Google Brain team developed this programming language which is very popular for designing deep neural networks and performing deep learning on the huge data sets. As explained in the previous tutorial "What is TensorFlow?" TensorFlow is open source programming language for machine learning and deep learning application development. TensorFlow programming langue uses computational graph which represents biological neural networks in machine language although its it not intelligent like bio-logical neural networks.
The TensorFlow graph consists of Tensors and the various mathematical operations on these Tensors. Mathematical operations such as multiply, add, divide etc.. are known as node.
In TensorFlow Tensors are simply the data in the network. It can be single dimensional or multi-dimentional array in the system on to which various mathematical operations are performed. Mathematical operations (node operation) takes one or tensors and may generate one or more tensors.
TensorFlow is called TensorFlow because it handles the flow (node/mathematical operation) of tensors (data). So, in TensorFlow we define the computational graph with Tensors and mathematical operation (node) to create system for machine learning and deep learning.
TensorFlow programming language provides tools to developer to create deep learning process by properly define flow of tensors. Developers can define machine learning routines within the flow.
This graph is executed by TensorFlow runtime over cluster of server consisting of CPU, GPU and TPU through TensorFlow session. TensorFlow programming language is designed to run the calculation with efficiency and in parallel over thousands of servers.
In future tutorials you will learn how to define TensorFlow graph and then run through TensorFlow session.
Posted on: October 1, 2017 If you enjoyed this post then why not add us on Google+? Add us to your Circles