Project

Snowfort is an open source data analytics system for wireless sensor network. It uses previously available open source hardware with a new open source communication scheme to allow for simple reliable, and flexible wireless sensor networks. It is developed by the faculties and students at Stanford University. The goal of this project is helping developers and researchers to rapidly develop a network for sensing without the time or background to develop the entire system for scratch.

Snowfort focus on:

• Reliable data collection for environment and infrastructure

• Standarized data management

• Simple implementation with scalability

• Real-time data analytics and visualization

[More Information]

We proposed a new multilayer artificial neural network model that is named No-Propagation (No-Prop). In this new model, the weights of the hidden-layer neurons are fixed with random values. Only the weights of the output-layer neurons are trained, using steepest descent to minimize mean squared error, with the LMS algorithm of Widrow and Hoff. The computational complexity of No-Prop is simpler than that of Back Propagation (Back-Prop) model. In addition, the performance of No-Prop is comparable with Back-Prop.

This Matlab class an optimized implementation of the No-prop Neural Network, an adaptive neural network model developed by Prof. Bernard Widrow's research group at Stanford University. The user can define the parameters and run the data for training and testing. In addition, the vectorized implementation of Back Propagation is included.