New machine learning from Alibaba and Netflix, mimicking animal vision, and more open source news

Catch up on the biggest open source headlines from the past two weeks.

Image by:

Opensource.com

In this edition of our open source news roundup, we take a look an open source election auditing tool, new open source from Alibaba and Netflix, mimicking animal vision, and more!

Alibaba and Netflix share machine learning and data science software

Two companies at the forefront of machine learning and data science have just released some of their tools under open source licenses.

Chinese ecommerce giant Alibaba just open sourced the algorithm libraries for its Alink platform. The algorithms "are essential to support machine learning tasks such as online product recommendations and smart customer services." According to Jia Yangqing, president of Alibaba Cloud, Alink is a good fit for "developers seeking big data and machine-learning tools." You can find the source code for Alink (which is under an Apache 2.0 license) on GitHub, with documentation in both Chinese and English.

Not to be outdone, streaming service Netflix just released its Metaflow Python library under an Apache 2.0 license. Metaflow enables data scientists to "see early on whether a prototyped model would fail in production, allowing them to fix whatever the issue was". It also works with a number of Python data science libraries, like SciKit Learn, Pytorch, and Tensorflow. You can grab Metaflow's code from its GitHub repository or learn more about it at the Metaflow website.

Open source software to mimic animal vision

Have you ever wondered how your dog or cat sees the world? Thanks to work by researchers at the University of Exeter in the UK and Australia's University of Queensland, you can find out. The team just released software that allows humans to see the world as animals do.

Called micaToolbox, the software can interpret digital photos and process images of various environments by mimicking the limitations of animal vision. Anyone with a camera, a computer, or smartphone can use the software without knowing how to code. But micaToolbox isn't just a novelty. It's a serious scientific tool that can help "help biologists better understand a variety of animal behaviors, including mating systems, distance-dependent signalling and mimicry." And, according to researcher Jolyon Troscianko, the software can help identify "how an animal's camouflage works so that we can manage our land to protect certain species."

You can download micaBox or browse its source code on GitHub.

New tool for post-election auditing

More and more aspects of our lives and institutions are being automated. With that comes an increased danger of systems breaking down or malicious someones tampering with those systems. Open source gives us an opportunity to look at exactly how the automation works.

Elections, in particular, are increasingly vulnerable. To combat election tampering, the US Cybersecurity and Infrastructure Security Agency (CISA) has joined forces with the non-profit organization VotingWorks to create a web-based application for auditing ballots.

Called Arlo, the application is designed to ensure that "elections are secure, resilient, and transparent," said CISA's director Chris Krebs. Arlo works with a range of automated voting systems to help "officials compare audited votes to tabulated votes, and providing monitoring & reporting capabilities." Arlo was used to verify the results of recent state and local elections and is being further field-tested in the states of Georgia, Michigan, Missouri, Ohio, Pennsylvania, and Virginia.

Arlo's source code, released under an AGPL-3.0 license, is available on GitHub.

Royal Navy debuts open source application development kit

Consistency across user interfaces is key to a successful set of applications and services. The UK's Royal Navy understands the importance of this and has released the open source NELSON standards toolkit to help its developers and suppliers "save time and give users a consistent experience."

Named after the legendary British admiral, NELSON is intended to "maintain high visual consistency and user-experience quality across the different applications developed or subcontracted by the Royal Navy." The toolkit consists of a set of components including visual styles, typographic elements, forms, elements like buttons and checkboxes, and notifications.

NELSON has its own GitHub repository, from which the Royal Navy encourages developers to make pull requests.