This article will help to understand the basics of deep learning, TensorFlow and mostly comparison between some top deep learning framework.
Deep Learning: Deep learning is a subset of machine learning which comes under the Artificial Intelligent, which aims to learn features to input data. Nowadays, researchers have deeply investigated deep learning algorithms for solving challenging problems in many areas such as image classification, Object recognition & Detection, and natural language processing. Basically, Deep learning has become popular since 2006.
- Deep Learning algorithm inspired by the structure and function of the brain called artificial neural network.
- There are many more problems that don’t solve by Machine learning like Huge Data, Multidimensional Array and others.
- This problem is solved by the Deep learning concept.
- Deep learning for image analysis using an algorithm called CNN (Convolution neural network).
Why Deep learning important?
- When we required the machine will work like the human brain and perform a task like the human.
- Deep Learning is required for the more powerful supervised learning (Like prediction based ).
- ML is not nearly as successful with unsupervised learning.
- Basically, deep learning comes after the ML to overcome the large complex problem of ML.
- And it is the next step of AI.
- DL has the ability to learn from unlabelled or unstructured data is as enormous benefits for those interested in the real-world application.
How Deep Learning Works?
Deep Learning framework is like libraries that provide support to the deep learning concept to implement and improve its performance.
What is Tensor flow:- Tensor flow is an open-source software library for high-performance numerical computation. Its flexible architecture allows easy deployment of computation across a variety of platforms (CPU’S GPU’S TPU’S) and from desktop to clusters of a server to mobile and edge devices.
Originally developed by researchers and engineers from the Google brain team within Google’s AI organization. It comes with the strong support of machine learning and deep learning.
- Open source
- Uses a platform such as a python, C/C++, Java.
- Written in C++/ Python.
- It Offers
pre-trained model Allow multi-nodeexecution
- The computation graph is pure python, therefore it is slow.
- No commercial support.
Comparison of some different DL framework which is top in industries nowadays.
|Framework||Open Source||Platform||Written –In||Interface||Pretrained Model||Creator|
|Caffe||YES||Linux,Mac Os,windows||C++||Python, MATLAB||YES||Berkeley Vision and Learning Center|
|Tensorflow||YES||Linux,Mac Os,windows||C++, Python||Python,C/C++ ,Java , R||YES||Google Brain Team|
|Theano||YES||Cross Platform||Python||Python(Keras)||NO||University de Montréal|
|Keras||YES||Linux,Mac Os,windows||Python||Python,R||YES||François Chollet|
|CNTK||YES||Cross Platform||C++, Python||C++,python, C#||YES||Micosoft|
This article is written after reading from the official site of the above DL frameworks.
Thank you :>)