Skip to content

Files

Latest commit

 

History

History
6 lines (4 loc) · 769 Bytes

README.md

File metadata and controls

6 lines (4 loc) · 769 Bytes

FeedForward Neural Network

Implementation of a FeedForward Neural Network (FFNN) using Matlab. The network is trained using Gradient Descent combined with the BackPropagation technique, and its performances can be evaluated on two datasets, namely the Monk dataset (https://archive.ics.uci.edu/ml/datasets/MONK's+Problems) and a dataset of sensor readings.

The file neural_network.m carries out the whole training process (also plotting the learning curves and saving them to a file) given a set of hyperparameters. If you want to perform a grid search to do hyperparametrs tuning, run main.m instead.

The Datasets directory contains a couple of dataset that can be used to train the network and evaluate its performances.