Summary of “Brain-Computer Interfaces Show That Neural Networks Learn by Recycling”

The hallmark of intelligence is the ability to learn.
The brain may be highly flexible and adaptive overall, but at least over short time frames, it learns by inefficiently recycling tricks from its neural repertoire rather than rewiring from scratch.
Now, while observing activity in the brain during learning, Yu and his colleagues have seen evidence of a similar lack of plasticity at the neural level.
In 2014, the researchers observed that test subjects could learn new tasks more easily if they involved patterns of neural activity within the intrinsic manifold rather than outside it.
Then the team switched the neural activity requirements for moving the cursor and waited to see what new patterns of neural activity, corresponding to new points in the intrinsic manifold, the animals would use to accomplish them.
Why would the brain use less than the best strategy for learning? The group’s findings suggest that, just as the neural architecture constrains activity to the intrinsic manifold, some further constraint limits how the neurons reorganize their activity during the experiments.
Chase likened the motor cortex to an old-fashioned telephone switchboard, with neural connections like cables linking inputs from other cortical areas to outputs in the brain’s cerebellum.
The researchers can’t yet rule out the possibility that reassociation is a fast interim way for the brain to learn new tasks; over a longer time period, realignment or rescaling might still show up.

The orginal article.

Summary of “Why Do Dogs Love Us? Science Explains”

You’re not just imagining it: There’s substantial research to support the claim that dogs truly adore their owners.
While we don’t know exactly how long ago humans started domesticating dogs, some scientists think our friendship could go as far back as 40,000 years.
“Of course dogs love their people!” animal behavior consultant Amy Shojai tells Inverse.
By utilizing functional magnetic resonance imaging scans – which measure brain nerve cell levels – the researchers got an inside look at how dogs responded to their humans’ scent versus familiar dogs, unfamiliar dogs, and unfamiliar people.
As Inverse previously reported, researchers at the University of York recently found that dogs respond more positively to dog-directed speech than when we talk to them like people.
Scientists had 37 dogs listen to people talking to them in “Dog-speak” – that high-pitched voice, coupled with “Dog-relevant” phrases.
Participants would then talk to dogs in a flat done about ordinary things.
The dogs overwhelmingly preferred dog-speak, which the researchers compared to the way people talk to babies.

The orginal article.

Summary of “The dark truth about chocolate”

The 19th century saw chocolate drinking become cheap enough to spread beyond the wealthy, the invention of solid chocolate and the development of milk chocolate.
The packets don’t say so, but the message we’re supposed to swallow is clear: this new, improved chocolate, especially if it is dark, is good for your health.
Studies published last year found chocolate consumers to be at reduced risk of heart flutters, and that women who eat chocolate are less likely to suffer from strokes.
Someone would need to consume about 12 standard 100g bars of dark chocolate or about 50 of milk chocolate per day to get that much.
The European Food Safety Authority has approved one rather modest chocolate-related health claim – that some specially processed dark chocolate, cocoa extracts and drinks containing 200mg of flavanols “Contribute to normal blood circulation” by helping to maintain blood vessel elasticity.
Then there’s the problem that, unlike in drug trials, those taking part in chocolate studies often know whether they are being given chocolate or a placebo.
“Efforts by many of the large chocolate companies to demonstrate health effects started side by side with the outcry over the use of child labour and slavery,” says Michael Coe, a retired anthropologist formerly of Yale University, co-author of The True History of Chocolate.
Research was making it increasingly clear that health benefits claims for commercial dark chocolate products were unrealistic because of their low flavanol content.

The orginal article.

Summary of “Hair dye: now with graphene to take away the frizz”

In a quest to make a less toxic hair dye, scientists created a dye using graphene.
Most permanent hair dyes use harsh chemicals to open up the outside layer of the hair so that other chemicals can get inside and change its color, Chemical & Engineering News reports.
A team of researchers at Northwestern University decided to use a different strategy: Instead of opening up the hair, why not just coat it with tiny colored particles made out of graphene?
The dye is made of chemically modified graphene particles, a sugar from the pulverized shells of crustaceans, and vitamin C. When sprayed and brushed on hair, the dye sticks in a couple of different ways: graphene clings to uneven surfaces, plus it sticks to the crustacean sugar in the dye, which in turn binds to a protein in hair, according to C&E News.
Charged hair strands can “Stand up on end, creating an uncomfortable ‘flyaway’ effect,” the study says.
Graphene conducts electricity, which means that hair coated in it doesn’t build up a charge.
To show this, researchers rubbed different hair samples with a rubber film, to create an effect similar to rubbing a balloon over your hair.
While natural hair and hair dyed with a conventional dye stood up, graphene-dyed hair stayed sleek and smooth.

