Regardless of what is happening around the world, the AI community are one productive bunch, and they have something interesting to share almost every day. So, here’s a compilation of all the important releases for the ML developers from top companies like Google and Uber.
Here’s what is new this week:
Google Open-Sources Neural Tangent Library
The short history of deep learning indicates the incredible effectiveness of infinitely wide networks. Insights from these infinitely wide networks can be used as a lens to study deep learning.
However, implementing infinite-width models in an efficient and scalable way requires significant engineering proficiency.
To address these challenges and accelerate theoretical progress in deep learning, Google’s AI team released Neural Tangents, a new open-source software library written in JAX. This library is aimed at helping researchers build and train infinitely wide neural networks as easily as finite neural networks. Neural Tangents, at its core, provides an easy-to-use neural network library that builds finite- and infinite-width versions of neural networks simultaneously.
3D Object Detection With MediaPipe
Most of the object detection usually deals with 2D objects. The bounding boxes are always rectangles and squares but never a cube. By extending prediction to 3D, one can capture an object’s size, position and orientation in the world, leading to a variety of applications in robotics, self-driving vehicles, image retrieval, and augmented reality.
In order to fuel more interest in the 3D aspect of object detection, Google AI released MediaPipe Objectron, a mobile real-time 3D object detection pipeline for everyday objects. This pipeline detects objects in 2D images, and estimates their poses and sizes through a machine learning (ML) model, trained on a newly created 3D dataset.
Implemented in MediaPipe, an open-source, cross-platform framework for building pipelines to process perceptual data of different modalities, Objectron computes oriented 3D bounding boxes of objects in real time on mobile devices.
EfficientNet-Lite For Mobiles By TensorFlow
Last year in May, Google released a new family of image classification models called EfficientNet. These models, as the name suggests, were able to achieve state-of-the-art accuracy with fewer computations and parameters. EfficientNet was designed to open up novel applications on mobile and IoT, where computational resources are constrained.
Today, to match the needs of edge devices, EfficientNet-Lite gets released. It runs on TensorFlow Lite and is designed to perform well on mobile CPU, GPU, and EdgeTPU. EfficientNet-Lite brings the power of EfficientNet to edge devices and comes in five variants, allowing users to choose from the low latency/model size option to the high accuracy option (EfficientNet-Lite4). EfficientNet-Lite4, achieved 80.4% ImageNet top-1 accuracy, while still running in real-time (e.g. 30ms/image) on a Pixel 4 CPU.
Introducing Piranha: An Open-Source Tool to Automatically Delete Stale Code
Feature flags are a key part of Uber’s coding customisations. They are used for customising mobile app execution and serving different features to different sets of users.
However, after a feature has been 100 per cent rolled out to users, the feature flag in the code becomes obsolete. These nonfunctional feature flags now become a technical debt in the ML pipeline, making things difficult for developers who work on the codebase.
Removing this debt can be time-intensive and to automate this process, Uber developed Piranha.
Piranha is a tool that scans source code to delete code related to stale, or obsolete, feature flags, leading to a cleaner, safer, more performant, and more maintainable codebase. This tool has now been open-sourced, and developers can make use of feature flags with more ease.
Pixar’s Pioneers Get 2019 Turing Award
Edwin Catmull and Patrick Hanrahan of Pixar have been awarded the prestigious Turing Award for the year 2019. Pixar has been revolutionising how 3D objects can be generated using computers. Catmull and Hanrahan have been key in creating near realistic effects like reflections and curved surfaces game-changing 3D computer graphics techniques, which are now widely used in the film industry. They have now won the highest distinction in computer science: the Turing Award.
AlphaGo The Movie
On March 9, 2016, the worlds of Go and artificial intelligence collided in South Korea for an extraordinary best-of-five-game competition.
With more board configurations than there are atoms in the universe, the ancient Chinese game of Go has long been considered a grand challenge for artificial intelligence.
Celebrating 4 years of the success of AlphaGo, DeepMind’s breakthrough was documented in an hour-long video directed by Greg Kohs with an original score by Academy Award nominee, Hauschka, AlphaGo chronicles a journey from the halls of Oxford, through the backstreets of Bordeaux, past the coding terminals of DeepMind in London, and ultimately, to the seven-day tournament in Seoul. As the drama unfolds, more questions emerge: What can artificial intelligence reveal about a 3000-year-old game? What can it teach us about humanity?