Code example usage for "Symbolic Audio Classification via Modal Decision Tree Learning", ITADATA2024.
This repo contains Jupyter Notebooks showcasing the proposed workflow for obtaining symbolic rules for symbolic audio classification. The two notebooks are
- ItaData2024_p_code.ipynb for extracting propositional audio classification rules, via the DecisionTree.jl package.
- ItaData2024_m_code.ipynb for extracting modal audio classification rules, via the ModalDecisionTrees.jl package.
The code uses the PyCall.jl Julia package, and the librosa python library; ensure you first execute pycall_librosa_install.jl to correctly setup the environment for executing the Jupyter Notebooks.
The code for extracting the rules from the learned trees relies Sole.jl, an open-source framework for symbolic machine learning, originally designed for machine learning based on modal logics.
The package is mantained by the ACLAI Lab @ University of Ferrara.