Google LLC recently introduced a raft of new products and capabilities within its cloud portfolio, including several new products and features in Contact Center AI and new versions of Document AI. It has also introduced improvements to the AI platform for Machine Learning Operations (MLOps) practitioners. It will allow machine learning developers, data scientists, and data engineers to combine their ideas with ML and develop them into real projects that they can quickly deploy in production without any extra costs.
Google’s AI platform is built on Kubeflow, an open-source platform developed by Google that allows developers to create portable ML pipelines that can run on-premise infrastructure or Google’s cloud offering. With this, developers can access Google’s machine learning framework TensorFlow, its BigQuery data store, and its cloud-based Tensor Processing Units.
In the latest blog post, Craig Wiley, director of product management at Google Cloud Artificial Intelligence, announced a long list of new features for the AI platform to simplify MLOps. It includes a new, fully managed service to build ML pipelines, the preview will be available in October.
The upcoming managed service will allow users to build ML pipelines using TensorFlow Extended’s pre-built components and templates, and lessen the effort required to deploy new models.
Google is also adding Continuous Evaluation service into the mix. It samples prediction input and output from deployed ML models and analyzes their performance. The search engine behemoth also made the promise that Continuous Monitoring will be available by December 2020. It is used to monitor ML model performance in production to warn customers if a particular model is going stale or any outliers, skews, or concept drifts need to be fixed.
Wiley also announced a new ML Metadata Management service for the AI platform, and it will be previewed in September. It can track important artifacts and experiments, and render a ledger of curated actions and detailed model lineage. He noted, “this will enable customers to determine model provenance for any model trained on AI Platform for debugging, audit, or collaboration.”
By the end of the year, the AI Platform will also get a new updated feature store that will serve as a centralized, organization-side repository of all historical and new feature values that can be bettered as desired by ML developers. Wiley also added, “this will boost productivity of users by eliminating redundant steps in feature engineering. “
Head of Google Cloud Artificial Intelligence & Industry, Mr. Andrew Moore revealed that Vizier, a new service that autotunes the hyperparameters of ML models to get the best output, is now available in beta. He added, “AI Platform’s Notebooks service, which provides an integrated and secure JupyterLab environment for data scientists and developers to experiment, develop, and deploy ML models into production, is now generally available.”
Other upcoming new features of the AI Platform include an update to Cloud AI Building Blocks. It provides access to commonly used models around AI-based vision, translation, and speech through an application programming interface. This service will also add AutoMl as an integrated function in the workflow. In simpler terms, more no-code and code-based options for developing custom ML models faster, Moore said.
Moore also mentioned Google’s Contact Center AI platform, a suite of services that applies artificial intelligence to automate enterprise contact center operations.
Here the main update is the new version of Google’s virtual agent named Dialogflow CX. It has been especially created for enterprises with large contact center operations. This new agent can also support complex and multiturn conversations, and is “truly omnichannel,” meaning it can be built once and deployed anywhere, Moore said.
Google also added some new AI-specific services, it includes Lending Document AI. It has been designed for the mortgage industry to speed up loan applications by automatically processing borrowers’ income and asset documents. However, Procure-to-pay Document AI is a new service that companies can use to automate procurement cycles, whereas the latest Media Translation API provides real-tech speech translation from audio data.
Lending Document AI is now available in Alpha, while Procure-to-pay Document AI and Media Translation API Moore are in beta.