AutoML – Databricks’ Automated End-To-End Machine Learning
Databricks , known for their Big Data Spark Contribution, has recently released AutoML. It is an automated toolkit for developers to enable them end-to-end deployment of Machine Learning models. AutoML is built completely on Spark. Although Spark already has capability of Machine Learning with it’s ML & MLLIB libraries , AutoML is one step ahead to make Data Science Solution building Automated and yet empowering the Data Scientists or Engineers based on how much they want to switch between complete or minimalistic control based on their expertise. Salient Features –
- Automating end-to-end machine learning pipelines, including model search, and deployment made available via Databricks Labs custom solutions
- Available for Integration with Microsoft Azure Machine Learning.
- Feature Engineering , Hyperparameter Optimization & Selection etc are supported.
- Integration with Hyperopt and MLflow gives optimized and distributed hyperparameter and model search capability.