Thanks to machine learning and new programming tools, the future smartphones will be a lot smarter. Dave Burke, Google’s VP of engineering announced that a TensorFlow Lite has arrived to bring machine learning to Android. And not just Android the new framework will also work on other devices.
Google Believes that in the future Android devices manufacturers will bring new models featuring new AI-focused chips called digital signal processors, specifically designed for neural network inference.
TensorFlow Lite will become a part of open-source TensorFlow project, And the next major updates of Android will come with Neural Network APIs. “We, at Google think these new capabilities will help power the next generation of on-device speech processing, visual search, augmented reality, and more,” said Dave Burke.
AI-focused chips digital signal processors could make possible for android devices to handle advanced deep learning Neural Networks without consuming much power.
Currently, it’s a painful job when it comes to implementing machine learning to mobile applications, especially, the training part which requires computational power, beefy hardware, a lot of time and a lot of power.
However, In future AI-focused chips would be able to fill full all the requirements for implementing machine learning to phones. Today, there’s only one mobile processor in the market with machine learning specific. The Qualcomm Snapdragon 835 the Hexagon DSP that also support TensorFlow.
DSPs are also popular for providing functionality like recognizing “OK, Google” to wakes up Google’s virtual assistant.
But Google isn’t the only company focusing on AI-based smartphones, Last year Facebook announced Caffe2Go — a mobile-oriented ML framework and the core of facebook real-time photo stylization tool.