The dominant language models (LMs) such as n-gram and neural network (NN) models represent sentence probabilities in terms of conditionals. In contrast, a new trans-dimensional random field (TRF) LM has been recently introduced to show superior performances, where the whole sentence is modeled as a random field. In this paper, we examine how the TRF models can be interpolated with the NN models, and obtain 12.1\% and 17.9\% relative error rate reductions over 6-gram LMs for English and Chinese speech recognition respectively through log-linear combination.

"Seven hundred years after glasses were invented there are still 2.5 billion people in the world with poor vision and no access to vision correction," says Hong Kong philanthropist James Chen. Chairman of his family's Nigeria-based manufacturing company, Wahum Group, Chen is funding a contest called the Clearly Vision Prize that will award a total of 250,000 to projects that improve eyesight, especially in poor countries. Thirty-six semifinalists were announced this week (the five winners will be awarded September 15). Among the contenders: 3D printed eyeglass frames, drones that deliver medical supplies, and several smartphone-based technologies. Some of the smartphones help nonexperts test vision, and one uses artificial intelligence to "see" for blind people. The Clearly Vision semifinalists represent just a sampling of the smartphone projects fighting vision loss, a growing field that is bringing critical care to remote regions far from hospitals and doctors offices.

The newest way to accurately predict poverty comes from satellite images and machine learning. This imaging technique could make it easier for aid organizations to know where and how to spend their money; it may also help governments develop better policy. We already know that the more lit up an area is at night, the richer and more developed it is. Researchers use this method to estimate poverty in places where we don't have exact data. But "night light" estimates are rough and don't tell us much about the wealth differences of the very poor.

Later this month, Uber will offer the world's first ride-hailing service in Pittsburgh, using a test fleet of 100 Volvo XC90 SUVs. I admit that I didn't see it coming this fast. Only yesterday, I wrote about a new trial of self-driving minibuses in Helsinki, and it seemed pretty darn forward-looking at the time. But those vehicles are really slow, they ply the same little route repeatedly, and they stop at every stop. Today Uber is talking about driving to points specified by whoever jumps into the rear seat.

Sleep is a naturally recurring state of mind characterized by altered consciousness, relatively inhibited sensory activity, inhibition of nearly all voluntary muscles, and reduced interactions with surroundings.[1] It is distinguished from wakefulness by a decreased ability to react to stimuli, but is more easily reversed than the state of hibernation or of being comatose. Mammalian sleep occurs in repeating periods, in which the body alternates between two highly distinct modes known as non-REM and REM sleep. REM stands for "rapid eye movement" but involves many other aspects including virtual paralysis of the body. During sleep, most systems in an animal are in an anabolic state, building up the immune, nervous, skeletal, and muscular systems. Sleep in non-human animals is observed in mammals, birds, reptiles, amphibians, and some fish, and, in some form, in insects and even in simpler animals such as nematodes. The internal circadian clock promotes sleep daily at night in diurnal species (such as humans) and in the day in nocturnal organisms (such as rodents). However, sleep patterns vary widely among animals and among different individual humans. Industrialization and artificial light have substantially altered human sleep habits in the last 100 years. The diverse purposes and mechanisms of sleep are the subject of substantial ongoing research.[2] Sleep seems to assist animals with improvements in the body and mind. A well-known feature of sleep in humans is the dream, an experience typically recounted in narrative form, which resembles waking life while in progress, but which usually can later be distinguished as fantasy. Sleep is sometimes confused with unconsciousness, but is quite different in terms of thought process. Humans may suffer from a number of sleep disorders. These include dyssomnias (such as insomnia, hypersomnia, and sleep apnea), parasomnias (such as sleepwalking and REM behavior disorder), bruxism, and the circadian rhythm sleep disorders. In mammals and birds, sleep is divided into two broad types: rapid eye movement (REM sleep) and non-rapid eye movement (NREM or non-REM sleep). Each type has a distinct set of physiological and neurological features associated with it. REM sleep is associated with dreaming, desynchronized and faster brain waves, loss of muscle tone,[3] and suspension of homeostasis[citation needed]. REM and non-REM sleep are so different that physiologists classify them as distinct behavioral states. In this view, REM, non-REM, and waking represent the three major modes of consciousness, neural activity, and physiological regulation.[4] According to the Hobson & McCarley activation-synthesis hypothesis, proposed in 1975–1977, the alternation between REM and non-REM can be explained in terms of cycling, reciprocally influential neurotransmitter systems.[5]

In the year 1820, a person could expect to live less than 35 years, 94% of the global population lived in extreme poverty, and less that 20% of the population was literate. Today, human life expectancy is over 70 years, less that 10% of the global population lives in extreme poverty, and over 80% of people are literate. These improvements are due mainly to advances in technology, beginning in the industrial age and continuing today in the information age. There are many exciting new technologies that will continue to transform the world and improve human welfare. Here are eleven of them.

Users arriving at, say, an airport would book a seat on a so-called zenHop "CityAirbus" drone, then proceed to a "zenHub" helipad, according to the concept. They'd be flown to their destination for about the same cost as a taxi, since the ride would be shared by several passengers. Luggage would be delivered by another service (zenLuggage, of course), and the whole thing would be safeguarded from hackers by (wait for it) zenCyber. The company said that the CityAirbus multi-rotor, electric aircraft design has been "kept under wraps," though it did supply an artist's impression (above). The Airbus Helicopter subsidiary has been working on the drone-like design for two years, and it "could soon become reality without having to wait for too many regulatory changes," according to the press release. Airbus is also working on a drone delivery service (below) and plans to start testing it at a Singapore university by mid-2017.

FOR frazzled teachers struggling to decide what to watch on an evening off, help is at hand. An online streaming service's software predicts what they might enjoy, based on the past choices of similar people. When those same teachers try to work out which children are most at risk of dropping out of school, they get no such aid. But, as Sendhil Mullainathan of Harvard University notes, these types of problem are alike. They require predictions based, implicitly or explicitly, on lots of data.

We've talked before about the prospects of artificial intelligence (AI) and how it will likely disrupt things like we've never seen before with some estimates predicting that up to 80% of all service jobs will be impacted. Healthcare is one area where AI is receiving a good chunk of funding. We looked before at one example of an artificial intelligence company called Enlitic that uses machine learning technology to read X-rays better than a human radiologist who makes 286,000 a year on average. There are actually quite a few artificial intelligence companies in healthcare and CB Insights recently identified 65 of them at various stages of funding. Founded just last year, Chinese company iCarbonX has taken in nearly 200 million in funding from investors that include the 200 billion Chinese internet giant Tencent.

Scientists in Japan reportedly saved a woman's life by applying artificial intelligence to help them diagnose a rare form of cancer. Faced with a 60-year-old woman whose cancer diagnosis was unresponsive to treatment, they supplied an AI system with huge amounts of clinical cancer case data, and it diagnosed the rare leukemia that had stumped the clinicians in just ten minutes. The Watson AI system from IBM matched the patient's symptoms against 20m clinical oncology studies uploaded by a team headed by Arinobu Tojo at the University of Tokyo's Institute of Medical Science that included symptoms, treatment and response. The Memorial Sloan Kettering Cancer Center in New York has carried out similar work, where teams of clinicians and data analysts trained Watson's machine learning capabilities with oncological data in order to focus its predictive and analytic capabilities on diagnosing cancers. IBM Watson first became famous when it won the US television game show Jeopardy in 2011.