KIP publications

year 2006
author(s) J. Fieres, J. Schemmel, K. Meier
title Training convolutional neural networks of threshold neurons suited for low-power hardware implementation
KIP-Nummer HD-KIP 06-21
KIP-Gruppe(n) F9
document type Paper
Keywords (shown) threshold neurons, artificial neural networks, low-power hardware, convolutional neural networks, image recognition
source Proceedings of the 2006 International Joint Conference onNeural Networks (IJCNN 2006), 21--28, IEEE Press (2006)
Abstract (en)

Convolutional neural networks are known to be powerful image classifiers. In this work, a method is proposed for training convolutional networks for implementation on an existing mixed digital-analog VLSI hardware architecture. The binary threshold neurons provided by this architecture cannot trained using gradient-based methods. The convolutional layers are trained with a clustering method, locally in each layer. The output layer is trained using the Perceptron learning rule. Competitive results are obtained on hand-written digits MNIST) and traffic signs. The analog hardware enables high integration and low power consumption, but inherent error sources affect the computation accuracy. Networks trained as suggested are highly robust against random changes of synaptic weights occuring on the hardware substrate, and work well even with only three distinct weight values (-1, 0, +1), reducing computational complexity to mere counting.

Datei article as pdf file
KIP - Bibliothek
Im Neuenheimer Feld 227
Raum 3.402
69120 Heidelberg