Air

At a demonstration of Amazon Web Services' new artificial intelligence image recognition tool last week, the deep learning analysis calculated with near certainty that a photo of speaker Glenn Gore depicted a potted plant. Artificial intelligence – in its application of deep learning neural networks, complex algorithms and probabilistic graphical models – has become a'black box' according to a growing number of researchers. While "humans are surprisingly good at explaining their decisions," said researchers at University of California, Berkeley and the Max Planck Institute for Informatics in Germany in a recent paper, deep learning models "frequently remain opaque". Their December paper Attentive Explanations: Justifying Decisions and Pointing to the Evidence, primarily focused on image recognition, makes a significant step towards AI that can provide natural language justifications of decisions and point to the evidence.

For the atmospheric scientist however, the eclipse provides a shining opportunity to directly study how the sun influences weather patterns by heating the atmosphere. To that end, a team of researchers from Oklahoma State University and the University of Nebraska is going to spend Monday tracking changes in the atmosphere in the path of the eclipse. And to get just how the eclipse changes the weather in the low sky, the team will fly drones during the totality. "There's an impact during what we call the diurnal cycle, the night-day boundary, the sun comes out, starts heating up the ground, and that's where a lot of our unstable weather phenomena starts to form," says Jamey Jacob, a professor of Mechanical and Aerospace Engineering at Oklahoma State University.

Dino Mehanovic, John Bass, Thomas Courteau, David Rancourt, and Alexis Lussier Desbiens from the University of Sherbrooke realized that perching with a fixed-wing aircraft doesn't need to involve a stall to achieve that vertical and ultra low-speed approach, as long as you can maintain control over the aircraft. We are thinking about various failure causes (unsuitable states during the approach, smooth surface for the microspines) and failure detection timing (before touchdown, at touchdown and after touchdown). You also have to consider numerous factors that are sometime hard to quantify: efficiency of gears, reuse of some components between flight and climbing, transition time, propeller size, operating away from the design point, battery size, etc. Autonomous Thrust-Assisted Perching of a Fixed-Wing UAV on Vertical Surfaces, by Dino Mehanovic, John Bass, Thomas Courteau, David Rancourt, and Alexis Lussier Desbiens from the University of Sherbrooke in Canada, was presented at the 2017 Living Machines Conference at Stanford, where it won a Best Paper award.

The advanced machine learning system, which goes by the faintly sinister name of Google Vizier, automatically tunes algorithms right across Google's parent company Alphabet. Google Vizier cuts short this tedious manual task by automatically optimising hyperparameters of machine learning models. "Our implementation scales to service the entire hyperparameter tuning workload across Alphabet, which is extensive," they write in a paper released this week, citing an example where Google researchers "used Vizier to perform hyperparameter tuning studies that collectively contained millions of trials for a research project… That research project would not be possible without effective black–box optimisation." As well as helping research, Google Vizier is being put to use at Alphabet, where, its creators write, it "has made notable improvements to production models underlying many Google products, resulting in measurably better user experiences for over a billion people".

Real-life flying car imagined for DeLorean's next model Stassi Schroeder shows off trip to Mexico with Rachael O'Brien The team combined two models to create the glider's predictive AI: the partially observable Markov decision process and another AI approach called Bayesian reinforcement learning. He and the team combined two models to create the glider's predictive AI: the partially observable Markov decision process and another AI approach called Bayesian reinforcement learning. Anything that will use sophisticated AU systems to operate real, unpredictable movements could benefit, including driving cars, keeping homes secure and even planning personal schedules. Anything that will use sophisticated AU systems to operate real, unpredictable movements could benefit, including driving cars, keeping homes secure and even planning personal schedules.

The advanced machine learning system, which goes by the faintly sinister name of Google Vizier, automatically tunes algorithms right across Google's parent company Alphabet. Google Vizier cuts short this tedious manual task by automatically optimising hyperparameters of machine learning models. "Our implementation scales to service the entire hyperparameter tuning workload across Alphabet, which is extensive," they write in a paper released this week, citing an example where Google researchers "used Vizier to perform hyperparameter tuning studies that collectively contained millions of trials for a research project… That research project would not be possible without effective black–box optimisation." As well as helping research, Google Vizier is being put to use at Alphabet, where, its creators write, it "has made notable improvements to production models underlying many Google products, resulting in measurably better user experiences for over a billion people".

The news web site BuzzFeed did just that, reporting this week that it employed a machine-learning algorithm to first recognize known spy planes, and then combine that model with a large set of flight-tracking data from a commercial web site. The AI project mapped thousands of surveillance flights operated by federal agencies over a four-month period, including a military contractor tracking terrorists in Africa that is also flying surveillance aircraft over U.S. cities, BuzzFeed reported. The ground radars sweep up a flight data transmitted by aircraft transponders, including unique identifiers for each plane. "Given that spy planes tend to fly in tight circles, it put most weight on the planes' turning rates," the web site reported.

Marshals Service along with military aircraft and surveillance flights operated by military contractors. It turned out the sorties over the Bay Area and southern California supported of U.S. Special Operations Command training missions. The machine-learning exercise also turned up a surprising number of local and state police aerial surveillance operations in Arizona, Florida, southern California and Ohio, the web site reported. It also spotted testing of special operations aircraft based in Ohio but detected flying over other parts of the U.S.

Each of those pilots – working with able and equally heroic crew mates - saved the lives of passengers aboard seriously damaged airliners over the last 25 years by doing things that no computer could come close to doing. Significant technical, regulatory, spectrum/bandwidth, artificial intelligence, financial and insurance barriers are but the most obvious challenges. Self-driving cars also face avery significant – and high cost – barrier that autonomous planes do not face: infrastructure. Robert Pearson and co-pilot Maurice Quintal saved the lives of all 69 people onboard Air Canada Flight 143, a Boeing 767, after the plane ran out of fuel at 41,000 feet (due to a ground crew error on July 23, 1983.

Airline passengers will give up leg room, overhead-bin space, and a healthy amount of dignity in exchange for a lower airfare. But many won't give up human pilots. "Technically speaking, remotely controlled planes carrying passengers and cargo could appear" by around 2025, the investment bank UBS said a report released Monday (Aug. A switch to full automation could save the air-transportation industry $35 billion a year and cut passenger fares by around 10%. Fully automated commercial flight won't likely take off until the 2040s, it says.