I’m a fan of using Machine Learning everywhere. And few months ago, when I learned there are some teams that are working on libraries that can run on web browsers, I checked them right away. There are few popular choices: ConvNetJS, DeepLearning4j, sushi and its sisters sashimi, sukiyaki.

Among these choices, I love the idea of ConvNetJS, and now I finally got the chance and time to use and hack it. This page is not just a tutorial for beginners, but also a chance for me to study. And I’m not gonna lie, I’m a terrible HTML coder, all the knowledge I knew about HTML is from a big book from Apache that I read approximately 7 years ago.

The future belongs to your internet and browser, not some sort of big-ass server. Billions of computers out there are the most powerful computing resources on this planet.

How to start

The reason why I love ConvNetJS is because it offered a clean design. If you are just a user that wants to use it, then you might be interested in its core build convnet-min.js. Once you include it in your HTML source, then you are ready to go:

1<!-- some script here -->
2<script src="/path/to/convnet-min.js"></script>

On the another hand, I suggest you to fork the GitHub repo if you planned to make some changes, it’s here.

Start with a simple test