Authors: Alan An, Fares El Tin, Victoria Madge Kaggle Team Name: MAE
Repo to host code for Project 3 of COMP-551 Applied Machine Learning.
The main module for the artifical neural network implemented is called aNN.py
It is run by the driver named aNN_driver.py
The main module for the convolutional neural network implemented is kcNN/trainer/kerascNN_final.py
The script that generates prediction for cNN result is predict/prediction.py the trained model is uploaded and called final2_model.h5
The toyaNN.py and toyaNN_driver.py are testing modules. We tried a vector memorization problem with it to see whether the aNN implemented works or not.
The module to run the scikit-learn MLP is sci_NN.py
The main module to run logistic regression is Log_Reg.py
The main module for feature design is get_feat_fun.py. Please note that Mahotas library needs to be installed for Zernike.
project_3_stats.m is a MATLAB file which plotted the class distribution and the cross-validation results for hidden layers of NN.