Science and IT news

DeepMind’s new computer can learn from its own memory

DeepMind, an artificial intelligence firm that was acquired by Google in 2014 and is now under the Alphabet umbrella, has developed a computer than can refer to its own memory to learn facts and use that knowledge to answer questions. That’s huge, because it means that future AI could respond to queries from humans without being taught every possible correct answer. DeepMind says its new AI model, called a differentiable neural computer (DNC), can be fed with things like a family tree and a map of the London Underground network, and can answer complex questions about the relationships between items in those data structures. For example, you could get responses to questions like, “Starting at Bond street, and taking the Central line in a direction one stop, the Circle line in a direction for four stops, and the Jubilee line in a direction for two stops, at what stop do you wind up?” DeepMind says its DNC could also help you plan an efficient route from Moorgate to Piccadilly Circus. Similarly, it could understand and answer questions about the relationships between people from a large family, like, ““Who is Freya’s maternal great uncle?” You can see a visualization of this below: {source} <iframe width="853" height="480" src="https://www.youtube.com/embed/B9U8sI7TcMY?rel=0" frameborder="0" allowfullscreen></iframe> {/source} This discovery builds on the concept of neural networks, which mimic the way the human mind works. They are great for machine learning applications where you want a computer to learn to do things by recognizing patterns. It’s these networks that helped DeepMind’s AlphaGo AI defeat world champions at the complex game of Go. But AlphaGo had to be trained by feeding it data about 30 million moves from historical games. By augmenting an AI’s capabilities with the power of learning from memory, it’ll likely be able to complete far more complex tasks on its own. DeepMind hopes that its DNC, which it describes as “a learning machine that, without prior programming, can organise information into connected facts and use those facts to solve problems,” will allow for further breakthroughs in computing. thenextweb.com

Strange signals from 234 stars could be ET – or human error

It’s a bold claim. Two astronomers think they have spotted messages from not just one extraterrestrial civilisation, but 234 of them. The news has sparked a lively debate in the field as other astronomers think the claim is premature and are working fast to get to the bottom of the signals.   In 2012, Ermanno Borra at Laval University in Quebec suggested that an extraterrestrial civilisation might use a laser as a means of interstellar communication. If the little green men simply flashed a laser toward the Earth like a strobe light, we would see periodic bursts of light hidden in the spectrum of their host star. They would be incredibly faint and rapid, but a mathematical analysis could uncover them.   “The kind of energy needed to generate this signal is not crazy,” says Borra. In fact, Borra showed that technology we have on Earth today – specifically the Helios laser at the Lawrence Livermore National Laboratory – could generate that kind of signal, should we want to reveal ourselves to the cosmos.   With this in mind, Borra’s graduate student Eric Trottier combed through 2.5 million stars recorded by the Sloan Digital Sky Survey in search of such a signal. He found it, down to the exact shape, in 234 stars.   The overwhelming majority of those stars are in the same spectral class as the sun, which Borra says supports his hypothesis that this signature must be the result of extraterrestrial intelligent life. And with the data in hand, he thinks that 234 distinct civilisations are beaming pulses of the same periodicity (roughly 1.65 picoseconds) toward the Earth.   Borra and Trottier ruled out other possible explanations for the pattern, like rapid pulsations in the atmospheres of the stars themselves and rotational transitions in molecules. “We have to follow a scientific approach, not an emotional one,” says Borra. “But intuitively – my emotion speaks now – I strongly suspect that it’s an ETI signal.”   Extraordinary claims   Other astronomers think that Borra’s intuition might have run away with him.   “They don’t consider every natural possibility and jump prematurely to the supernatural – so to speak – conclusion,” says Peter Plavchan at Missouri State University in Springfield. “I think it’s way too premature to do that.”   “There is perhaps no bolder claim that one could make in observational astrophysics than the discovery of intelligent life beyond the Earth,” says Andrew Siemion, the director of the SETI Research Centre at the University of California Berkeley. “It’s an incredibly profound subject—and of course that’s why many of us devote our lives to the field and put so much energy into trying to answer these questions. But you can’t make such definitive statements about detections unless you’ve exhausted every possible means of follow-up.”   So that’s exactly what the Breakthrough Listen Initiative—a project headed by Siemion that searches for signs of intelligent life beyond Earth—will do. The team plans to observe several stars from Borra’s sample with the 2.4-meter Automated Planet Finder telescope at the Lick Observatory in California.   Borra is excited to see that others are taking the reins. “At this stage, the signal is so strange, that although our detailed analysis seems to indicate that it is a real signal, it has to be validated with more work,” he says.   Still, the Breakthrough Listen team doesn’t share Borra’s enthusiasm. According to a statement, they have rated the detection as a zero to 1 on the Rio Scale for SETI observations, meaning that it is insignificant.   In fact, Siemion thinks the spectral patterns were likely caused by errors in calibration or data analysis. And Plavchan agrees. He points to several steps in the team’s data analysis that “scared him” because they didn’t consider how those steps might affect their results—a red flag in any scientific claim. At the end of the day, the signal probably comes down to a human error, he says.   “It’s not a bad idea to look for a signal, it’s just that they didn’t do their homework,” says Plavchan. newscientist.com  

