Google Releases Machine Learning Library

New deeplearn.js library runs in the browser and requires no programming.

The Google PAIR (People + AI Research) initiative has released an open source, WebGL-accelerated JavaScript library for machine learning. The deeplearn.js library requires no installation or back end and runs entirely in the browser.

According to the blog post by Google engineers Nikhil Thorat and Dnaiel Smikov, “While web machine learning libraries have existed for years, … they have been limited to the speed of JavaScript, or have been restricted to inference rather than training (e.g. TensorFire). By contrast, deeplearn.js offers a significant speedup by exploiting WebGL to perform computations on the GPU, along with the ability to do full backpropogation.” The library lets you train a convolution neural network to recognize photos and handwritten digits “all in the browser without writing a single line of code.”

The deeplearn.js home page includes links to demos showing deeplearn.js at work on real-world machine learning tasks, such as classifying photos.