Google has announced a new version of TensorFlow, the open source software library for machine learning, which is optimised for mobile. Called TensorFlow lite, the new library enables developers to build enhanced deep learning models to run on Android smartphones.
The launch is part of a move by Google to expand its portfolio of Artificial Intelligence (AI) enabled services on Android. TensorFlow Lite has been created to make running such services less bloated.
The open source software library can be used for a wide variety of machine learning tasks such as computation, data verification and analysis as well as designing machine learning algorithms.
Dave Burke, VP Engineering (Android) at Google, said:
“TensorFlow Lite will leverage a new neural network API to tap into silicate specific accelerators, and over time we expect to see DSPs (Digital Signal Processors) specifically designed for neural network inference and training. We think these new capabilities will help power the next generation of on-device speech processing, visual search, augmented reality, and more.”
Facebook rolled out a similar service last year which is now part of Style Transfer, the company’s photo stylisation tool. Caffe2Go has been designed to improve deep learning models across mobile devices.
Google’s and Facebook’s AI efforts are likely to pave the way for a whole series of new deep learning frameworks for mobile.