SR Quantum Machine Learning

CHANGELOG.md

[0.1.12] - 06-05-2026

Added

  • Implemented a first-class quantum autoencoder workflow in qml.autoencoder
  • Added autoencoder CLI support via python -m qml autoencoder
  • Added smoke, artifact, CLI, and import coverage for the quantum autoencoder
  • Added autoencoder documentation and example notebook support
  • Added a GitHub Pages workflow for publishing a custom static documentation site
  • Added a Pages site generator and stylesheet matched to the root portfolio site
  • Added generated RESULTS.md reference outputs and a Results web page
  • Added generated result images to RESULTS.md and the Results web page
  • Added a Pages workflow status badge to README.md
  • Added documentation for the Pages tooling in docs/pages/README.md
  • Added README and usage-documentation links to the published Pages site

Maintenance

  • Ignored local static-site build output directories (_site/ and site_docs/)
  • Configured pytest to include the repository root on the import path for local test runs
  • Regenerate reference results during the Pages workflow before building the site

Summary

New core QML capability:

  • variational quantum classification (VQC)
  • variational quantum regression (VQR)
  • quantum convolutional neural networks (QCNN)
  • quantum autoencoders
  • quantum kernel methods
  • trainable quantum kernels
  • quantum metric learning

[0.1.11] - 10-04-2026

Added

  • Implemented a first-class QCNN workflow in qml.qcnn
  • Added qcnn CLI support via python -m qml qcnn
  • Added QCNN benchmark support in classification benchmarks
  • Added QCNN smoke, CLI, benchmark, and import coverage
  • Added QCNN documentation across README, usage docs, theory notes, and a dedicated algorithm page
  • Added QCNN example notebook: notebooks/quantum_convolutional_neural_network.ipynb

Summary

New core QML capability:

  • variational quantum classification (VQC)
  • variational quantum regression (VQR)
  • quantum convolutional neural networks (QCNN)
  • quantum kernel methods
  • trainable quantum kernels
  • quantum metric learning

[0.1.10] - 10-04-2026

Fixed

  • metric-learning now honors --save in the CLI and API
  • Added JSON/plot artifact saving for quantum metric learning
  • Normalised VQR artifact output paths to results/vqr/ and images/vqr/
  • Updated API docs to reflect that metric learning returns a dataclass result

Added

  • Regression tests for metric-learning artifact saving
  • Regression tests for VQR default artifact path selection

Maintenance

  • Ignored local .codex file and .codex/ directory
  • Removed stale local build artifacts before the next release cut

[0.1.9] - 06-04-2026

Added

Quantum metric learning

  • Implemented supervised quantum metric learning using contrastive loss
  • Trainable data re-uploading embedding circuits
  • Nearest-centroid classification in learned quantum feature space
  • CLI workflow:
python -m qml metric-learning --samples 200 --layers 2 --steps 50 --plot
  • Notebook:
notebooks/quantum_metric_learning.ipynb
  • Documentation:
docs/qml/metric_learning.md

Visualisation support

  • Added plot_metric_learning_embeddings(...) to qml.visualize
  • Standardised plotting via shared visualisation utilities
  • Automatic embedding plots when plot=True

Benchmark integration

  • Added quantum_metric_learning to classification benchmark framework
  • Supports multi-seed comparison with VQC, quantum kernel, and classical baselines
  • Compatible with per-model hyperparameter overrides

CLI integration

  • Added metric-learning subcommand
  • Consistent interface with other QML workflows

Testing

  • Added smoke tests for:

  • API workflow

  • CLI execution
  • Ensures reproducibility with small-step configurations

Documentation updates

  • README feature list updated
  • USAGE.md includes API and CLI usage examples
  • THEORY.md extended with contrastive learning formulation
  • Added dedicated docs page:
docs/qml/metric_learning.md

Internal improvements

  • Unified plotting interface across models
  • Improved result dataclass structure for embedding-based workflows
  • Added label outputs (y_train, y_test) to metric learning results
  • Improved compatibility of benchmark framework with dataclass-based outputs

Summary

New core QML capability:

  • variational quantum classification (VQC)
  • variational quantum regression (VQR)
  • quantum kernel methods
  • trainable quantum kernels
  • quantum metric learning

Metric learning provides a flexible representation-learning approach compatible with classical classifiers and similarity-based workflows.

[0.1.7] - 06-04-2026

Added

  • unified training loop via qml.training.run_training_loop
  • shared utilities in qml.utils
  • centralised path handling via qml.io_utils.ensure_dir

Refactored

  • removed duplicated optimisation loops across VQC, VQR, and kernel workflows
  • improved package modularity and internal consistency
  • simplified experiment output handling

Removed

  • deprecated qml.datasets
  • redundant local helper functions

[0.1.5] - 06-04-2026

Added

  • Multiple dataset support via qml.data

  • classification datasets:

    • moons
    • circles
    • blobs
    • xor

    • regression datasets:

    • linear

    • sine
    • polynomial
    • Dataset selection exposed across public APIs:

  • run_vqc(dataset=...)

  • run_vqr(dataset=...)
  • run_quantum_kernel_classifier(dataset=...)
  • run_trainable_quantum_kernel_classifier(dataset=...)
  • compare_classification_models(dataset=...)
  • compare_regression_models(dataset=...)
  • CLI support for dataset selection:

bash python -m qml vqc --dataset circles python -m qml regression --dataset sine python -m qml benchmark classification --dataset xor - Dataset smoke tests ensuring end-to-end compatibility - Deterministic dataset generation with seeded NumPy RNG

Changed

  • Benchmark framework updated to support model-specific kwargs alongside dataset selection
  • Classification and regression runners now return consistent "dataset" metadata
  • Improved separation between dataset specification and data tensors
  • Benchmark dispatch filters unsupported kwargs for classical baselines

Fixed

  • Finite-shot determinism preserved across datasets
  • Regression benchmark default dataset corrected to "linear"
  • Removed dataset shadowing bug where dataset dict replaced dataset name
  • CLI dataset argument now correctly propagates to runners

0.1.4 - 06-04-2026

Added

  • Noise-aware benchmark support via per-model model_kwargs
  • Finite-shot benchmark smoke tests
  • Deterministic finite-shot benchmark execution with fixed seeds
  • Extended dataset utilities for multiple classification and regression dataset types

Changed

  • Updated benchmark dispatch to support model-specific kwargs cleanly
  • Refined README, USAGE, and THEORY documentation to reflect current package capabilities
  • Generalised qml.data dataset generation with lightweight dispatch helpers

0.1.2

Added

  • benchmark CLI workflow
  • multi-seed comparison utilities
  • benchmark smoke tests
  • documentation for benchmarking workflows

Improved

  • consistency of classical vs quantum comparisons