
Google AI tool simplifies machine learning for businesses
Called Cloud AutoML, the tools are designed to enable developers with limited machine learning expertise to train high-quality models by leveraging the company’s transfer learning and neural architecture search technology. The first product to be released – Cloud AutoML Vision – is offered as a simple, secure, and flexible ML service that lets users train custom vision models for their own image recognition use cases.
Starting with as few as a few dozen photographic samples, users can upload them with the tool’s drag-and-drop interface, train and manage models, and then deploy the trained models directly on Google Cloud. According to the company, early results using the tool to classify popular public datasets like ImageNet and CIFAR have shown more accurate results with fewer misclassifications than generic ML APIs.
For customers with images but no labels yet, Google has a team of in-house labelers that will review a customer’s custom instructions and classify their images accordingly. The data will remain private, says Google, while being of the same quality and throughput that Google achieves with its own products.
Cloud AutoML is fully integrated with other Google cloud services. Users can store their training data in Google Cloud Storage. To generate a prediction on a trained model, users can use the existing Vision API by adding a parameter for their custom model, or use Cloud ML Engine’s online prediction service.
Several companies have been testing the Cloud AutoML tool for the past several months, says Google, including Disney and Urban Outfitters, to enhance the search and shopping functions on their websites.
“Cloud AutoML’s technology is helping us build vision models to annotate our products with Disney characters, product categories, and colors,” says Mike White, CTO, SVP, Disney Consumer Products and Interactive Media. “These annotations are being integrated into our search engine to enhance the impact on guest experience through more relevant search results, expedited discovery, and product recommendations on shopDisney.”
“Creating and maintaining a comprehensive set of product attributes is critical to providing our customers relevant product recommendations, accurate search results, and helpful product filters,” says Alan Rosenwinkel Ph.D., Data Scientist at Urban Outfitters. “However, manually creating product attributes is arduous and time-consuming. To address this, our team has been evaluating Cloud AutoML to automate the product attribution process by recognizing nuanced product characteristics like patterns and neck lines styles. Cloud AutoML has great promise to help our customers with better discovery, recommendation, and search experiences.”
Other Cloud AutoML services for all other major fields of AI will be released soon, Google says.
Related articles:
Google AI demo lets anyone explore machine learning without coding
Google acquires AI vision app startup
Google acquires image-recognition startup
Self-taught Google AI achieves ‘superhuman’ chess proficiency in hours
Google AI fund invests in algorithm marketplace startup
