Project Description
Resonance is a system to train neural networks, it allows to automate the train of a neural network, distributing the calculation on multiple machines.

Main Components
  • TrainNode: is the service that actually performs the processing of learning algorithms. Train Node can be installed on multiple machines, each node performs the work assigned by the Resonance Server. The main server and nodes communicate via WCF services.
  • Resonance Server: is the core service that coordinates the work of processing nodes and tracks all data related to neural networks managed by the system. Client applications communicate with the server to run the learning process of neural networks and to manage the archive of neural networks stored in the database. The data access layer is implemented using Linq to SQL, the interface is exposed as a WCF service, this allows complete interoperability with other operating systems and development environments. The server also comes with a management console that lets you configure the system (defining the Train Nodes) and monitor the main parameters of the system.
  • Client: users can develop client applications with their preferred languages and development environments, given that the services of Resonance Server are exposed as SOAP web services.
  • Currently, for training the network, using the framework AForge.

Last edited Jun 10, 2010 at 10:20 AM by carlobruscolotti, version 10