Final report for Cognitive Computing for Security LDRD 165613. It reports on the development of hybrid of general purpose/ne uromorphic computer architecture, with an emphasis on potential implementation with memristors.
Proceedings of the International Joint Conference on Neural Networks
Some next generation computing devices may consist of resistive memory arranged as a crossbar. Currently, the dominant approach is to use crossbars as the weight matrix of a neural network, and to use learning algorithms that require small incremental weight updates, such as gradient descent (for example Backpropagation). Using real-world measurements, we demonstrate that resistive memory devices are unlikely to support such learning methods. As an alternative, we offer a random search algorithm tailored to the measured characteristics of our devices.
Abstract not provided.
Abstract not provided.
Abstract not provided.
This report discusses aspects of neuromorphic computing and how it is used to model microsystems.