Keras
lastmod 2025-02-11created 2018-10-253 edits
Keras is an open-source neural network library written in Python. It can run on top of MXNet, Deeplearning4j, TensorFlow, Microsoft Cognitive Toolkit, or Theano. It is designed to enable fast experimentation with deep neural networks and focuses on minimal, modular extensibility. It was developed as part of the research effort of the ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System) project, and its primary author and maintainer is Google engineer Francois Chollet.
Key features of Keras
Keras has the following four key features.
- Modularity
- The modules Keras provides are independent and configurable, and can be connected to each other with as few restrictions as possible. A model is a composition of these modules as a sequence or a graph.
- In particular, neural network layers, cost functions, optimizers, initialization schemes, activation functions, and regularization schemes are all independent modules, and you can combine them to build new models.
- Minimalism
- Each module is short and simple.
- All code should be understandable at a single glance.
- However, iteration speed and innovation may suffer somewhat.
- Easy extensibility
- Modules are very easy to add as new classes or functions.
- This allows the wide range of expression needed for advanced research.
- Python-based
- Unlike Caffe, no separate model configuration file is needed; models are defined in Python code.
The person who develops and maintains this great Keras is Google engineer François Chollet.