DEA Approves First Trial Of Medical Marijuana For PTSD

In a historic move, the United States Drug Enforcement Agency (DEA) has granted full approval for the first ever U.S. clinical trial to develop the marijuana plant into a prescription medicine to treat PTSD symptoms. The randomized controlled trial—sponsored by the Multidisciplinary Association for Psychedelic Studies (MAPS)—will document the effects of smoked marijuana on 76 war veterans who have chronic, treatment-resistant PTSD. The research aims to dig even deeper by testing four separate strains of marijuana to compare dosing, composition, side effects, and different benefits between plants with varying levels of tetrahydrocannabinol (THC) and cannabidiol (CBD)—two of the major psychoactive components of cannabis.

With the DEA already on board, the study been given the green light from every relevant federal agency, including the Food and Drug Administration (FDA), the National Institute on Drug Abuse (NIDA), and the Public Health Service (PHS). According to MAPS, the study is expected to start this year, once they receive the medical marijuana from the NIDA.

Although there is already significant evidence that marijuana could have a positive effect on PTSD patients, the U.S. Department of Veterans Affairs currently prohibits doctors from prescribing marijuana to patients even in states where it is legal, noting on its website that “controlled studies have not been conducted to evaluate the safety or effectiveness of medical marijuana for PTSD.” This has caused an unknown number of PTSD-affected veterans returning to the U.S. to circumvent the VA system, and also seek out illegal means of purchasing marijuana.

from Popular Science http://ift.tt/1Td2sS7
via IFTTT

Watch Tim Cook and JJ Lin Play A Song on GarageBand

Sometimes it’s nice to take a moment and enjoy the product of your company’s innovations. That’s just what Apple CEO Tim Cook did with GarageBand.

After the release of a new App update, Cook and artist JJ Lin had a bit of a jam session on one of Lin’s songs.

Lin, for his part, is a Singaporean singer/songwriter based in Taiwan. He seems to have taken the more challenging section of the song, but it makes sense that Cook would play backup on someone else’s music.

No word yet on whether Tim Cook’s performance was all for fun, or if he’s putting together a collaborations album that will eventually be downloaded onto all of our iPhones without our permission.

from Popular Science http://ift.tt/1NyvPMz
via IFTTT

How Google’s New A.I. Microchips Take A Page From Bitcoin Miners

Yesterday at Google’s I/O developers conference, CEO Sundar Pichai briefly spoke about a custom-built chip that helps give Google its edge in machine learning and artificial intelligence. The chip, dubbed a TPU or Tensor Processing Unit (in keeping with Google’s A.I. platform TensorFlow), is specifically wrought for running Google’s decision-making algorithms. Most companies like Facebook and Microsoft use GPUs for their machine learning and artificial intelligence.

But Pichai’s speech and the accompanying blog post only reveal a few details about TPUs. About the only useful thing we know about the chip is that it’s an ASIC, or application-specific integrated circuit. ASIC chips aren’t bought off the shelves, but designed specifically to do one task very well without using a lot of power. They’re used in applications that never change, like the controlling how a phone battery charges.

ASIC designers customize the size and arrangement of the chip’s transistors to closely match the kind of computations needed. Norm Jouppi, a hardware engineer at Google who wrote the TPU blog post, says that the chip was designed to have fewer transistors per operation, because machine learning requires less precise computations.

Pichai’s big statistic is performance per watt—that’s where Google is saying the TPU is an “order of magnitude” better

It’s entirely possible that the TPU chips are simply performing at slightly better speeds, but consuming significantly less power. ASIC chips are extremely power efficient, which is one of the main reasons a recent surge of interest in ASIC chips has been in Bitcoin mining. Mining, like machine learning, requires a computer to quickly perform a large amount of computations and has also typically relied on GPUs.

In the Bitcoin world, an ASIC-based mining rig can consume twenty times less power while doing the same work as a GPU-based rig, as broken down on this Bitcoin forum. In this example, six AMD R9 290 graphics cards (which pull an average of 250 watts each, plus computer components), are stacked against 78 DualMiner ASIC cards (which run off USB and pull 2.5 watts each). Both setups are rated to run the same computations (5.4 Mh/s), but the GPU-based rig conservatively pulls more than 1500 watts while the ASIC rig pulls less than 200 watts.

Basically, the takeaway is that statistics can be misleading, and without seeing objective benchmarks on the TPUs, we should be skeptical of these claims.

ASIC chips have been made since around since the 1960s, and also have some noted downsides. For instance, ASIC chips are designed to run certain code a very specific way. If the machine learning architecture at Google changes, the TPUs could be rendered useless. This is unlikely, as Google uses their Tensorflow platform for everything, but still a possibility.

