JSR 381 Ballot Results

#JSR381 “VisRec” is a standard high-level API that allows Java developers a Java-centric API for basic Machine Learning (ML), image classification, and object detection. One of the goals of JSR #381 is to provide a common reusable design for Java machine learning development in different domains and use cases.

Over the past 12 months, we have engaged a wide collection of contributors for ideas, implementations, and tests. Using the VisRec mailing list (on groups.io) and our weekly status meetings (minutes available on the mailing list), we have had valuable discussions from several senior Java developers and JCP corporate members.

There are already several implementations. The reference implementation is based on Deep Netts, a pure Java deep learning library. Another exciting implementation was completed by Amazon (a member of our Expert Group) using their Deep Java Library (DJL), an open-source library to build and deploy ML applications natively in Java.

Our beta preview releases include basic hello-world examples for supported machine learning tasks (classification and regression) and image classification. Besides Amazon, we have had serious inquiries from GridGain and the Apache Ignite team to join the Expert Group. There are several blog posts and magazine articles on JSR381 pending over the next few weeks.

We’re working on finalizing the specification document and javadocs, performing extensive testing, updating TCK and integrating final changes based on the feedback that we’re getting from implementers.

Repos
https://github.com/JavaVisRec/visrec-api
https://github.com/JavaVisRec/visrec-ri/
https://github.com/JavaVisRec/jsr381-examples

Getting Started doc https://github.com/JavaVisRec/visrec-api/wiki/Getting-Started-Gui