Neural network technology for pattern recognition, stock prediction and market forecasting

Matlab MEX Interface to Assembler-based Neural Network Training and Simulation

Download now Matlab source code
Requirements: Matlab.

ANNS is a neural network simulator based on assembler language with a simple and intuitive Matlab interface. ANNS package dynamically generates ASM code, automatically embedding it into Matlab MEX file (Matlab C interface) for a fast and highly optimized code for neural network training and simulation. ANNS package is particularly suitable for pattern recognition tasks, when speed becomes the most relevant issue. ANNS package implements the backpropagation algorithm and includes:

• arbitrary length of input feature vectors
• one hidden layer (arbitrary number of hidden neurons)
• hyperbolic tangent activation function on the hidden layer
• linear activation function on the output layer
• scalar output

ANNS package uses the Netwide Assembler.
The Netwide Assembler is available at http://sourceforge.net/projects/nasm.

Index Terms: Matlab, assembler, Mex, c, interface, neural network, asm, Netwide Assembler, ann, artificial neural networks, nn.

Release 1.0 Date 2007.01.18
Major features:

• Neural Network training and simulation based on the back-propagation algorithm
• Easy and intuitive MEX C interface
• C code included
• Assembler code included
• Demo code (protected P-files) available for performance evaluation