AI weekly (51/2019)

My selection of news on AI/ML and Data Science

AI weekly (51/2019)

My selection of news on AI/ML and Data Science

+++ Facebook has a neural network that can do advanced math +++ Google’s AI can identify wildlife from trap-camera footage with up to 98.6% accuracy +++ Intel acquires AI chip startup Habana Labs for $2 billion +++ eBay open-sources AI platform Krylov +++ AI experts urge machine learning researchers to tackle climate change +++ Top 10 Python libraries of 2019 +++

Breakthrough — Or So They Say

Facebook has a neural network that can do advanced math. An article by Technology Review features the preprint arXiv:1912.01412 by Guillaume Lample and François Charton, from Facebook AI Research in Paris, that I had mentioned in an earlier post.

Google’s AI can identify wildlife from trap-camera footage with up to 98.6% accuracy. Google has partnered with Conservation International and several other organizations in order to process one of the world’s largest and most diverse databases of photographs taken from motion-activated cameras. As of today, the fruits of their labor is available through Google Cloud as a part of Wildlife Insights, an AI-enabled platform that streamlines conservation monitoring by expediting trap-camera photo analysis. Google says that its AI Platform Prediction service, which hosts trained AI models in the cloud and runs data through those models with as little latency as possible, has made it possible for Wildlife Insights to analyze up to 3.6 million photos an hour. Using TensorFlow, they automatically remove photos that don’t contain an animal and achieved high accuracy across 614 species, with an 80% to 98.6% classification accuracy for species like jaguars and African elephants, as well as rarer wildlife, like white-lipped peccaries.

Tools and Frameworks

eBay will be open-sourcing Krylov, its AI platform. According to a VentureBeat interview with Sanjeev Katariya, eBay’s VP and chief architect of AI and platforms, eBay intends to open source Krylov, a machine learning platform that enables data scientists and machine learning engineers to ship models for all kinds of data quickly into production. Some, but not too many, details are mentioned in an eBay tech blog post.

Top 10 Python libraries of 2019. tryo labs, a US based machine learning consultancy, lists their top 10 Python libraries every year in a blog post. Among this year’s winners are Modin which scales your pandas workflow by changing a single line of code, Streamlit which lets you create apps for your machine learning projects with simple Python scripts, Transformers which provides state-of-the-art general-purpose architectures for NLP with over 32+ pretrained models and interoperability between TensorFlow 2.0 and PyTorch, Detectron2 by facebook that implements state-of-the-art object detection algorithms based on PyTorch, and Metaflow from Netflix which helps engineers build and manage real-life data science projects.

Business News and Applications

Intel acquires AI chip startup Habana Labs for $2 billion. Intel Corporation today announced that it has acquired Habana Labs, an Israel-based developer of programmable deep learning accelerators for the data center for approximately $2 billion. Founded in 2016, Habana has raised about $120 million from investors. In June, the startup debuted Gaudi, an eight-core processor designed for the sole purpose of training AI models. Habana showed in an internal benchmark test that Gaudi managed to train ResNet-50 models 3.8 times faster than Nvidia’s flagship V100 graphics card for data centers. The startup has a second chip called Goya that’s touted as having record-beating performance as well. It’s based on the same eight-core architecture as Gaudi but features a different design optimized for inference, or running AI models once they’re already trained and have started working with live data. Prior to this move, Intel had acquired AI chip producers Altera, Nervana and Movidius, and earlier this year Vertex.ai, a startup developing a platform-agnostic AI model suite.

Publications

AI experts urge machine learning researchers to tackle climate change. At the Tackling Climate Change workshop at this year’s NeurIPS conference, some of the top minds in machine learning came together to discuss the effects of climate change on life on Earth, how AI can tackle the urgent problem, and why and how the machine learning community should join the fight. The panel included Yoshua Bengio, MILA director and University of Montreal professor; Jeff Dean, Google’s AI chief; and Andrew Ng, cofounder of Google Brain and founder of Landing.ai. In June, Bengio and Ng had already joined a cadre of more than 20 Climate Change AI steering committee and advisor members, including DeepMind founder Demis Hassabis. Together they published a paper titled “Tackling Climate Change with Machine Learning,” which contains 650 references.

Tutorials

See also