Google announces TensorFlow 2.0 Alpha and more from TWIML & AI

1024 684 The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)

Bits & Bytes

  • Google announces TensorFlow 2.0 Alpha, TensorFlow Federated, TensorFlow Privacy. At the 3rd annual TensorFlow Developer Summit, Google announced the first alpha release of TensorFlow 2.0 and several other new releases such as: TensorFlow Federated – a new open-source framework that allows developers to use all the ML-training features from TF while keeping the data local; TensorFlow Privacy – which uses differential privacy to process data in a private manner; extensions to TensorFlow Extended (TFX), a platform for end-to-end machine learning; and Activation Atlases – which attempts to visualize and explain how neural networks process images.
  • Google open sources GPipe, a library for parallel training of large-scale neural networks. GPipe, which is based on the Lingvo (a TensorFlow framework for sequence modeling), is applicable to any network consisting of multiple sequential layers and allows researchers to “easily” scale performance. [Paper]
  • Facebook AI researchers create a text-based adventure to study how AI speak and act. Researchers from Facebook and University College London specifically investigated the impact of grounding dialogue – a collection of mutual knowledge, beliefs, and assumptions essential for communication between two people–on AI agents.
  • Google announces Coral platform for building IoT hardware with on-device AI. Coral targets developers creating IoT hardware from prototyping to production. It is powered by a TPU that is specifically designed to run at the edge and is available in beta.
  • Google and DeepMind are using AI to predict the energy output of wind farms. Google announced that it has made energy produced by wind farms more viable using DeepMind’s ML algorithms to better predict the wind output.
  • Ben-Gurion U. develops new AI platform for ALS care. Researchers at Ben-Gurion University have used ML models to develop a new method of monitoring and predicting the progression of neurodegenerative and help identify markers for personalized patient care and improve drug development.
  • Google rolls out AI grammar checker for G Suite users. Google applies ML techniques to understand complex grammar rules and identify “tricky” grammatical errors by G Suite users.


Dollars & Sense

To receive the Bits & Bytes to your inbox, subscribe to our Newsletter.

Leave a Reply

Your email address will not be published.