Skip to content

Python source code, property of Federica Cavaliere and UNISA's Natural Computing Laboratory, necessary for the master thesis work “Parkinson's Disease Diagnosis: Towards Grammar-based Explainable Artificial Intelligence”, on March 3 2020.

Notifications You must be signed in to change notification settings

Natural-Computation-Lab/Tesi_FedericaCavaliere

 
 

Repository files navigation

MasterThesis

This repository contains 4 project folders:

  1. "Classifier_Comparison" for the benchmarking work in which the performances (as the hyperparameters vary) of 6 machine learning techniques (SVM, Naive Bayes, MLP, AdaBoost, Random Forest, Decision Tree) were compared on the data sets meanders and spirals HandPD;
  2. "Classifier_Comparison_plots" for the graphs of point 1);
  3. "PonyGE2-master" for the performance analysis of the Grammatical Evolution technique with the homonymous Python library (PonyGE2) on the HandPD dataset, designed two grammars (basic grammar, advanced grammar);
  4. "GE_plots" for the graphs of point 3).

For project 3),

  • the executable source file is recoverable for testing at the following path: PonyGE2-master / src / big_run.py

For a correct execution,

  • set as execution parameters the name of the evolution classification file chosen (--parameters namefile), which can be viewed in: PonyGE2-master / parameters /, a classification file for the meadri (classification_meander.txt) and one for the spirals (classification_spiral.txt)
  • the grammar used is specified in the classification file, which can be viewed in: PonyGE2-master / parameters / grammars, a file for basic grammar (test_basic.pybnf) and one for advanced grammar (test_adv.pybnf)
  • the function for calculating and evaluating the fitness of each individual can be viewed at: PonyGE2-master / src / fitness / supervised_learning / adf_supervised_learning.py

About

Python source code, property of Federica Cavaliere and UNISA's Natural Computing Laboratory, necessary for the master thesis work “Parkinson's Disease Diagnosis: Towards Grammar-based Explainable Artificial Intelligence”, on March 3 2020.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.8%
  • Shell 0.2%