Release initially on November 9, 2015, Google’s open source platform for Artificial Intelligence, Machine Learning and Deep Learning called TensorFlow, has had a huge impact on AI research. TensorFlow has helped researchers predict cancer, ease machine translation, enhance image recognition, and speed up speech recognition. In February this year, Google announced that they were “excited to see people using TensorFlow in over 6000 open-source repositories online.“ Today, major companies like Uber, Twitter, Snapchat, Qualcomm, Airbnb, Ebay and several others implement TensorFlow.

The Launch of TensorFlow Lite

Unfortunately, TensorFlow was primarily for computers, until now. In Google’s I/O, Dave Burke, the vice president of engineering for Android, announced the soon-to-be-released TensorFlow Lite for mobile. He 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.” He further added “We think these new capabilities will help power the next generation of on-device speech processing, visual search, augmented reality, and more.”

This new library will allow developers to build Machine Learning apps which run faster while being less expensive on the system. The framework will enable development of deep learning models, and thus bring AI with a greater impact to the mobile platform. We could expect social media apps to become smarter by giving better recommendations. Photo, video or document editing apps could soon have cooler tools, enabling higher customizability and enhancing ease of use. New games could have better Augmented Reality graphics and smoother gameplay. Speech recognition and translation apps would become more user friendly, and way smarter. Most importantly, TensorFlow Lite will enable smaller companies to develop high quality AI apps, thus giving them an opportunity to grow significantly.

Any AI software requires training a huge dataset, which remains a major task, and a task which is computationally too expensive to be handled by smartphones. So in spite of TensorFlow Lite, the training of the models would still have to be performed on the cloud. But as Burke mentioned, Android O will introduce “a new framework” to hardware to enable accelerated neural computations. We can be sure that the future will see several modifications not only in software, but in hardware too. Thus, powered with these new technologies, we can push AI to new limits.