Profile cover photo
Profile photo
Ward Plunet
60,162 followers
60,162 followers
About
Posts

Post has attachment
People assigned to take cold showers didn’t feel any healthier, but they missed less work

There’s something impressive about taking cold showers. It summons up images of military discipline, and people claim that it makes you healthier. But how much of that is true? In my quest to find out, I came across a 2016 paper published in PLoS One on how cold showers affected health and work. The researchers recruited about 3,000 participants who, like most of us, were not used to taking cold showers. These participants were assigned to four groups. Every day for 30 consecutive days, the first group had to end their normal, blessedly warm shower with a cold shower for 30 seconds. The second group had to do this for 60 seconds, the third for 90 seconds, and the fourth was the control group that did not suffer for science. The participants had to time themselves, too, so if they couldn’t complete the full 30–90 seconds, they had to record how long they had subjected themselves to the water. (Keep in mind that this happened in the Netherlands, in January.) After the 30 days, the participants spent another month showering however they wanted. During this time, volunteers were asked to take a wide variety of surveys. At the end of the period, the researchers discovered an interesting finding: the people who had taken the cold showers took 29 percent fewer sick days than the others. But, crucially, they didn’t actually feel ill (as self-reported) any less often.
Add a comment...

Post has attachment
A Gentle Intro to Transfer Learning

Nowadays most applications of Deep Learning rely on Transfer Learning. This is especially true in the domain of Computer Vision. We will explore what Transfer Learning is, how to do it and what the potential pitfalls are. To do this, we’ll go on a little startup quest....The idea is to take the knowledge learned in a model and apply it to another task. Transfer learning sounds like what we want to do. We decide to reuse the already trained Pet-cognizer®. Scenario 1: New dataset is similar to initial dataset: Our new breed dataset is close to the ImageNet dataset we first trained on. It’s similar in the sense that they both contain pictures of the “real world” (as opposed to images of documents or medical scans). Thus the filters in the CNN can be reused, and we don’t have to learn them again. For Breed-cognizer®, we swap the only the last layer. Instead of telling us what animal is in the picture, it will give us probabilities for each dog breed . Because the weights on the last layer are initially random we have to train it. But training these random weights might also change the great filters in the earlier layers. To avoid this, we freeze all layers but the last. Freezing means that the layer weights won’t be updated during training. Rather miraculously, this works to an extent. We are able to achieve 80% accuracy on over 120 classes, just by training the last layer.
Add a comment...

Post has attachment
Dog ownership linked to lower mortality

A team of Swedish scientists have used national registries of more than 3.4 million Swedes aged 40 to 80 to study the association between dog ownership and cardiovascular health. Their study shows that dog owners had a lower risk of death due to cardiovascular disease or to other causes during the 12-year follow-up. A total of more than 3.4 million individuals without any prior cardiovascular disease in 2001 were included in the researchers' study linking together seven different national data sources, including two dog ownership registers. The results are being published for the first time in Scientific Reports. The goal was to determine whether dog owners had a different risk of cardiovascular disease and death than non-dog owners. "A very interesting finding in our study was that dog ownership was especially prominent as a protective factor in persons living alone, which is a group reported previously to be at higher risk of cardiovascular disease and death than those living in a multi-person household. Perhaps a dog may stand in as an important family member in the single households. The results showed that single dog owners had a 33 percent reduction in risk of death and 11 percent reduction in risk of cardiovascular disease during follow-up compared to single non-owners. Another interesting finding was that owners to dogs from breed groups originally bred for hunting were most protected," says Mwenya Mubanga, lead junior author of the study and PhD student at the Department of Medical Sciences and the Science for Life Laboratory, Uppsala University.
Add a comment...

Post has attachment
Giving agriculture a global do-over could feed nearly a billion more people

Globally optimizing which crops we grow where cuts water use and feeds more people. As the human population grows and the human middle class grows in developing countries, we are going to need more. More food, more meat, more energy. And producing more is going to require more resources. Since we are just about tapped out of the resources required for food production—namely water and land—we are going to have to figure out how to use these limited resources as efficiently as possible. A number of suggestions have been made to try to achieve this, from the lower tech—like curbing animal consumption and minimizing food waste—to the higher tech, like planting GMOS that might improve yields, developing better fertilizers, and maximizing irrigation efficiency. A new analysis in Nature Geoscience offers up one more: switching what we grow where. The authors write: “We find that the current distribution of crops around the world neither attains maximum production nor minimum water use.” This is hardly surprising, since agriculture developed in a haphazard, piecemeal way, pushed by different political entities with different agendas over centuries. No one ever sat down with the whole globe before to determine what would grow best in each region. But these scientists claim that shuffling around crops based on how much rain and irrigation water they need could feed 825 million more people than the planet can currently sustain, while cutting water use by about 13 percent. Their primary goal was to optimize food production (in terms of calories grown) while saving water. But they also attempted to render rain-fed crops more resilient to droughts, to leverage local knowledge and technologies, and to maintain crop diversity.
Add a comment...

