Skip to content

CHANGELOG.md

Latest commit

 

History

History
220 lines (181 loc) · 9.43 KB

CHANGELOG.md

File metadata and controls

220 lines (181 loc) · 9.43 KB
Raw

Changelog

All notable changes to QBioCode will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • Quantum Backend: Noisy simulation support based on IBM device noise models

    • New backend format: noisy_<device_name> (e.g., noisy_ibm_cleveland)
    • Extracts noise model from actual IBM Quantum devices for realistic local simulation
    • Configurable simulation method via sim_method parameter
    • Supports all AerSimulator methods: statevector, matrix_product_state, tensor_network, etc.
    • Enables testing quantum algorithms under realistic noise without hardware queue times
    • No IBM Quantum compute credits consumed (simulation runs locally)
    • Updated qbiocode.utils.qutils.get_backend_session() to support noisy backends
    • Updated qbiocode.utils.ibm_account.py for improved credential handling
    • Documentation updated with comprehensive examples and best practices

  • Data Generation: New blob dataset generator

    • generate_blobs_datasets(): Create isotropic Gaussian blob datasets
    • generate_default_blobs_datasets(): Quick generation with default parameters
    • Follows QBioCode data generation patterns
    • Useful for clustering and classification benchmarks

  • Evaluation Metrics: Added generalized evaluation_metrics() function to qbiocode.evaluation.model_evaluation

    • Supports multiple metrics: accuracy, brier, f1, precision, recall, auc
    • Configurable via metrics parameter (default: ['accuracy', 'brier'])
    • Backward compatible: returns (accuracy, brier) tuple by default
    • Supports both binary and multi-class classification
    • Provides calibration quality assessment via Brier score
    • Handles edge cases (e.g., single class in test set)
    • Now available in main API (previously only in tutorial helpers)

  • Quantum Ensemble Learning: New unified quantum ensemble classifier

    • compute_qensemble(): Quantum ensemble with configurable construction methods
      • Implements quantum ensemble using controlled operations and superposition
      • Two ensemble methods via ensemble_method parameter:
        • "swap" (default): Fixed controlled-SWAP operations (faster, deterministic)
        • "random_unitary": Haar-random unitaries (more general, potentially better generalization)
      • Three ensemble modes: balanced, unbalanced, and pair_sample
      • Support for configurable ensemble depth (d) and operations per qubit (n_swap)
      • Quantum cosine similarity classifier using SWAP test
    • New utility functions in qbiocode.utils:
      • normalize_data(): Normalize data for quantum state encoding
      • label_to_array(): Convert binary labels to one-hot encoding
      • prepare_training_set(): Prepare balanced training subsets
      • retrieve_probabilities(): Extract probabilities from measurement counts (generic quantum utility)
      • execute_circuit(): General-purpose Aer simulator execution (reusable across quantum algorithms)
    • Based on Macaluso et al., "A variational algorithm for quantum ensemble learning" (2023)
    • Integrated from tutorial/QEnsemble with full API compatibility
    • Code organization: Extracted reusable functions to utils for broader applicability

  • Testing Infrastructure: Comprehensive test suite for core functionality

    • tests/test_data_generation.py: Tests for data generation utilities
    • tests/test_file_utilities.py: Tests for file operations and utilities
    • tests/test_generator_dispatch.py: Tests for generator dispatch logic
    • tests/conftest.py: Pytest configuration and fixtures
    • Test coverage for utility modules and data generation helpers

  • Code Quality Tools: Enhanced development tooling

    • isort integration for consistent import ordering
    • Configuration in pyproject.toml for isort settings
    • Added to dev and all dependency groups

  • Documentation: Testing instructions in README

    • Added "Running Tests" section with pytest usage
    • Instructions for installing development dependencies

Changed

  • Code Formatting: Applied consistent code style across entire codebase

    • Ran black formatter on all Python files
    • Ran isort for standardized import ordering
    • Fixed invalid escape sequences in visualization module
    • Improved code readability and maintainability

  • CI/CD Improvements: Stabilized continuous integration pipeline

    • Updated GitHub Actions workflows to Node.js 24
    • Fixed CI code quality checks for import ordering
    • Fixed CI type-check issues
    • Fixed documentation build process
    • Fixed Pandoc compatibility issues
    • Enhanced workflow reliability across all platforms

  • Testing: Improved test reliability

    • Fixed path-order assumptions in duplicate-file tests
    • Tests now work consistently across different file systems

