Metadata-Version: 2.4
Name: hone-io
Version: 2.1.0
Summary: Library for quantum cognition machine learning.
Requires-Python: >=3.12
Description-Content-Type: text/markdown
Requires-Dist: torch==2.7.0
Requires-Dist: structlog>=24.4.0
Requires-Dist: pandas>=2.2.3
Requires-Dist: qbism>=0.1.0
Requires-Dist: numpy>=2.1.3
Requires-Dist: rich
Requires-Dist: scikit-learn
Requires-Dist: sentry_sdk
Requires-Dist: tensordict>=0.7.2
Requires-Dist: cryptlex-lexactivator>=3.33.0
Requires-Dist: cachetools>=6.1.0
Requires-Dist: pywavelets>=1.9.0
Requires-Dist: tqdm>=4.67.1

# Hone-io Community Edition - QCML Scikit-Learn Integration

Welcome to the **Honeio Community Edition**! This package provides scikit-learn compatible wrappers for Quantum Cognition Machine Learning (QCML) models. Documentation available [here](https://qognitive-qcog-core.readthedocs-hosted.com/en/latest/).

## 🚀 What is QCML?

**Quantum Cognition Machine Learning** is a new form of machine learning that is inspired by quantum cognition.
QCML models learn a representation of the input data into quantum states, and the outputs of the models reflect the outcomes of quantum measurements.
QCML is highly effective on datasets with a large number of input features and a large number of classes (for classification) or targets (for regression).

For more details you can check out some of our publications:
- [Quantum Cognition Machine Learning AI Needs Quantum](https://www.qognitive.io/papers/QCML%20-%20Qognitive,%20Inc.pdf)
- [Robust estimation of the intrinsic dimension of data sets with quantum cognition machine learning](https://www.nature.com/articles/s41598-025-91676-8)
- [Quantum Cognition Machine Learning: Financial Forecasting](https://www.qognitive.io/papers/Qognitive_Financial_Forecasting.pdf)
- [Supervised Similarity for High-Yield Corporate Bonds with Quantum Cognition Machine Learning](https://arxiv.org/abs/2502.01495)
- [Quantum Cognition Machine Learning for Forecasting Chromosomal Instability](https://arxiv.org/abs/2506.03199)
- [Quantum Geometry of Data](https://arxiv.org/abs/2507.21135)
- [QCML Patient Similarity - UTI Prediction](https://www.qognitive.io/blog/urinalysis/)
- [Breaking the curse of dimensionality: QCML vs. tree-based models](https://www.qognitive.io/blog/tree-benchmarks/)

## 📦 Available Classes

### `QCMLRegressor`
A scikit-learn compatible regressor for continuous target prediction tasks.

### `QCMLClassifier`
A scikit-learn compatible classifier for discrete classification tasks.

Both classes wrap the underlying QCML layers with scikit-learn wrapper and provide familiar scikit-learn interfaces.

## 🎯 Key Features

- **Scikit-learn compatibility**: Drop-in replacement for sklearn estimators
- **Quantum-inspired learning**: Represents data with quantum states
- **Adaptive weighting**: Learnable input feature weights for automatic feature selection
- **GPU support**: Train on CPU or CUDA devices
- **Model persistence**: Save and load trained models
- **Flexible batching**: Support for batch training or full-batch optimization

## ⚠️ Community Edition Limitations

The community edition has the following restrictions:

| Parameter | Limit | Description |
|-----------|-------|-------------|
| **Input Features** | 100 | Maximum number of input features/operators |
| **Output Features** | 12 | Maximum number of output features/operators |
| **Hilbert Space Dimension** | 8 | Maximum dimension of the quantum state space |
| **Training Samples** | 1,000 | Maximum number of training samples per batch |

> 💡 **Need more capacity?** Contact [support@qognitive.io](mailto:support@qognitive.io) for information about commercial licensing to remove these limitations.

## 🛠️ Installation

```bash
pip install hone-io
```

Documentation available [here](https://qognitive-qcog-core.readthedocs-hosted.com/en/latest/)
