EUVIMEDEUVIMED entdecken
Crossref

A Multi-Bit ECRAM-Based Analog Neuromorphic System With High-Precision Current Readout Achieving 97.3% Inference Accuracy

Minseong Um, Minil Kang, Kyeongho Eom, Hyunjeong Kwak, Kyungmi Noh, Jimin Lee, Jeonghoon Son, Jiseok Kwon, Seyoung Kim, Hyung-Min Lee

IEEE Transactions on Biomedical Circuits and Systems · 2025-06 · 19 (3) · 590-604

DOI: 10.1109/tbcas.2024.3465610 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Biomedical Circuits and Systems, 2025-06. Verfasst von Minseong Um und 9 weiteren Autor:innen. Hauptthemen laut MEDLINE: Neural Networks, Computer, Signal Processing, Computer-Assisted. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 37 Literaturangaben. Gefördert von: Technology Innovation Program, National Research Foundation of Korea.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article proposes an analog neuromorphic system that enhances symmetry, linearity, and endurance by using a high-precision current readout circuit for multi-bit nonvolatile electro-chemical random-access memory (ECRAM). For on-chip training and inference, the system uses activation modules and matrix processing units to manage analog update/read paths and perform precise output sensing with feedback-based current scaling on the ECRAM array. The 250nm CMOS neuromorphic chip was tested with a 32 × 32 ECRAM synaptic array, achieving linear and symmetric updates and accurate read operations. The proposed circuit system updates the 32 × 32 ECRAM across 100 levels, maintaining consistent synaptic weights, and operates with an output error rate of up to 2.59% per column. It consumes 5.9 mW of power excluding the ECRAM array and achieves 97.3% inference accuracy on the MNIST dataset, close to the software-confirmed 97.78%, with only the final layer (64 × 10) mapped to the ECRAM.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerSignal Processing, Computer-AssistedEquipment DesignHumans

Förderung

  • Technology Innovation Program
  • National Research Foundation of Korea

Zitieren

APA

Um, M., Kang, M., Eom, K., Kwak, H., Noh, K., Lee, J., Son, J., Kwon, J., Kim, S., & Lee, H. M. (2025). A Multi-Bit ECRAM-Based Analog Neuromorphic System With High-Precision Current Readout Achieving 97.3% Inference Accuracy. IEEE Transactions on Biomedical Circuits and Systems, 19(3), 590-604. https://doi.org/10.1109/tbcas.2024.3465610

Vancouver

Um M, Kang M, Eom K, Kwak H, Noh K, Lee J, et al. A Multi-Bit ECRAM-Based Analog Neuromorphic System With High-Precision Current Readout Achieving 97.3% Inference Accuracy. IEEE Transactions on Biomedical Circuits and Systems. 2025;19(3):590-604. doi:10.1109/tbcas.2024.3465610

BibTeX
@article{Um2025,
  author = {Um, Minseong and Kang, Minil and Eom, Kyeongho and Kwak, Hyunjeong and Noh, Kyungmi and Lee, Jimin and Son, Jeonghoon and Kwon, Jiseok and Kim, Seyoung and Lee, Hyung-Min},
  title = {A Multi-Bit ECRAM-Based Analog Neuromorphic System With High-Precision Current Readout Achieving 97.3% Inference Accuracy},
  journal = {IEEE Transactions on Biomedical Circuits and Systems},
  year = {2025},
  volume = {19},
  number = {3},
  pages = {590--604},
  doi = {10.1109/tbcas.2024.3465610}
}
Kostenlos starten →