Posted by David So, Software Engineer, Google AI Since it was introduced a few years ago, Google’s Transformer architecture has been ap...
Google AICheck out Evolved Transformer, a new state-of-the-art Transformer architecture for natural language processing, derived using a fresh approach to evolution-based neural architecture search. Learn all about it here →
Posted by Karol Kurach, Research Lead and Olivier Bachem, Research Scientist, Google Research, Zürich The goal of reinforcement learning...
Google AICheck out a novel #ReinforcementLearning environment where agents aim to master the world’s most popular sport—football! The Google Research Football Environment includes benchmarks & progressive RL training scenarios, and is available in open source beta→goo.gle/2Mz8IqGpic.twitter.com/sTx6oWc3Br
Posted by Mingxing Tan, Staff Software Engineer and Quoc V. Le, Principal Scientist, Google AI Convolutional neural networks (CNNs) are c...
Google AIEfficientNet is an open source library that uses a new compound model scaling method and leverages recent progress in #AutoML to improve #neuralnetwork scaling techniques, achieving state-of-the-art accuracy with up to 10x better efficiency. Learn more ↓
Posted by Tali Dekel, Research Scientist and Forrester Cole, Software Engineer, Machine Perception The human visual system has a remarka...
Google AIReconstructing scene depth from video with moving subjects can be very challenging. Now there’s a #DeepLearning approach that generates accurate 3D depth maps from ordinary video, even when both camera and subjects are in motion. Learn more at → goo.gle/2HxYqCmpic.twitter.com/OOxeZTHseH
Posted by Ye Jia and Ron Weiss, Software Engineers, Google AI Speech-to-speech translation systems have been developed over the past sever...
Google AITranslatotron is our experimental model for direct end-to-end speech-to-speech translation, which demonstrates the potential for improved translation efficiency, fewer errors, and better handling of proper nouns. Learn all about it below!
Posted by Yifeng Lu, Software Engineer, Google AI Machine learning (ML) for tabular data (e.g. spreadsheet data) is one of the most acti...
Google AIInterested in applying machine learning to tabular data? Check out our work on an end-to-end AutoML solution, making ML scalable and accessible for both research and industry applications, and see how it performed in the recent @Kaggle_Days challenge!
Library of deep learning models and datasets designed to make deep learning more accessible and accelerate ML research. - tensorflow/tensor2tensor
Google AIJoin @anselmlevskaya and @lukaszkaiser at 4:15 at the #ICLR2019 Google booth to learn how to train #NeuralNetworks with Trax, a library of layers and models written in JAX, enabling simpler code to be compiled and run on GPUs and TPUs. Get the code at ↓
Building better products with on-device data and privacy by default. An online comic from Google AI.
Google AIFederated learning enables many devices to collaboratively train a global ML model without centralizing their data. Check out a comic book style introduction to FL, and dive deeper into details at g.co/federatedpic.twitter.com/0i3mNoxSHY
Posted by Bingyi Cao and Tobias Weyand, Software Engineers, Google AI Last year we released Google-Landmarks , the largest world-wide la...
Google AIAnnouncing Google-Landmarks-v2, a new dataset with 5M+ images of 200k+ landmarks from across the world, sourced and annotated by the @WikiCommons community. Learn more, including how to take part two associated @Kaggle challenges at ↓
Posted by Christine Kaeser-Chen, Software Engineer and Serge Belongie, Visiting Faculty, Google AI In recent years, fine-grained visual ...
Google AIWe’re excited to co-organize the 6th annual @cvpr2019 Workshop on Fine-Grained Visual Categorization (FGVC6), bringing together computer vision experts with specialists to address the challenges of applying FGVC to real-life settings. Learn more at ↓
Posted by Olivier Bachem, Research Scientist, Google AI Zürich The ability to understand high-dimensional data, and to distill that know...
Google AILearning disentangled representations of a scene is critical for many machine vision tasks. In collaboration with @ETH and @MPI_IS, we present a broad examination of the field, examine the role of implicit biases and provide direction for future research.
Google AICongratulations to the Google, @ETH and @MPI_IS authors of "Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations" (goo.gle/2IyFqTO), recipient of an #ICML2019 Best Paper Award! Learn more in the blog post at goo.gle/2KaMs48.
Posted by Daniel S. Park, AI Resident and William Chan, Research Scientist Automatic Speech Recognition (ASR), the process of taking an ...
Google AIAutomatic Speech Recognition (ASR) struggles in the absence of an extensive volume of training data. We present SpecAugment, a new approach to augmenting audio data that treats it as a visual problem rather than an audio one. Learn more at → goo.gl/KAPS5dpic.twitter.com/8t79Uc7uMU
Posted by Andrew Poon, Senior Software Engineer and Dhyanesh Narayanan, Product Manager, Google AI Perception Deep neural networks (DNNs...
Google AINeed to fine tune your #neuralnetwork to a specific task? Check out MorphNet, an open source technique for neural network model refinement that takes an existing network as input and produces a new one that is smaller and faster with better performance.
Posted by Navid Shiee, Senior Software Engineer and Aseem Agarwala, Staff Research Scientist, Google AI Taking a good group selfie can b...
Google AITaking good group selfies can be tricky. With Photobooth, a new shutter-free mode in the Pixel 3 Camera app, you can now shoot selfies that automatically capture you and your friends when you’re all looking your best! Learn how at ↓ goo.gl/KLFKsdpic.twitter.com/dTMc6OjBNm