Oracle Launches Cloud Data Science Service
Technology: This new service aims to provide data specialists with a collaboration platform that guides data science and machine learning projects throughout their life cycle.
Oracle announced Wednesday the launch of the new service “Cloud Infrastructure Data Science”, a native service on Oracle Cloud Infrastructure, designed to allow teams of data scientists to collaborate in the development, deployment and maintenance of machine learning models.
As Oracle expands the footprint of its “second generation” cloud, the new service aims to surpass the services that other public cloud providers already offer – and the problems encountered by data scientists in their daily lives.
“One of the traditional problems in data science, and I think that’s still what we generally see in almost all organizations, is that data scientists really work in silos, in isolation,” said told ZDNet Greg Pavlik, Oracle vice president for AI product, data and service development. “The objective of the service is to bring them together as a team, in a collaborative environment, allowing them to work together and allow organizations to monitor their work … without creating obstacles for data scientists”. Providing customers with a complete data science platform is a “critical” part of Oracle’s cloud strategy, he adds. Oracle already owns one of the world’s largest SaaS companies,
Make data science more collaborative
“In both cases, using this data to guide business decisions, either in the application or in the individual databases themselves, is one of the first things customers try to do,” says Greg Pavlik. Other public cloud providers offer tools to make data science more collaborative. Google, for example, introduced Kubeflow Pipelines and its AI Hub to maximize the impact of data science in an organization.
Greg Pavlik says Oracle’s new service stands out for its emphasis on teamwork. Chief data officers and CIOs “wonder how to make their teams effective. How to make these teams responsible, how to ensure that they understand that data scientists are building data that they can really use? It’s all part of the service. ”
The key capabilities of the new services can be classified into four categories, he adds. First, it offers a collaborative space for data exploitation, machine learning experimentation and model training. Second, it includes what Oracle calls an “accelerated data science toolkit”. Python’s native library offers multiple key capabilities, such as access to the underlying cloud resources, as well as productivity tools like advanced visualization capabilities. It also includes Oracle Auto ML – capabilities that allow data scientists to largely automate model selection and optimization.
The toolkit also includes model explanation capabilities, allowing users to explain which datasets and which inputs are behind the results of a model. “It is very important to be able to explain frankly what drives these decisions at the industry level,” says Greg Pavlik. “But in companies that are regulated in areas where governance requirements are strict, you also need to be able to explain why decisions are made. “
The third component of the Data Science service is a catalog of models through which a data science specialist can make models available to other users, including other specialists, business analysts or application developers. “Anyone who tries to use these models to make business decisions, or to build reports or logical applications, can very simply consume the model and integrate it into their own context without having to have specialized skills or knowledge in machine learning ”.
Finally, the service offers the deployment of models in the context of loosely coupled services. Users can monitor the effectiveness of a model and update it without disrupting the application or consumers of the model. “This paradigm corresponds to an emerging practice on how to manage the complete life cycle of a machine learning model,” concludes Greg Pavlik.