Microsoft has joined the SciKit-learn Consortium, the organisation supporting construction of the free software machine-learning library for Python.
The software giant will support central contributions and tackle new features and has “committed to helping” SciKit-learn users in training and production through its own services and open-source projects.
The company is supporting SciKit-learn in the Open Neural Network Exchange Runtime, with a SKlearn-ONNX converter that exports common SciKit-learn pipelines to ONNX-ML for use on Linux, Windows and Mac.
Microsoft last year open-sourced the ONNX interface engine for machine models built using the ONNX format it developed in conjunction with AWS and Facebook.
“We are extremely supportive of what the SciKit-learn community has accomplished so far and we want to see it continue to thrive and expand,” Microsoft AI frameworks principal program manager Prasanth Pulavarthi wrote here.
Pulavarthi also promised “strong support” for SciKit-learn training on Microsoft’s Azure ML, with users able to take their existing training scripts and scale them up on the cloud service before tuning them, and logging experiments. The platform could also generate the “best” SciKit-learn pipeline “according to your training data and problem scenario,” he claimed.
SciKit-learn, a Google summer of code project, started life in 2007. The BSD-licensed effort provides tools for data mining and analysis and spans classification, regression, clustering, dimensionality reduction, model selection and pre-processing.
Users include Spotfy, JP Morgan, Evernote, OK Cupid and Mars.