News

Google Enhances ML Kit, for Mobile Machine Learning

• By David Ramel
• April 22, 2019

Google's Android developer team updated ML Kit, which packages up the company's machine learning expertise and technology for mobile developers creating Android or iOS apps.

ML Kit provides base APIs ready to be used "out of the box" for tasks such as creating custom models used on-device or in the cloud. The kit is used in conjunction with Google's Firebase mobile development platform that includes a host of tools that can add back-end functionality for the development of mobile apps, including analytics, storage, syncing, cloud functions, authentication and more.

Optimized for mobile, the kit -- still in beta -- provides functionality for things like image labeling, barcode scanning, text, or optical character recognition (OCR), face detection and so on.

Thanks to a recent update, ML Kit now offers two more pieces of functionality: Language Identification and Smart Reply.

"You might notice that both of these features are different from our existing APIs that were all focused on image/video processing," the team said in a post earlier this month. "Our goal with ML Kit is to offer powerful but simple-to-use APIs to leverage the power of ML, independent of the domain. As such, we are excited to expand ML Kit with solutions for Natural Language Processing (NLP)!

"NLP is a category of ML that deals with analyzing and generating text, speech, and other kinds of natural language data. We're excited to start out with two APIs: one that helps you identify the language of text, and one that generates reply suggestions in chat applications. Both of these features work fully on-device and are available on the latest version of the ML Kit SDK, on iOS (9.0 and higher) and Android (4.1 and higher)."

The new Smart Reply API provides functionality that is increasingly being found in messaging apps: a list of suggested responses that can be sent as an action taken in reply to a notification or from within a mobile app.

"The API provides suggestions based on the last 10 messages in a conversation, although it still works if only one previous message is available," Google said. "It is a stateless API that fully runs on-device, so we don't keep message history in memory nor send it to a server."

The post promised more news to come about ML Kit at the company's upcoming Google I/O conference, set for May 7-9.