A new suite of three algorithms, implementing a Supervised Machine Learning Workflow based on Feed Forward Neural Networks, is now available. It can be used for general Classification and Regression tasks as well as correlation studies and biological simulations requiring some kind of information extraction from labelled data.
The three algorithms are briefly described below:
- Feed Forward Neural Network Trainer: The algorithm trains a Feed Forward Artificial Neural Network on a given dataset using an online Back Propagation procedure and returns the training error and a binary file containing the trained network. It is possible to specify the following simulation parameters: learning rate, training error threshold, maximum number of training iterations, number and dimensions of hidden layers. Try it now!
- Feed Forward Neural Network Regressor: This algorithm simulates a real-valued vector function using a trained Feed Forward Artificial Neural Network and returns a table containing the function actual inputs and the predicted outputs. Try it now!
- Feed Forward Neural Network Cloud Regressor: This algorithm implements the same procedure as the previous one, but exploits the cloud computing potential provided by DataMiner in order to carry out faster calculations, splitting the input dataset into chunks and processing them in parallel. Try it now!