Terahertz radiation could speed up computer memory by 1000 times

One area limiting speed in personal computing speed is memory -- specifically, how quickly individual memory cells can be switched, which is currently done using an external magnetic field. European and Russian scientists have proposed a new method using much more rapid terahertz radiation, aka "T-rays," the same things used in airport body scanners. According to their research, published in the journal Nature, swapping out magnetic fields for T-rays could crank up the rate of the cell-resetting process by a factor of 1000, which could be used to create ultrafast memory.   The radiation is actually a series of short electromagnetic pulses pinging the cells at terahertz frequencies (which have wavelengths of about 0.1 millimeter, lying between microwaves and infrared light, according to the scientists' press release). Most of the recent T-ray experiments have dealt with quick, precise inspections of organic and mechanical material. Aside from quickly scanning you for contraband and awkward bulges at airports, other proposals have involved using terahertz radiation to look into broken microchip innards, peer into fragile texts and even comb airport luggage for bombs.   But similar to those hypothetical applications, you won't see T-rays in your PCs any time soon. The scientists have successfully demonstrated the concept on a weak ferromagnet, thulium orthoferrite (TmFeO₃), and even found that the terahertz radiation's effect was ten times greater than a traditional external magnetic field, meaning the new method is both far faster and more efficient. But the scientists have yet to publish tests on actual computer memory cells, so it's unknown when, or if, T-rays will buzz around inside your machine.   engadget.com  

The science world is freaking out over this 25-year-old's answer to antibiotic resistance

A 25-year-old student has just come up with a way to fight drug-resistant superbugs without antibiotics. The new approach has so far only been tested in the lab and on mice, but it could offer a potential solution to antibiotic resistance, which is now getting so bad that the United Nations recently declared it a "fundamental threat" to global health. Antibiotic-resistant bacteria already kill around 700,000 people each year, but a recent study suggests that number could rise to around 10 million by 2050. In addition to common hospital superbug, methicillin-resistant Staphylococcus aureus (MRSA), scientists are now also concerned that gonorrhoea is about to become resistant to all remaining drugs. But Shu Lam, a 25-year-old PhD student at the University of Melbourne in Australia, has developed a star-shaped polymer that can kill six different superbug strains without antibiotics, simply by ripping apart their cell walls. "We’ve discovered that [the polymers] actually target the bacteria and kill it in multiple ways," Lam told Nicola Smith from The Telegraph. "One method is by physically disrupting or breaking apart the cell wall of the bacteria. This creates a lot of stress on the bacteria and causes it to start killing itself." The research has been published in Nature Microbiology, and according to Smith, it's already being hailed by scientists in the field as "a breakthrough that could change the face of modern medicine". Before we get too carried away, it's still very early days. So far, Lam has only tested her star-shaped polymers on six strains of drug-resistant bacteria in the lab, and on one superbug in live mice. But in all experiments, they've been able to kill their targeted bacteria - and generation after generation don't seem to develop resistance to the polymers. The polymers - which they call SNAPPs, or structurally nanoengineered antimicrobial peptide polymers - work by directly attacking, penetrating, and then destabilising the cell membrane of bacteria. Unlike antibiotics, which 'poison' bacteria, and can also affect healthy cells in the area, the SNAPPs that Lam has designed are so large that they don't seem to affect healthy cells at all. "With this polymerised peptide we are talking the difference in scale between a mouse and an elephant," Lam's supervisor, Greg Qiao, told Marcus Strom from the Sydney Morning Herald. "The large peptide molecules can't enter the [healthy] cells." You can see the SNAPPs (green) surrounding and ripping apart bacterial cells below: While the results are positive so far, it's too early to get excited about what this could mean for humans, says Cyrille Boyer from the University of New South Wales in Australia, who wasn't involved in the research. "The main advantage seems to be they can kill bacteria more effectively and selectively [than other peptides]" Boyer told Strom, before adding that the team is a long way off clinical applications. But what's awesome about the new project is that, while other teams are looking for new antibiotics, Lam has found a completely different approach. And it could make all the different in the coming 'post-antibiotic world'. That's what she's hoping, anyway. "For a time, I had to come in at 4am in the morning to look after my mice and my cells," she told The Telegraph. "I wanted to be involved in some kind of research that would help solve problems ... I really hope that the polymers we are trying to develop here could eventually be a solution."