Written in Python, Keras is an artificial neural network rapid prototyping interface. It is based on the deep learning libraries Tensorflow, Microsoft Cognitive Toolkit, PlaidML and Theano.
Keras, what is it?
Written in Python, Keras is an open source library for rapid prototyping deep learning models. Accessible to beginners in AI, it revolves around a high-level API supporting different libraries of recurrent or convolutional artificial neural networks, such as Tensorflow, Microsoft Cognitive Toolkit, PlaidML or Theano.
The goal of Keras is to provide a framework for developing artificial neural networks as quickly as possible. Initiated in 2015, this technology is based on the work of François Chollet, a Google developer. It is part of the Oneiros project (for Open-ended Neuro-Electronic Intelligent Robot Operating System).
Why using Keras?
Keras is designed to make deep learning projects accessible. Its API is intended to be relatively simple to learn. It is designed to minimize the number of actions to be performed when creating an artificial neural network, while delivering explanations in the event of errors. Another advantage of Keras, its machine learning models are supported by a wide range of platforms: iOS, Android, Raspberry Pi or Google Cloud.
Under the MIT open source license, Keras contains many neural network bricks: layers, activation functions, optimizers… The library also includes a multitude of tools to facilitate work with image and text data. Keras is characterized by excellent multi-GPU support in order to distribute the training of a model on several graphics cards and thus ingest a massive volume of training data. Keras code is hosted on GitHub.
Is Keras cross-platform?
Keras models are deployed on:
- iOS, through Apple’s CoreML,
- Android, through the TensorFlow Android runtime,
- RaspberryPi,
- A browser via JavaScript GPU accelerators,
- A Java virtual machine.
How does Keras work?
In Keras, a model comes in the form of an arrangement of configurable modules: layers of neurons, optimizers, activation functions, regularization… New models can be further developed, and added as needed.
Keras delegates low-level operations to the deep learning frameworks on which it relies: TensorFlow, Theano, Microsoft Cognitive Toolkit…
How to download Keras?
Keras can be downloaded from GitHub. The open-source project site offers extensive documentation on the Keras API, as well as a series of guides covering various use cases.
What is the difference between Keras and TensorFlow?
Developed by Google, TensorFlow is an open-source deep learning platform designed to design artificial neural networks. As for Keras, it is a high-level API tailored for the rapid prototyping of artificial neural networks that can use TensorFlow as an implementation base.
ABOUT LONDON DATA CONSULTING (LDC)
We, at London Data Consulting (LDC), provide all sorts of Data Solutions. This includes Data Science (AI/ML/NLP), Data Engineer, Data Architecture, Data Analysis, CRM & Leads Generation, Business Intelligence and Cloud solutions (AWS/GCP/Azure).
For more information about our range of services, please visit: https://london-data-consulting.com/services
Interested in working for London Data Consulting, please visit our careers page on https://london-data-consulting.com/careers