One of the bigger challenges associated with building an advanced application that makes extensive use of machine learning algorithms is that everything has to be built from the ground up. There is no readily available database platform that developers can communally invoke to run this class of applications.
Simon Chan, senior director of product management for PredictionIO at Salesforce, says the goal is to make it simpler to build machine learning applications by allowing developers to reuse a database platform that includes much of the middleware needed to integrate those applications with traditional legacy enterprise and cloud native applications.
In general, the vast majority of machine learning applications are making use of algorithms that have been around in one form or another for years. What has changed is that the availability of inexpensive compute and storage in the cloud has made it less expensive to deploy these applications while also providing access to a massive pool of data that enables them to become more accurate.
Chan notes that PredictionIO will become an Apache project alongside Hadoop and the Spark in-memory computing framework for building analytics applications. Collectively, Chan says, Salesforce is betting that the availability of a reusable platform will accelerate the development of machine learning applications that leverage the full pantheon of Big Data Apache projects.
It may to be too early to say just what impact PredictionIO will have as an open source project. But one thing that is for certain is that the machine learning playing field is getting more level with each passing day.