This repository contains Jupyter Notebooks and python modules implementing some classifiers for music genres trained on the GTZAN dataset.
It was developed as an assignment for the "Numerical Analysis for Machine Learning" course in the Artificial Intelligence track of the Computer Science and Engineering Masters Degree at Politecnico di Milano. The course was taught by Edie Miglio and Francesco Regazzoni. Group Details: Group ID: 14 Academic Year: 2021/2022
WARNING: This project was meant to evaluate the practical implementation aspect of the requested algorithms. For this reason, no crossvalidation, model selection or other performance enhancements were applied to the classifiers.
Implemented Classifiers
- k-Nearest Neighbours
- Nearest Centroid
- Multiclass Support Vector Machine
- One-To-One (with decision tree)
- One-To-Rest (ties resolved with highest "score"
All details on the implementation, the notebooks, the module documentation are available in the report.pdf file.