Google announces TensorFlow 2.0 Alpha, TensorFlow Federated and new TensorFlow APIs at 2019 TensorFlow Dev Summit. The 2019 TensorFlow Dev Summit was held on March 6 and 7, 2019 at Sunnyvale, CA.
The continuous focus of Google in churning out innovative tech products and solutions based on artificial intelligence is continuing. Recently the tech giant has come with an array of development tools, documentation, tutorials, and development platforms to allow the web and mobile app developers incorporating machine learning technology in their applications. Among these new projects, TensorFlow seems to be the most exciting project with far-reaching implications.
TensorFlow has been unrolled as the open-source app development platform that allows individual developers and development teams to build and train models through Machine Learning and utilise the same in their app development projects. Google for the last two years is already holding developer summit for this platform and in the latest and third annual TensorFlow Developer Summit, the company made a few significant announcements including the release of TensorFlow 2.0. version. Let us briefly take a look at the different releases at the summit.
TensorFlow 2.0 Alpha
In the summit, the first publicly available alpha version of TensorFlow 2.0 is launched. The latest version reportedly made a huge step forward by simplifying the workflow and making the development a lot easier for everyone. The new version of TensorFlow will run more aptly and almost as soon as it is summoned. This has just been the alpha version and the full and definitive version of the platform is going to release in the second quarter of 2019.
In the summit, TensorFlow developers also came with a solution to address the shortcomings of utilising Big data. Big Data While is considered to offer huge potential by tracking numerous facets of data across devices, having been subjected to centralised control most of the times, there are tremendous privacy concerns for the end user. Acknowledging this, shortcoming developer launched TensorFlow Federated which is a sophisticated open-source framework to allow developers using Machine Learning insights while handling the data locally.
In the recently concluded TensorFlow summit Google developers also made some significant announcements about new approaches concerning privacy and data security. The TensorFlow Privacy approach is all about processing user data without violating user privacy. This will be achieved by not accessing sensitive portion of user data in the Machine Learning based model.
Coral Development Platform
Apart from the above-mentioned frameworks and approaches Google also made some crucial announcements concerning a new development platform. Google now introduced Coral that allows developers to go beyond the web and build localised hardware solutions. This new Coral development platform will help developers bypassing the constraints of the web and build solutions locally with ready to use software and hardware components.
Coral Dev Board comes as the first hardware component incorporated with this platform. The component comes loaded with an Edge TPU (Tensor Processing Unit). The best thing about the component is that it has robust and performance-savvy Machine Learning capabilities. It comes equipped with all a developer needs including RAM, eMMC modules, WiFi, and Bluetooth. Moreover, it also allows developers to add some hardware as they need.