A grid based neural network execution service
Leander Krammer, Erich Schikuta, Helmut Wanek · International Conference on Parallel and Distributed Computing and Networks · 2006
Grid Computing needs some application scenarios proved by operative implementations that should lead towards profitable real business applications. NeuroGrid represents the first step of a long way to connect the immense area of neural networks and its high potential applications to the fast emerging infrastructure possibilities of Grid computing. Based on the de-facto standard Grid middleware Globus Toolkit and the platform independent Java technology NeuroGrid demonstrates a proof-of-concept implementation for training and evaluating neural networks within a Grid environment.