Issue: Create a quantum machine learning tutorial in a Jupyter notebook using the Python package HierarQcal for classifying music genres.
Overview
This tutorial will guide users on using HierarQcal to create quantum circuit models for classification tasks. The goal is to provide a quantum machine learning template, where a user can obtain baseline model performance for a classification task. While focusing on the GTZAN dataset's country vs. rock genre pair, the tutorial should emphasize a robust pipeline for both classical and quantum aspects of QML.
Resources:
Paper: The methods section in the hierarqcal paper serves as a blueprint for constructing a machine learning pipeline. The Architectural Impact section will also help guide model development and provide something to compare against.
qml_example.py: A code snippet is provided as an MVP for this tutorial, and should be expanded into a comprehensive, well-explained template.
Data: Use the GTZAN dataset to distinguish between country and rock songs as done in the paper. Use the 30 second window data with PCA(8 components) encoded on 8 qubits with AngleEmbedding. If alternative preprocessing yields better results, feel free to implement it.
Key Notes
Ensure the tutorial is adaptable to other classification tasks despite the specific example used.
Suggested sections (feel free to modify as needed):
Data Loading
Data Cleaning and Preprocessing
Model Creation (with HierarQcal)
Training Pipeline
Evaluation and Visualization
Implement cross-validation for a comprehensive model evaluation.
Feel free to re-use images/graphs from the paper to include in the tutorial.
Requirements
Code Base: Utilize the provided code, and expand or improve upon it as needed. The code includes essential parts for getting a model that works.
Explanation: Provide detailed explanations for each code section to clarify the quantum circuit architecture and machine learning pipeline.
Visualization: Incorporate plots that illustrate model architecture, training loss curves, and overall performance.
References
Lourens et al. (2023). Hierarchical Quantum Circuit Representations for Neural Architecture Search.
@matt-lourens Created a PR on QML for music genre classification. I will need your help on the different QCNN architectures used in the paper to extend the tutorial and benchmark them.
Issue: Create a quantum machine learning tutorial in a Jupyter notebook using the Python package HierarQcal for classifying music genres.
Overview
This tutorial will guide users on using HierarQcal to create quantum circuit models for classification tasks. The goal is to provide a quantum machine learning template, where a user can obtain baseline model performance for a classification task. While focusing on the GTZAN dataset's country vs. rock genre pair, the tutorial should emphasize a robust pipeline for both classical and quantum aspects of QML.
Resources:
Paper: The methods section in the hierarqcal paper serves as a blueprint for constructing a machine learning pipeline. The Architectural Impact section will also help guide model development and provide something to compare against.
qml_example.py: A code snippet is provided as an MVP for this tutorial, and should be expanded into a comprehensive, well-explained template.
Data: Use the GTZAN dataset to distinguish between country and rock songs as done in the paper. Use the 30 second window data with PCA(8 components) encoded on 8 qubits with AngleEmbedding. If alternative preprocessing yields better results, feel free to implement it.
Key Notes
Requirements
References
qml_example.py