Toward Neuromorphic Chips at Scale
Lists
Only you see this
Abstract
Neuromorphic computing aims to mimic the brain to build energy-efficient and adaptive systems. It offers an alternative to traditional AI by using event-driven, spike-based models. Recent advances have produced powerful chips, but scaling these systems remains a major challenge. This paper surveys the core features that enable scalable neuromorphic systems. We discuss architectures, learning algorithms, and communication models that support large-scale operation. Key topics include sparsity, dynamic reconfigurability, asynchronous processing, and heterogeneous integration. We also examine the current limitations in software tools and standardization. Usability, interoperability, and integration with sensors are major barriers to adoption. The paper outlines future directions for research in scalable learning, unified toolchains, and real-world deployment. Our goal is to guide researchers toward building neuromorphic systems that can scale efficiently, adapt in real time, and solve complex tasks across domains.
BibTeX
@misc{lamaakal2025toward,
author = {Ismail Lamaakal and Chaymae Yahyati and Ibrahim Ouahbi and Khalid El Makkaoui and Yassine Maleh},
title = {{Toward Neuromorphic Chips at Scale}},
year = {2025},
doi = {10.1109/icesa66763.2025.11280943},
}Loading related content...