Ai blog 2 (3)

Vertex Delivers Managed Machine Learning

At last week’s annual developer conference, Google I/O, Google announced Vertex AI, a managed machine learning platform designed to make it easier for developers to deploy and maintain their artificial intelligence models.

An announcement like this is huge at a time when companies are facing all kinds of challenges when deploying and using machine learning. As reported by market data aggregator, Statista, the nature of those hurdles cover a broad range. Meeting IT, governance, security, auditability requirements presents problems for greater than 50% of companies. Thirty-seven percent of companies surveyed report monitoring model performance as a chief concern. Still, other businesses (36%) say the chores of updating and maintaining model quality and performance are their biggest concerns.

Google’s announcement is well-timed with the market’s enthusiasm for artificial intelligence (and specifically machine learning) pitted against the challenges of using that technology. Here again, Statista shines a light on the acceptance of AI. Nearly a fourth of 1,032 businesses surveyed say that they will have fully deployed AI into business operations by the end of 2021.

AI-blog

The Need for Vertex

The goal of any business is to see a return on investment (ROI). So far, however, the returns from machine learning have been elusive. The reasons for this are varied, including those challenges stated above, and include things like cultural commitment and acceptance within organizations. Regardless of the reason, Craig Wiley, the director of product management for Google Cloud’s AI Platform, says the cycle of investment without return needs to come to an end. As reported by Techcrunch, Wiley calls what’s happening with machine learning in enterprises a crisis;

 “As someone who has worked in that space for a number of years, if you look at the Harvard Business Review or analyst reviews, or what have you — every single one of them comes out saying that the vast majority of companies are either investing or are interested in investing in machine learning and are not getting value from it.”

In its announcement about the release of Vertex, Google says the time of data scientists grappling with the challenge of manually piecing together machine learning solutions must come to an end. That means deploying solutions with fewer lines of code, having ML accessible to engineers at all levels, making the training of models easier, and facilitating easier deployment and maintenance.

How Google Has Made ML Different

Google Vertex AI uses a singular environment to handle all aspects of machine learning building, training, deploying, and maintenance. Using one unified user interface and an API, businesses can go from concept to production more efficiently. And because Vertex also facilitates the discovery of patterns and anomalies, enterprises are better prepared to adjust their models to meet dynamic market conditions.

In addition to the features Google advertises with Vertex, it also says engineering teams will be more empowered with:

Access to the same AI toolkit Google uses internally, including computer vision, language, conversation, and structured data.

MLOps features like Vertex Vizier, which increases the rate of experimentation, access to the Vertex Feature Store, and Vertex ML Edge Manager for deploying and monitoring models on the edge.

MLOps tools like Vertex Model Monitoring, Vertex ML Metadata, and Vertex Pipelines enable trouble-free model management.

AI-pic-for-blog

Vertex In Practice

Google’s goals for Vertex are twofold. First, the company wants to make things easier for data scientists and engineers. Second, it hopes to get businesses to help businesses get out of neutral and accelerate their projects into full-scale production. Early evidence suggests that it is succeeding.

In 2018, French personal care company L’Oréal acquired ModiFace, an augmented reality, and artificial intelligence startup. ModiFace lets consumers try beauty products such as hair color, makeup, and nail color, virtually, in real-time. The skin diagnostic application is trained on thousands of images from L’Oréal’s Research & Innovation using the Vertex AI platform. The result is a technology that allows users to have incredibly real-to-life experiences with cosmetics and other beauty products.

In other arenas, global data and measurement-driven media agency, Essence, uses Vertex to create AI models that dynamically adjust with human behavior. The MLOps capabilities in Vertex means the company can keep pace with its clients’ rapidly evolving needs.

Getting started with Vertex AI

To jump-start engagement with Vertex, Google has documented best practices for implementing machine learning on Google Cloud as well as an MLOps whitepaper. For expert insights and an evaluation of your AI and ML needs, WALT Labs can help. Contact one of our experts today to get started.