Mushroom Neural Network

Quickstart

Interactive Neural Network Training

1. Navigate to neural_network/
2. Ensure the presentation_mode parameter in run_interactive.m is false
3. Run run_interactive.m in Octave or MATLAB
4. Using the interactive console, one can choose to load a prexisting configuration or train a new one

Presentation Mode

1. Navigate to neural_network/
2. Ensure the presentation_mode parameter in run_interactive.m is true
3. Run run_interactive.m in Octave or MATLAB

Training Multiple Neural Networks

1. Navigate to neural_network/
2. Run run_interactive.m in Octave or MATLAB
3. Once training is complete, trained networks will show up in neural_network/trained_networks

Overview

datasets/

• Contains the datasets used to train and validate the neural network

docs/

• Planning documents used in the process of creating the neural network

prototypes/

• Initial prototypes of the neural network

neural_network/

train.m

• Function to train neural network

autorun.m

• Trains multiple neural networks and saves them uniquely by training parameters and timestamp

predict.m

• Uses input and expected output to check neural network's ability to predict

find_error.m

• Compares the neural network to the validation set correction-set.csv and computes its accuracy

neural_network/trained_networks

• Directory of all trained networks
• Can be loaded via run_interactive to explore stats and accuracy