The Wall Street Journal reports that Google has been using TPUs since April 2015 to speed up how fast Street View reads signs, and Pichai mentioned in his speech that they are being used in Google’s Cloud Machine Learning platform.

Google says they’re not giving any more information about the TPU chips at this time.

from Popular Science http://ift.tt/25cAagb
via IFTTT

Saab’s Underwater Drone Will Hunt For Aquatic Explosives

Landmines are a bad day waiting to happen. The classic explosives, used in war for centuries, hide in plain sight, waiting for someone unfortunate enough to set them off. Sea mines are a similar peril, only threatening whole boats at a time. And while there are plenty of robots that can find the bombs on land, there aren’t that many which can do the trick in water, especially close to shore. The Sea WASP stands for “Waterborne Anti-IED Security Platform.” Made by Sweden defense and auto giant Saab, the Sea WASP is an underwater drone made to find Improvised Explosive Devices under the water’s surface.

It takes two people to operate the Sea Wasp. The robot is tethered, with 500 feet of cord to let it explore depths of up to 200 feet. On land, the whole 5.5-foot long machine weighs about 200 pounds. It has forward-looking sonar, several sensors for depth and navigation, and two cameras: a big one on the front of the vehicle, and another one on the grabber arm.

From Saab:

Sea Wasp represents a significant change in underwater operations against IEDs and similar threats by allowing bomb technicians to conduct underwater intervention for both improvised and conventional munitions. Designed to be operated by a small EOD/IEDD team – as few as two persons – the system can easily be configured to meet the specific requirements of any mission. Transportable over land in a light support vehicle, the Sea Wasp can be deployed from harbour walls or the beach. It can also be fitted to bespoke surface support vessels, and is flexible enough to be loaded into multiple varieties of boat, for example vessels of opportunity (VOO), rigid-hulled inflatable boats (RHIBs) and work boats, depending on the demands of the mission.

Once the Sea Wasp has found the device, it will disposes of them similar to how ground robots would. That means human technicians would watch video as the robot handles the explosive, and then determine if it’s harmless, can be dismantled, and/or needs to be blown up far away from people and ships.

Mines are an old weapon of war, and improvised explosive devices are no different. Finding a safe way to identify and neutralize those weapons is important for any modern military, and making sure that bomb disposal doesn’t stop at the water’s edge is in everyone’s best interests, too.

from Popular Science http://ift.tt/1TZ6xDD
via IFTTT

NASA And Australia Successfully Test A Hypersonic Rocket

Humans have only flown for a few centuries. Balloons, the earliest human fliers, were not what anyone would call fast. Airplanes, first flown in 1903, started slow, and even now, the majority of human flight is subsonic, with only the highest-end military jets regularly clearing the sound barrier.

To probe the edges of flight, NASA, together with Australia’s Department of Defence and America’s Air Force Research Laboratory, wants to see if they can make aircraft work flying many times the speed of sound. Together, they’re working on a project called HIFiRE, for Hypersonic International Flight Research Experimentation Program.

This week, a successful test brought the HIFiRE rocket to a height of 172 miles, with a maximum speed 7.5 times the speed of sound.

From the Australian Department of Defense:

“The success of this test launch takes us one step closer to the realisation of hypersonic flight,” Dr Zelinsky said.
Hypersonic flight, involving speeds of more than five times the speed of sound, has the potential to provide immense social and economic benefits.
“It is a game-changing technology identified in the 2016 Defence White Paper and could revolutionise global air travel, providing cost-effective access to space,” Dr Zelinsky said.

Several earlier tests of hypersonic machines have failed, like this Army missile that exploded prematurely.

But the potential for a hypersonic plane is huge: with that much speed, it would be impossible for an enemy to move anything on the ground after radar detected the plane, letting it either capture pictures immediately or drop bombs almost unimpeded.

Here’s what NASA said about the HIFiRES program in 2012:

NASA, AFRL and Australia’s Defence Science and Technology Organisation (DSTO) are working with a number of partners on the HIFiRE (Hypersonic International Flight Research Experimentation Program) program to advance hypersonic flight — normally defined as beginning at Mach 5 — five times the speed of sound. The research program is aimed at exploring the fundamental technologies needed to achieve practical hypersonic flight. Being able to fly at hypersonic speeds could revolutionize high speed, long distance flight and provide more cost-effective access to space.

Check out another angle on the rocket below.

from Popular Science http://ift.tt/1YGzo3I
via IFTTT