MMLSpark header graphic - polygons on a purple background

SynapseML

Simple and Distributed Machine Learning

Synapse Machine Learning (SynapseML) expands the distributed computing framework Apache Spark (opens in new tab) in several new directions. SynapseML adds several machine learning frameworks to the SparkML Ecosystem, including LightGBM (opens in new tab)Vowpal Wabbit (opens in new tab)OpenCV (opens in new tab)Isolation Forest, (opens in new tab) and the Open Neural Network Exchange (ONNX) (opens in new tab). These tools allow users to craft powerful and highly-scalable models that span multiple ML ecosystems.

SynapseML also brings new networking capabilities to the Spark ecosystem. With the HTTP on Spark project, users can embed any web service into their SparkML models and use their Spark clusters for massive networking workflows. In this vein, SynapseML provides easy-to-use SparkML transformers for a wide variety of Microsoft Cognitive Services (opens in new tab). Finally, the Spark Serving project enables high throughput, sub-millisecond latency web services, backed by your Spark cluster.

Visit the SynapseML web site (opens in new tab) or Github repository (opens in new tab) to learn more.