Fixed

  • Invalid escape sequence in qbiocode/visualization/visualize_correlation.py
  • Import ordering issues throughout codebase
  • Type-check errors in CI pipeline
  • Documentation build failures
  • Path handling in cross-platform tests

Planned Features

  • Additional quantum ML algorithms
  • Enhanced meta-learning capabilities
  • More dataset complexity metrics
  • Performance optimizations
  • Extended Galaxy tool integration

0.1.0 - 2026-04-06

⚠️ Breaking Changes

  • Minimum Python version increased to 3.10 (was 3.9)
    • Required for compatibility with latest Qiskit ecosystem (qiskit-ibm-runtime 0.44.0+)
    • Python 3.9 reaches end-of-life in October 2025

Added

Core Features

  • QProfiler: Automated machine learning benchmarking with data complexity analysis

    • Support for multiple ML models (RF, SVM, LR, DT, NB, MLP, XGBoost)
    • Support for quantum ML models (QSVC, PQK, VQC, QNN)
    • Comprehensive data profiling and complexity metrics
    • Batch processing mode for large-scale experiments
    • CLI interface for easy usage

  • QSage: Meta-learning framework for intelligent model selection

    • Model recommendation based on dataset characteristics
    • Pre-trained models for quick predictions
    • Integration with QProfiler results

  • Data Generation: Artificial dataset generation with controlled complexity

    • Multiple dataset types (circles, moons, spirals, S-curve, Swiss roll, spheres)
    • Configurable noise levels and complexity parameters
    • Support for multi-class classification problems

  • Embeddings: Dimensionality reduction and feature extraction

    • Autoencoder implementations
    • Integration with classical embedding methods

  • Evaluation: Comprehensive model and dataset evaluation

    • Multiple metrics (accuracy, F1-score, AUC, training time)
    • Cross-validation support
    • Statistical analysis tools

  • Visualization: Publication-quality plotting functions

    • Correlation analysis plots with scientific styling
    • Heatmaps with customizable colormaps
    • High-resolution output (600 DPI) for publications

Documentation

  • Complete API documentation with Sphinx
  • Tutorials for QProfiler, QSage, and data generation
  • Installation guides for multiple platforms
  • Galaxy integration documentation
  • Example notebooks and workflows

Project Infrastructure

  • GitHub issue templates (bug report, feature request, documentation, question)
  • Pull request template with comprehensive checklists
  • Security policy (SECURITY.md)
  • Support documentation (SUPPORT.md)
  • Contributing guidelines (CONTRIBUTING.md)
  • Code of conduct (CODE_OF_CONDUCT.md)
  • Citation file (CITATION.cff)
  • Zenodo metadata for DOI generation (.zenodo.json)

CI/CD

  • GitHub Actions workflow for continuous integration
    • Multi-OS testing (Ubuntu, macOS, Windows)
    • Multi-Python version testing (3.10, 3.11)
    • Code quality checks (flake8, black, isort, mypy)
    • Test coverage reporting
    • Documentation building
  • GitHub Actions workflow for automated releases
    • PyPI publishing
    • Zenodo archiving
    • Release asset management

Command-Line Tools

  • qprofiler: Run QProfiler experiments
  • qprofiler-batch: Batch processing mode
  • qsage: Model recommendation tool

Changed

  • Updated visualization functions for publication quality
    • Enhanced scatter plots with better colormaps
    • Improved heatmaps with professional styling
    • Better legend positioning and labeling
    • Increased DPI for high-quality output

Fixed

  • NaN handling in correlation visualization
  • Windows path compatibility issues
  • Automated QSage model download

Dependencies

  • Python >= 3.10, < 3.13
  • Qiskit for quantum computing functionality
  • scikit-learn for classical ML algorithms
  • pandas, numpy for data manipulation
  • matplotlib, seaborn for visualization
  • XGBoost for gradient boosting
  • hydra-core for configuration management

Release Notes

Version 0.1.0 - Initial Public Release

This is the first public release of QBioCode, a comprehensive framework for quantum machine learning applications in healthcare and life sciences. The release includes:

  • Complete implementation of QProfiler and QSage applications
  • Support for both quantum and classical machine learning models
  • Extensive documentation and tutorials
  • Professional project infrastructure for open-source collaboration
  • Automated CI/CD pipelines
  • Ready for PyPI distribution and Zenodo archiving

Note: This is an alpha release. APIs may change in future versions. Please report any issues on our GitHub issue tracker.