The orginal article.

Summary of “Uber, Lyft Drivers Earning A Median Profit Of $3.37 Per Hour, Study Says”

Uber, Lyft Drivers Earning A Median Profit Of $3.37 Per Hour, Study Says : The Two-Way Researchers at MIT said 30 percent of Uber and Lyft drivers are actually losing money after taking car expenses into account, while most drivers earn less than minimum wage.
The vast majority of Uber and Lyft drivers are earning less than minimum wage and almost a third of them are actually losing money by driving, according to researchers at the Massachusetts Institute of Technology.
A working paper by Stephen M. Zoepf, Stella Chen, Paa Adu and Gonzalo Pozo at MIT’s Center for Energy and Environmental Policy Research says the median pretax profit earned from driving is $3.37 per hour after taking expenses into account.
Drivers earning the median amount of revenue are getting $0.59 per mile driven, researchers say, but expenses work out to $0.30 per mile, meaning a driver makes a median profit of $0.29 for each mile.
“If drivers are fully able to capitalize on these losses for tax purposes, 73.5% of an estimated U.S. market $4.8B in annual ride-hailing driver profit is untaxed,” they add.
According to MIT researchers, 80 percent of drivers said they work less than 40 hours per week.
Recode listed the initiatives Uber rolled out in 2017 in order to appeal to drivers, including 24-hour phone support, paid wait time and paying drivers if customers cancel after a certain amount of time.
Both Uber and Lyft have been fighting legal battles for years against initiatives to classify their drivers as “Employees” instead of “Independent contractors” – meaning drivers don’t receive benefits like health care or sick leave.

The orginal article.

Summary of “Where Millennials end and post-Millennials begin”

As we enter 2018, it’s become clear to us that it’s time to determine a cutoff point between Millennials and the next generation.
In order to keep the Millennial generation analytically meaningful, and to begin looking at what might be unique about the next cohort, Pew Research Center will use 1996 as the last birth year for Millennials for our future work.
At 16 years, our working definition of Millennials will be equivalent in age span to their preceding generation, Generation X. By this definition, both are shorter than the span of the Baby Boomers – the only generation officially designated by the U.S. Census Bureau, based on the famous surge in post-WWII births in 1946 and a significant decline in birthrates after 1964.
For analytical purposes, we believe 1996 is a meaningful cutoff between Millennials and post-Millennials for a number of reasons, including key political, economic and social factors that define the Millennial generation’s formative years.
Generation X grew up as the computer revolution was taking hold, and Millennials came of age during the internet explosion.
Pew Research Center is not the first to draw an analytical line between Millennials and the generation to follow them, and many have offered well-reasoned arguments for drawing that line a few years earlier or later than where we have.
As has been the case in the past, this means that the differences within generations can be just as great as the differences across generations, and the youngest and oldest within a commonly defined cohort may feel more in common with bordering generations than the one to which they are assigned.
In the coming weeks, we will be updating demographic analyses that compare Millennials to previous generations at the same stage in their life cycle to see if the demographic, economic and household dynamics of Millennials continue to stand apart from their predecessors.

The orginal article.

Summary of “Does money buy happiness? Yes, up to a point.”

The technical term for this cutoff is the income “Satiation point.”
The researchers analyzed the relationship between this score and household income.
The incomes are converted to US dollars and adjusted for variations in spending power across countries.
These psychologists, from Purdue University and the University of Virginia, are not the first to study how income relates to life satisfaction.
Dan Sacks is an economist at Indiana University who studies the relationship between income and subjective well-being.
The surveys rely on self-reported income, and previous research shows that just because people say they make a certain amount of money, it doesn’t mean they actually do.
“It could be true that on average, people who say they have income of $150,000 are no happier than people who say they have income of $100,000,” writes Sacks.
“But I’m not convinced that people who actually have income of $150,000 are no happier than people who have income of $100,000.” Also, it’s possible that rich people have a tendency to underemphasize their happiness compared with poorer people.

