What is TensorFlow.js?
TensorFlow.js (tfjs) is a JavaScript library for creating, training, and deploying ML models in the browser or in Node.js. It was created at Google by Nikhil Thorat and Daniel Smilkov and was initially called Deeplearn.js, before being merged into the TensorFlow team in 2018 and renamed as TensorFlow.js.
TensorFlow.js provides two main layers, outlined as follows:
- CoreAPI: This is the low-level API that deals directly with tensors—the core data structure of TensorFlow.js.
- LayerAPI: A high-level layer built on top of the CoreAPI layer for easily building ML models.
In later sections, Tensors and basic operations on tensors and Building a simple regression model with TensorFlow.js, you will learn more details about the CoreAPI and LayerAPI layers.
With TensorFlow.js, you can do the following:
- Perform hardware-accelerated mathematical operations
- Develop ML models for the browser or Node.js
- Retrain existing ML models...
