The rules governing the union recognition process can change depending on which political party controls the White House. The Trump administration made it more difficult for employees to form so-called micro unions, arguing that workers could simply cherry-pick a group that was more likely to vote for a union. Trump's NLRB overturned an Obama-era precedent where the burden of proof had been on employers to show that workers excluded from a union vote shared enough in common to become included -- a hard task when job positions often differ in hours, wages and duties.

The Defense Department's recent efforts to raise its artificial intelligence game have revealed a few obstacles. There are no cohesive goals across the military branches, and there is no way of knowing whether each service has enough people with the right skills. DOD should work with the services to establish AI-specific goals for cultivating technical talent, make it easier for all personnel to learn about AI and put it to use, and enable AI "rock stars" to succeed. It is currently impossible for the DOD to assess its AI posture, let alone assert leadership in AI. That's because posture assessment requires measurement.

With 5G, real-time computing will become a reality. The high speeds, high data throughput, and high number of connections that 5G enables will effectively erase the lag time between when data gets generated to when we can act on it. And, while self-driving vehicles might be the most visible new example of real-time processing most of us see, they are really only the tip of the iceberg, especially as private networks and network slicing roll out to bring the power of carrier-grade infrastructure to more locations and situations. IDC estimates that real-time data will grow by 50 times between 2000 and 2030 and constitute 30% of all data by then. Manufacturing will be one of the first places where a real-time data revolution takes place.

There's no easy fix to America's supply chain problem, but a major issue is a lack of drivers. Now a new type of scanner called 4-D LiDar offers a possible solution, which could be a huge step forward for self-driving cars and trucks. Anyone who has ever been temporarily blinded by high-beam headlights from an oncoming car will be happy to hear this. Audi's lighting tech can be focused enough to project images on a wall. U.S. highway safety regulators are about to allow new high-tech headlights that can automatically tailor beams so they focus on dark areas of the road and don't create glare for oncoming drivers.

As more and more companies adopt artificial intelligence (AI) in their operations, investors may be considering AI stocks in the stock market. After all, AI remains to be an increasingly relevant area of research now. For the uninitiated, AI deals with machines that attempt to emulate human intelligence. Most of us may think of sentient robots and futuristic gadgets upon hearing the term. But in reality, AI often revolves around complex algorithms and software which enable enterprises to make more effective decisions.

It may sound obvious, perhaps even clichéd, but this mantra is something that must be remembered in ongoing political negotiations over Horizon Europe, which could see Switzerland and the UK excluded from EU research projects. We need more, not fewer, researchers collaborating to solve today's and tomorrow's challenges. By closely working with Swiss and British researchers, who have long played key roles, Horizon Europe projects will benefit – as they have in the past. This is the motivation behind ETH Zurich, which collaborates with IBM Research on nanotechnology, leading the Stick to Science campaign. This calls on all three parties – Switzerland, the UK and the EU – to try and solve the current stalemate and put Swiss and British association agreements in place.

A new experiment showed how faces generated by artificial intelligence were indistinguishable from actual human faces. If that's not exciting (or scary) enough, the humans involved in the study also found AI-generated faces to be more trustworthy. As deep fakes become more common on social networking portals, watchdogs and critics alike are urging specialists to take note of its dangers - especially in crime. The two researchers also urged caution against AI-synthesised text, audio, video, and image that are used to commit fraud and propaganda. The study was published in Proceedings of the National Academy of Sciences USA and its co-author Hany Farid, a professor at the University of California, Berkeley was quoted by Scientific American as saying the following - "We found that not only are synthetic faces highly realistic, they are deemed more trustworthy than real faces... these faces could be highly effective when used for nefarious purposes."

The controversial facial recognition company Clearview AI reportedly told investors that it aims to collect 100 billion photos--supposedly enough to ensure that almost every human will be in its database. "Clearview AI is telling investors it is on track to have 100 billion facial photos in its database within a year, enough to ensure'almost everyone in the world will be identifiable,' according to a financial presentation from December obtained by The Washington Post," the Post reported today. There are an estimated 7.9 billion people on the planet. The December presentation was part of an effort to obtain new funding from investors, so 100 billion facial images is more of a goal than a firm plan. However, the presentation said that Clearview has already racked up 10 billion images and is adding 1.5 billion images a month, the Post wrote.

If not previously completed, a background security investigation will be required for all appointees. Appointment will be subject to the applicant's successful completion of a background security investigation and favorable adjudication. Failure to successfully meet these requirements may be grounds for appropriate personnel action. In addition, if hired, a background security reinvestigation or supplemental investigation may be required at a later time. Applicants are also advised that all information concerning qualifications is subject to investigation. False representation may be grounds for non-consideration, non-selection, or appropriate disciplinary action.

This paper describes the principle of "General Cyclical Training" in machine learning, where training starts and ends with "easy training" and the "hard training" happens during the middle epochs. We propose several manifestations for training neural networks, including algorithmic examples (via hyper-parameters and loss functions), data-based examples, and model-based examples. Specifically, we introduce several novel techniques: cyclical weight decay, cyclical batch size, cyclical focal loss, cyclical softmax temperature, cyclical data augmentation, cyclical gradient clipping, and cyclical semi-supervised learning. In addition, we demonstrate that cyclical weight decay, cyclical softmax temperature, and cyclical gradient clipping (as three examples of this principle) are beneficial in the test accuracy performance of a trained model. Furthermore, we discuss model-based examples (such as pretraining and knowledge distillation) from the perspective of general cyclical training and recommend some changes to the typical training methodology. In summary, this paper defines the general cyclical training concept and discusses several specific ways in which this concept can be applied to training neural networks. In the spirit of reproducibility, the code used in our experiments is available at \url{https://github.com/lnsmith54/CFL}.