Keras

Keras
Original author(s)François Chollet
Developer(s)ONEIROS
Initial release27 March 2015; 9 years ago (2015-03-27)
Stable release
3.7.0[1] / 26 November 2024; 32 days ago (26 November 2024)
Repository
Written inPython
PlatformCross-platform
TypeFrontend for TensorFlow, JAX or PyTorch (and more)
LicenseApache 2.0
Websitekeras.io Edit this on Wikidata

Keras is an open-source library that provides a Python interface for artificial neural networks. Keras was first independent software, then integrated into the TensorFlow library, and later supporting more. "Keras 3 is a full rewrite of Keras [and can be used] as a low-level cross-framework language to develop custom components such as layers, models, or metrics that can be used in native workflows in JAX, TensorFlow, or PyTorch — with one codebase."[2] Keras 3 will be the default Keras version for TensorFlow 2.16 onwards, but Keras 2 can still be used.[3]

History

The name 'Keras' derives from the Ancient Greek word κέρας (Keras) meaning 'horn'.[4]

Designed to enable fast experimentation with deep neural networks, Keras focuses on being user-friendly, modular, and extensible. It was developed as part of the research effort of project ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System),[5] and its primary author and maintainer is François Chollet, a Google engineer. Chollet is also the author of the Xception deep neural network model.[6]

Up until version 2.3, Keras supported multiple backends, including TensorFlow, Microsoft Cognitive Toolkit, Theano, and PlaidML.[7][8][9]

As of version 2.4, only TensorFlow was supported. Starting with version 3.0 (as well as its preview version, Keras Core), however, Keras has become multi-backend again, supporting TensorFlow, JAX, and PyTorch.[10]

Features

Keras contains numerous implementations of commonly used neural-network building blocks such as layers, objectives, activation functions, optimizers, and a host of tools for working with image and text data to simplify programming in deep neural network area.[11] The code is hosted on GitHub, and community support forums include the GitHub issues page, and a Slack channel.[citation needed]

In addition to standard neural networks, Keras has support for convolutional and recurrent neural networks. It supports other common utility layers like dropout, batch normalization, and pooling.[12]

Keras allows users to produce deep models on smartphones (iOS and Android), on the web, or on the Java Virtual Machine.[8] It also allows use of distributed training of deep-learning models on clusters of graphics processing units (GPU) and tensor processing units (TPU).[13]

See also

References

  1. ^ "Release 3.7.0". 26 November 2024. Retrieved 25 December 2024.
  2. ^ "Keras: Deep Learning for humans". keras.io. Retrieved 2024-04-30.
  3. ^ "What's new in TensorFlow 2.16". Retrieved 2024-04-30.
  4. ^ Team, Keras. "Keras documentation: About Keras 3". keras.io. Retrieved 2024-02-10.
  5. ^ "Keras Documentation". keras.io. Retrieved 2016-09-18.
  6. ^ Chollet, François (2016). "Xception: Deep Learning with Depthwise Separable Convolutions". arXiv:1610.02357 [cs.CV].
  7. ^ "Keras backends". keras.io. Retrieved 2018-02-23.
  8. ^ a b "Why use Keras?". keras.io. Retrieved 2020-03-22.
  9. ^ "R interface to Keras". keras.rstudio.com. Retrieved 2020-03-22.
  10. ^ Chollet, François; Usui, Lauren (2023). "Introducing Keras Core: Keras for TensorFlow, JAX, and PyTorch". Keras.io. Retrieved 2023-07-11.
  11. ^ Ciaramella, Alberto; Ciaramella, Marco (2024). Introduction to Artificial Intelligence: from data analysis to generative AI. ISBN 9788894787603.
  12. ^ "Core - Keras Documentation". keras.io. Retrieved 2018-11-14.
  13. ^ "Using TPUs | TensorFlow". TensorFlow. Archived from the original on 2019-06-04. Retrieved 2018-11-14.