Hi guys, today I will tell you about an amazing library launched a few months ago, TensorFlow JS aka tfjs

TensorFlowJS is a library for machine learning development with javaScript, the entire library was built following the most usual and clean methods to develop an entire machine learning application, you can treat data, process, learning, test, and production. This means that the process to use and apply machine learning and data science methods will become easier for developers.

ETL Process

With the TensorflowJS data function, we can load, transform, create, and access local data from webcam and microphone easily, and this is amazing! we don’t need to know the best ways to apply an ETL process, the tensor flow js library does it for us. Also, we can do operations directly in the tfjs, this is really nice because we just need one library to make an entire solution.

Backends

or the processing phase we have many backends available to use with TensorFlowJS, the two backends more common and recommended to use is WebGL and TensorFlow NodeJS, the tfjs will manage and choose the best backend to use if you have the two backends available, this means we can create more performative applications. We can see below that WebGL is faster than wasm and CPU backends, for that reason they recommend and built the TensorFlowJS to run better in this backend. See below the benchmark between the three backends.

#tensorflow #tensorflowjs #machine-learning #javascript #nodejs