predictiveanalyticstoday.com
ConvNetJS in 2024 - Reviews, Features, Pricing, Comparison - PAT RESEARCH: B2B Reviews, Buying Guides & Best Practices
ConvNetJS is a Javascript library for training Deep Learning models (Neural Networks) entirely in users’ browsers. Users just open a tab and they are training. No software requirements, no compilers, no installations, no GPUs, no sweat. The library allows users to formulate and solve Neural Networks in Javascript, and was originally written by @karpathy (a PhD student at Stanford). However, the library has since been extended by contributions from the community. The code is available on Github under MIT license. Pull requests for new features / layers / demos and miscellaneous improvements are encouraged. The library is also available on npm for use in Nodejs, under name convnetjs. There are two ways to use the library: inside the browser, or on a server using node.js. The fastest way to obtain the library in a plug-and-play way for users who don't care about developing is through a link to convnet-min.js, which contains the minified library. Alternatively, users can also choose to download the latest release of the library from Github. The file that they would probably be most interested in is build/convnet-min.js, which contains the entire library. And to use it users must create a bare-bones index.html file in some folder and copy build/convnet-min.js to the same folder. The entire library is based around transforming 3-dimensional volumes of numbers. These volumes are stored in the Vol class, which is at the heart of the library. The Vol class is a wrapper around a 1-dimensional list of numbers (the activations, in field .w), their gradients (field .dw) and lastly contains three dimensions (fields .sx, .sy, .depth).
PredictiveAnalyticsToday ReviewDesk