Machine learning: Apache Flink ML 2.0 opens for Python

Share your love

The team behind Apache Flink has released Apache Flink ML in version 2.0. This is an accompanying library for machine learning purposes for the framework for processing data streams. Apache Flink ML provides both APIs and infrastructure to build stream-batch unified ML algorithms. These should be easy to use and offer almost real-time latency. The current release should make a significant contribution to expanding Apache Flink to new use cases from the machine learning area, in particular real-time ML scenarios.

In the second major version, the Apache Flink team not only carried out a major refactoring of the library, but also implemented new features: For example, Apache Flink ML now allows the use of stages in an ML workflow with multiple inputs and outputs and thus creating a Directed Acyclic Graph (DAG) with predefined levels.

Native online machine learning scenarios should now be possible because the getModelData()-API makes model data available as an unbounded stream. This model data can be transferred to web servers in real time and used for online inference.

In addition, a new Python SDK is available – taking into account that Python is often used for machine learning purposes. It can be installed with pip (pip install apache-flink-ml) and provides APIs for building ML algorithms that are similar to their Java counterparts. Future interoperability of the Python SDK with the Apache Flink ML Java library is already being planned.

With the new main version, the library is moving to its own GitHub repository for the first time nimble-ml under the Flink project around. This step is intended to accelerate the further development of the open source library due to lighter contribution workflows and separate release cycles.

Read Also   JavaScript framework: Next.js 11.1 fixes an open redirect vulnerability

Since Apache Flink ML 2.0 increases the dependency to the current version Apache Flink 1.14, the breaking changes introduced there go hand in hand.

All other details about version 2.0 are the official announcement such as den Release Notes refer to.


(May)

Article Source

Share your love