Post has attachment
Artist Builds VR Game In Two Weeks Using Google Blocks And Unreal

Artist Jarlan Perez recently worked with Google to test out a VR-first creation process. Perez built a fully interactive game in two weeks that started in intuitive VR creation tool Blocks, with another week spent polishing and squashing bugs. You can now download the short puzzler, called Blocks Isle, and play it in an HTC Vive. I’ve been following closely this type of process because VR tools use intuitive or natural behavior to unlock creativity. As the tools improve, this approach could dramatically lower the barrier to entry for making compelling interactive software. That’s essentially what Google is exploring through tests like the one with Perez. Earlier this year the game Paulo’s Wing was built using Google’s other VR art app, Tilt Brush. Blocks was released more recently and makes it easy to create simple solid objects that can be stacked up over time like Lego to make eye-catching scenes.
Add a comment...

Post has attachment
Introduction to AI for Video Games

Welcome to my new reinforcement learning course! For the next 10 weeks we're going to go from the basics to the state of the art in this popular subfield of machine learning using video game environments as our testbed. RL is a huge reason DeepMind and OpenAI have been so successful thus far in creating world changing AI bots. Make sure to subscribe so you'll get updated with every new video I release. And don't worry if you don't understand policy iteration or value iteration just yet, I merely wanted to introduce these phrases in this video, next week i'm going to really dive into what these 2 methods look like programmatically.
Add a comment...

Post has attachment
Building Dota Bots That Beat Pros - OpenAI's Greg Brockman, Szymon Sidor, and Sam Altman
Add a comment...

Post has attachment
Scientists claim to diagnose football-related brain injury in living patients for first time

For the first time, scientists have confirmed a diagnosis of chronic traumatic encephalopathy (CTE) — a neurological disease linked to head injuries from sports like football — in a living person. Until now, we’ve only been able to diagnose CTE in dead patients. Finding the disease while the patient is still alive could help scientists find a way to treat it. CTE develops from repeated hits to the head and has been linked to severe memory loss, depression, and dementia. It’s been found in 99 percent of the donated brains of NFL players. In a study published in the journal Neurosurgery, researchers found a telltale sign of CTE, a specific protein, in the brains of 14 retired NFL players who underwent a brain scan. Now that one of the players has died and doctors have been able to take a closer look at his brain, they have confirmed the CTE diagnosis. Many former National Football League players like Aaron Hernandez and Junior Seau have been found to have the progressive brain disease. Last year, the NFL reached a billion-dollar settlement, the largest in sports history, over a lawsuit from former players who suffered concussions and now have severe neurological diseases like amyotrophic lateral sclerosis (ALS, or Lou Gehrig’s disease). The subject of the paper was former NFL player Fred McNeill, according to CNN. McNeill played 12 seasons in the National Football League for the Minnesota Vikings, reporting one concussion. But by the time he was 59, McNeill was already experiencing serious problems with his motor skills, and so he sat for the brain scan. Scientists found increased levels of certain proteins in his brain, including the tau protein, which has been linked to Alzheimer’s. After McNeill died two years ago, at 63, detailed brain-tissue analysis confirmed that his brain had other physical signs of CTE, suggesting that the tau protein is linked to the disease. A study from September suggested that the levels of another protein, called CCL11, may also be used to diagnose CTE in living patients.
Add a comment...

Post has attachment
MEET THE TESLA SEMITRUCK, ELON MUSK'S MOST ELECTRIFYING GAMBLE YET

Musk believes that going after the big boys is the best way to have a real impact on climate change. In the five years since Tesla started producing its Model S sedan, it has sold about 200,000 cars. The US has more than 250 million passenger cars on the road, making the impact of this, roughly, zero. Even if Tesla scales up production of its “affordable” Model 3 sedan, it will still be a very long time before the Silicon Valley automaker can change the way humanity moves about enough for any dip in emissions to register as more than a blip. Trucks offer a more effective way to do that, because they are particularly toxic. “Heavy-duty vehicles make up a small fraction of the vehicles on the road, but a large fraction of their emissions,” says Jimmy O’Dea, who studies clean vehicles at the Union of Concerned Scientists. In California, that category (which includes buses as well as trucks) accounts for 7 percent of total vehicles, but produces 20 percent of transportation-related greenhouse gas emissions and a third of all NOx emissions (those are the ones linked to asthma attacks and respiratory illnesses). Every truck you move with electricity instead of diesel has an outsize effect on the health of the planet and everything living on it. 18-wheelers are the ultimate force multiplier. Musk has done the math. And while lots of players are moving into electric trucking space, none have the star power of Tesla, the kind of clout that makes the whole country pay attention.
Add a comment...

Post has attachment
Study finds consuming nuts strengthens brainwave function

A new study by researchers at Loma Linda University Health has found that eating nuts on a regular basis strengthens brainwave frequencies associated with cognition, healing, learning, memory and other key brain functions. An abstract of the study—which was presented in the nutrition section of the Experimental Biology 2017 meetings in San Diego, California, and published in the FASEB Journal. In the study titled "Nuts and brain: Effects of eating nuts on changing electroencephalograph brainwaves," researchers found that some nuts stimulated some brain frequencies more than others. Pistachios, for instance, produced the greatest gamma wave response, which is critical for enhancing cognitive processing, information retention, learning, perception and rapid eye movement during sleep. Peanuts, which are actually legumes, but were still part of the study, produced the highest delta response, which is associated with healthy immunity, natural healing, and deep sleep.
Add a comment...
Wait while more posts are being loaded