TensorFlow Fundamentals: What are Tensors in TensorFlow.js?
TensorFlow is an open-source machine learning library developed by Google. It is widely used for building and training machine learning and deep learning models. In TensorFlow, data is represented as tensors, which are multi-dimensional arrays.
TensorFlow.js is the JavaScript version of TensorFlow that allows developers to build and train machine learning models directly in the browser. Tensors play a crucial role in TensorFlow.js, as they are used to store and manipulate the data that is fed into the models.
What are Tensors?
In TensorFlow.js, a tensor is a multi-dimensional array that can store numeric data. Tensors can have any number of dimensions, known as rank. For example, a scalar (single number) is a 0-dimensional tensor, a vector is a 1-dimensional tensor, a matrix is a 2-dimensional tensor, and so on.
Here is an example of creating a tensor in TensorFlow.js:
const tensor = tf.tensor([1, 2, 3, 4]);
This creates a 1-dimensional tensor with the values [1, 2, 3, 4]. Tensors can also be created from existing arrays, typed arrays, or other tensors.
Operations with Tensors
Once tensors are created, you can perform various operations on them, such as addition, subtraction, multiplication, division, and more. These operations are carried out element-wise, meaning that the corresponding elements of the tensors are combined using the specified operation.
Here is an example of adding two tensors together:
const tensor1 = tf.tensor([1, 2, 3, 4]); const tensor2 = tf.tensor([5, 6, 7, 8]); const result = tensor1.add(tensor2);
This will create a new tensor with the values [6, 8, 10, 12], which are the sum of the corresponding elements of tensor1 and tensor2.
Conclusion
Tensors are a fundamental concept in TensorFlow.js and are used to represent and manipulate data in machine learning models. Understanding how to create and work with tensors is essential for building and training models effectively. By leveraging the power of tensors, developers can create advanced machine learning applications directly in the browser using TensorFlow.js.
Catch more episodes from Machine Learning for Web Developers (Web ML) → https://goo.gle/learn-WebML
@GoogleDevelopers when we say "3d array" are we really means "an array which can be a 3dimensional vector but got converted to a 1d tensor"? I am having trouble understand the relationship between the 3d describing an array vs the 3d describing a vector, are they all the same?
Brilliant, lucid explanation of tensors. Thanks so much!
I didn't understand how you are drawing grayscale image only with white and black color? because grayscale image must have shades of gray, and that could only be achieve through RGB values(3D Array). Are you referring here vector images in place of grayscale which are only combination of black and white pixels? that would be appropriate example to achieve with 2D array with [x, y, black/white]. Am I getting it correct here or am I wrong?
Genius & crystal clear. Thanks Sir
This series of videos has greatly raised my understanding of Tensor, thank you!
Learning lot from you Sir.Thank.