Tensors can be thought to be generalized matrixes or, more specifically, mathematical entities living in structures and interacting with other mathematical entities. If the other entities in the structure are transformed in any way, then the tensor must also be transformed by that transformation rule.

What does the preceding definition mean? Perhaps thinking of a Tensor as multidimensional array is easier to grasp, or consider the following, comparing Scalar, Vector, Matrix, and Tensor:

Building on the topic of TensorFlow, TensorFlow is an open source software library (also called a framework) originally created by Google for creating deep learning models.

You can visit https://www.tensorflow.org/ for more details on TensorFlow.

In the next section, we will talk about the connection between deep learning, neural networks, and TensorFlow.