matrix Multiplication Accelerator. In Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO) (pp. 1148-1162). [LINK]
- Jang, M., Kim, J., Kim, J., & Kim, S. (2022, March). Encore compression: Exploiting narrow-width values for quantized deep neural networks. In 2022 Design, Automation & Test in Europe Conference & Exhibition (DATE) (pp. 1503-1508). IEEE. [LINK]
- Jang, M., & Hong, J. (2021). MATE: Memory-and Retraining-Free Error Correction for Convolutional Neural Network Weights. Journal of information and communication convergence engineering, 19(1), 22-28. [LINK]
- (Master Thesis) Jang, Myeongjae. "Neural network-based FMCW radar system for detecting and tracking a drone." (2019). [LINK]
- (Patent) 인공 신경망 기반 FMCW 레이다 드론 추적 기법, OBJECT DETECTION METHOD AND APPARATUS OF RADAR SYSTEM, 2021-06-22, 10-2269732-0000, (Korean). [LINK]
한국과학기술원, 김순태, 장명재, 강민철, 이원영