Oracle has announced the availability of the Oracle Cloud Data Science Platform, which offers users the ability to deploy, train and manage machine learning algorithms in the cloud.
Platform for data science
Available as a native service on Oracle Cloud Infrastructure, the platform is built on DataScience.com acquired by Oracle in 2018. The latter offers collaborative capabilities for data scientists to work together using shared projects, model catalogs and the ability to implement team security policies.
Oracle’s platform supports Python and offers various open-source tools and libraries such as TensorFlow, Keras and Jupyter. Created as a collaborative space for data exploration, it supports machine learning experimentation and model training by data scientists in a way that is easily reproduced and audited.
Aside from support for Oracle Cloud SQL, the Oracle Cloud Data Science Platform includes Oracle Big Data Service which offers a full Cloudera Hadoop implementation. Organizations can also run Spark machine learning in-memory within one product for minimal data movement.
To accelerate data science, Oracle says the platform makes it easy for data scientists to access multiple services in the cloud, including cloud resources, and tools that deliver advanced visualization capabilities. It also includes Oracle’s auto ML technology which lets data scientists automate model selection and model optimization.
The platform also allows data scientists to explain data sets and identify what inputs are driving a model’s outputs, ensuring that decisions in relation to the line of business are clearly articulated. This is vital in regulated businesses with strict governance requirements and serves to ensure clarity in larger organizations as to why certain decisions were made.
Because model deployment is loosely coupled with the service context, a model’s effectiveness can be monitored and updated without disrupting the applications it might be enabling.
Photo credit: iStockphoto/photo_Pawel