EUVIMEDEUVIMED entdecken
Open Access

Deep random forest with ferroelectric analog content addressable memory

Xunzhao Yin iD, Franz Müller iD, Ann Franchesca Laguna, Chao Li iD, Qingrong Huang, Zhiguo Shi iD, Maximilian Lederer iD, Nellie Laleni, Shan Deng, Zijian Zhao iD, Mohsen Imani iD, Yiyu Shi und 5 weitere

Science Advances · 07.06.2024 · 10 (23)

DOI: 10.1126/sciadv.adk8471 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science Advances, erschienen am 07.06.2024. Verfasst von Xunzhao Yin und 16 weiteren Autor:innen. Beteiligte Einrichtung: Zhejiang University, Hangzhou, Zhejiang, China. Die Arbeit wurde bisher 36-mal zitiert (Crossref) und nennt 61 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep random forest (DRF), which combines deep learning and random forest, exhibits comparable accuracy, interpretability, low memory and computational overhead to deep neural networks (DNNs) in edge intelligence tasks. However, efficient DRF accelerator is lagging behind its DNN counterparts. The key to DRF acceleration lies in realizing the branch-split operation at decision nodes. In this work, we propose implementing DRF through associative searches realized with ferroelectric analog content addressable memory (ACAM). Utilizing only two ferroelectric field effect transistors (FeFETs), the ultra-compact ACAM cell performs energy-efficient branch-split operations by storing decision boundaries as analog polarization states in FeFETs. The DRF accelerator architecture and its model mapping to ACAM arrays are presented. The functionality, characteristics, and scalability of the FeFET ACAM DRF and its robustness against FeFET device non-idealities are validated in experiments and simulations. Evaluations show that the FeFET ACAM DRF accelerator achieves ∼10 6 ×/10× and ∼10 6 ×/2.5× improvements in energy and latency, respectively, compared to other DRF hardware implementations on state-of-the-art CPU/ReRAM.

Autor:innen und Institutionen

  • Zhejiang University, Hangzhou, Zhejiang, China.
  • Key Laboratory of CS&AUS of Zhejiang Province, Hangzhou, China.
  • Fraunhofer IPMS, Dresden, Germany.
  • De La Salle University, Manila, Philippines.
  • University of Notre Dame, Notre Dame, IN 46614, USA.
  • University of California, Irvine, CA 92697, USA.

ORCID-Profile: Xunzhao Yin, Franz Müller, Chao Li, Zhiguo Shi, Maximilian Lederer, Zijian Zhao, Mohsen Imani, Xiaobo Sharon Hu, Cheng Zhuo, Thomas Kämpfe, Kai Ni

Zitieren

APA

Yin, X., Müller, F., Laguna, A. F., Li, C., Huang, Q., Shi, Z., Lederer, M., Laleni, N., Deng, S., Zhao, Z., Imani, M., Shi, Y., Niemier, M., Hu, X. S., Zhuo, C., Kämpfe, T., & Ni, K. (2024). Deep random forest with ferroelectric analog content addressable memory. Science Advances, 10(23). https://doi.org/10.1126/sciadv.adk8471

Vancouver

Yin X, Müller F, Laguna AF, Li C, Huang Q, Shi Z, et al. Deep random forest with ferroelectric analog content addressable memory. Science Advances. 2024;10(23). doi:10.1126/sciadv.adk8471

BibTeX
@article{Yin2024,
  author = {Yin, Xunzhao and Müller, Franz and Laguna, Ann Franchesca and Li, Chao and Huang, Qingrong and Shi, Zhiguo and Lederer, Maximilian and Laleni, Nellie and Deng, Shan and Zhao, Zijian and Imani, Mohsen and Shi, Yiyu and Niemier, Michael and Hu, Xiaobo Sharon and Zhuo, Cheng and Kämpfe, Thomas and Ni, Kai},
  title = {Deep random forest with ferroelectric analog content addressable memory},
  journal = {Science Advances},
  year = {2024},
  volume = {10},
  number = {23},
  doi = {10.1126/sciadv.adk8471}
}
Kostenlos starten →