https://doi.org/10.1140/epjs/s11734-025-01750-5
Regular Article
Memristor-based convolutional neural network for efficient image recognition on edge computing devices
1
College of Electrical Engineering and Automation, Shandong University of Science and Technology, 266590, Qingdao, Shandong, China
2
Beijing Orient Institute of Measurement and Test, 100094, Beijing, China
3
Doctoral Workstation, Guangdong Songshan Polytechnic College, 512126, Shaoguan, Guangdong, China
Received:
27
February
2025
Accepted:
3
June
2025
Published online:
30
June
2025
Traditional von Neumann architecture-based Internet of Things (IoT) edge devices cannot meet the computational resource demands of convolutional neural networks (CNNs). To improve computational efficiency, a complete convolutional neural network circuit based on memristor crossbar arrays is constructed, which is suitable for image recognition in edge computing devices. The memristor convolutional neural network (MCNN) circuit encompasses convolutional layers, fully connected layers, activation function circuits, and pooling layers. Moreover, the robustness of the MCNN circuit is tested and analyzed under different conductance error levels, Gaussian noise levels, and pretzel noise levels. Finally, the results show that the MCNN circuit exhibits good fault tolerance and achieves efficient and low-power image recognition on edge devices.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2025
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
