# uncertainty-wizard

> Uncertainty-Wizard is a plugin on top of tensorflow.keras, allowing to easily and efficiently create uncertainty-aware deep neural networks. Also useful if you want to train multiple small models in parallel.

**Wikidata**: [Q127485425](https://www.wikidata.org/wiki/Q127485425)  
**Source**: https://4ort.xyz/entity/uncertainty-wizard

## Summary
Uncertainty-Wizard is a free software plugin designed for tensorflow.keras that enables the creation of uncertainty-aware deep neural networks. It allows users to easily and efficiently quantify model uncertainty and is also optimized for training multiple small models in parallel. The library is open-source, released under the MIT License, and is described in the academic context as a tool for "Fast and User-Friendly Neural Network Uncertainty Quantification."

## Key Facts
- **Entity Type:** Free Software / Software Plugin.
- **Primary Function:** Creates uncertainty-aware deep neural networks on top of tensorflow.keras.
- **Secondary Function:** Facilitates the parallel training of multiple small models.
- **License:** MIT License.
- **Latest Stable Version:** 0.4.0 (released February 16, 2023).
- **Initial Release:** Version 0.1.0 released on December 20, 2020.
- **Source Code Repository:** Hosted on GitHub at `testingautomated-usi/uncertainty-wizard`.
- **Documentation:** Available at `uncertainty-wizard.readthedocs.io`.
- **Academic Source:** Described by "Uncertainty-Wizard: Fast and User-Friendly Neural Network Uncertainty Quantification."

## FAQs
### Q: What framework does Uncertainty-Wizard integrate with?
A: Uncertainty-Wizard functions as a plugin built specifically on top of tensorflow.keras. It is designed to integrate seamlessly with this framework to extend its capabilities.

### Q: What are the main use cases for Uncertainty-Wizard?
A: The primary use case is creating deep neural networks that are aware of their own uncertainty (uncertainty quantification). Additionally, it is useful for efficiently training multiple small models in parallel.

### Q: Is Uncertainty-Wizard free to use?
A: Yes, it is classified as free software released under the MIT License, allowing users to freely run, study, change, and distribute it.

## Why It Matters
Uncertainty-Wizard addresses a critical gap in the deployment of deep neural networks: the ability to quantify confidence. Standard deep learning models often act as "black boxes," providing predictions without indicating how confident the model is in those results. By serving as a plugin for the widely used tensorflow.keras ecosystem, Uncertainty-Wizard democratizes access to uncertainty quantification, making it easier and more efficient for developers to build robust, safety-critical AI systems.

Furthermore, its utility extends to computational efficiency through the support of training multiple small models in parallel. This dual functionality—enhancing model reliability through uncertainty awareness and optimizing training processes—makes it a significant tool for researchers and engineers working with deep learning architectures.

## Notable For
- **Seamless Integration:** Functions as a direct plugin on top of tensorflow.keras, lowering the barrier to entry for uncertainty quantification.
- **User-Friendly Design:** Cited in academic sources as a "Fast and User-Friendly" solution for complex neural network tasks.
- **Parallel Training:** Offers distinct utility for training multiple small models concurrently, a feature not always standard in uncertainty libraries.
- **Open Source Accessibility:** Distributed under the permissive MIT License, encouraging broad adoption and modification.

## Body
### Overview and Functionality
Uncertainty-Wizard is a software library classified as free software. It is designed to operate as a plugin layer on top of `tensorflow.keras`. Its core objective is to simplify the process of generating deep neural networks that are capable of uncertainty quantification. Beyond uncertainty, the tool provides functionality to train multiple small models in parallel.

The tool is formally described by the source "Uncertainty-Wizard: Fast and User-Friendly Neural Network Uncertainty Quantification."

### Version History
The development of Uncertainty-Wizard has seen regular updates since its initial release in late 2020. The version progression highlights active maintenance and feature expansion.

*   **0.1.x Series:**
    *   **v0.1.0:** Released December 20, 2020 (Initial stable version).
    *   **v0.1.1:** Released February 9, 2021.
    *   **v0.1.2:** Released June 30, 2021.
    *   **v0.1.3:** Released July 22, 2021.
*   **0.2.x Series:**
    *   **v0.2.0:** Released December 16, 2021.
*   **0.3.x Series:**
    *   **v0.3.0:** Released May 3, 2022.
    *   **v0.3.1:** Released February 3, 2023.
    *   **v0.3.2:** Released February 3, 2023.
*   **Current Version:**
    *   **v0.4.0:** Released February 16, 2023. This is marked as the stable version and the preferred release as of the provided data.

### Technical Details and Location
*   **Source Code:** The code is maintained in a GitHub repository under the user `testingautomated-usi` with the project name `uncertainty-wizard`.
*   **Documentation:** Official documentation and user guides are hosted on Read the Docs at `uncertainty-wizard.readthedocs.io/en/latest/`.
*   **Licensing:** The software is fully open-source, governed by the MIT License.

## References

1. [Release 0.1.0. 2020](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.1.0)
2. [Release 0.1.1. 2021](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.1.1)
3. [Release 0.1.2. 2021](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.1.2)
4. [Release 0.1.3. 2021](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.1.3)
5. [Release 0.2.0. 2021](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.2.0)
6. [Release 0.3.0. 2022](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.3.0)
7. [Release 0.3.1. 2023](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.3.1)
8. [Release 0.3.2. 2023](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.3.2)
9. [Release 0.4.0. 2023](https://github.com/testingautomated-usi/uncertainty-wizard/releases/tag/v0.4.0)
10. [Source](https://api.github.com/repos/testingautomated-usi/uncertainty-wizard)