The orginal article.

Summary of “How Our Beliefs Can Shape Our Waistlines”

A recent epidemiological study suggests that our beliefs about how much we exercise may substantially influence our health and longevity, even if those beliefs are objectively inaccurate – which hints that upending our thinking about exercise might help us whittle away pounds, whether we work out more or not.
Crum and her co-author studied 84 female hotel-room attendants, who told the researchers that they felt they completed little or no daily exercise, although their work consisted mostly of physical labor.
Crum and her colleague explained to half of them that they were meeting or exceeding national recommendations for 30 minutes of daily exercise; a month later, when the researchers checked back, the women said they believed they were getting more exercise than before.
For the new study, Crum and a different co-author, Octavia Zahrt, turned to two federal databases, the National Health Interview Survey and the National Health and Nutrition Examination Survey, which contain health data about representative samples of Americans.
The scientists homed in on information from 61,141 participants who answered questions about whether they felt they were getting more, less or about the same amount of exercise as most people their age.
Risk of early death was up to 71 percent higher than for the group that, correctly or not, felt confident that they exercised more than their peers.
This type of study cannot prove that exercise beliefs directly cause life spans to shorten or grow; it can show only that the two issues are related.
Self-comparisons might also dampen exercise motivation, leading to declining health.

The orginal article.

Summary of “The GANfather: The man who’s given machines the gift of imagination”

The goal of GANs is to give machines something akin to an imagination.
When future historians of technology look back, they’re likely to see GANs as a big step toward creating machines with a human-like consciousness.
Yann LeCun, Facebook’s chief AI scientist, has called GANs “The coolest idea in deep learning in the last 20 years.” Another AI luminary, Andrew Ng, the former chief scientist of China’s Baidu, says GANs represent “a significant and fundamental advance” that’s inspired a growing global community of researchers.
In one widely publicized example last year, researchers at Nvidia, a chip company heavily invested in AI, trained a GAN to generate pictures of imaginary celebrities by studying real ones.
Once it’s been trained on a lot of dog photos, a GAN can generate a convincing fake image of a dog that has, say, a different pattern of spots; but it can’t conceive of an entirely new animal.
Researchers at Yale University and Lawrence Berkeley National Laboratory have developed a GAN that, after training on existing simulation data, learns to generate pretty accurate predictions of how a particular particle will behave, and does it much faster.
Hany Farid, who studies digital forensics at Dartmouth College, is working on better ways to spot fake videos, such as detecting slight changes in the color of faces caused by inhaling and exhaling that GANs find hard to mimic precisely.
Researchers are already highlighting the risk of “Black box” attacks, in which GANs are used to figure out the machine-learning models with which plenty of security programs spot malware.

The orginal article.

Summary of “Why Self-Taught Artificial Intelligence Has Trouble With the Real World”

In 2016, Google DeepMind’s AlphaGo thrashed champion Lee Sedol at the ancient board game Go after poring over millions of positions from tens of thousands of human games.
The past year also saw otherworldly self-taught bots emerge in settings as diverse as no-limit poker and Dota 2, a hugely popular multiplayer online video game in which fantasy-themed heroes battle for control of an alien world.
One characteristic shared by many games, chess and Go included, is that players can see all the pieces on both sides at all times.
An even more daunting game involving imperfect information is StarCraft II, another multiplayer online video game with a vast following.
Before the release of AlphaGo and its progeny, the DeepMind team achieved its first big, headline-grabbing result in 2013, when they used reinforcement learning to make a bot that learned to play seven Atari 2600 games, three of them at an expert level.
Within the larger category of reinforcement learning, board games and multiplayer games allow for an even more specific approach.
In game after game, an algorithm in a self-play system faces an equally matched foe.
Since 2008, hundreds of thousands of human players have attempted Foldit, an online game where users are scored on the stability and feasibility of the protein structures they fold.

The orginal article.