Being able to reason through potential future events is something humans are pretty good at doing, but that kind of ability is a real challenge when it comes to training AI. Taking those reasoning skills and using them to create a plan is even more difficult, but the Google DeepMind team has begun to tackle this problem. In a recent blog post, researchers describe new approaches they’ve developed for introducing "imagination-based planning" to AI.
Other programs have been able to work in planning abilities, but only within limited environments. AlphaGo, for example, can do this well, as the researchers note in the blog post, however, they add that "environments like Go are ‘perfect’ – they have clearly defined rules which allow outcomes to be predicted very accurately in almost every circumstance." Facebook also created a bot that could reason through dialogue before engaging in conversation, but again, that was in a fairly restricted environment. "But the real world is complex, rules are not so clearly defined and unpredictable problems often arise. Even for the most intelligent agents, imagining in these complex environments is a long and costly process," said the blog post.
DeepMind researchers created what they’re calling "imagination-augmented agents," or I2As, that have a neural network trained to extract any information from its environment that could be useful in making decisions later on. These agents can create, evaluate and follow through on plans. To construct and evaluate future plans, the I2As "imagine" actions and outcomes in sequence before deciding which plan to execute. They can also choose how they want to imagine, options for which include trying out different possible actions separately or chaining actions together in a sequence. A third option allows the I2As to create an "imagination tree," which lets the agent choose to continue imagining from any imaginary situation created since the last action it took. And an imagined action can be proposed from any of those previously imagined states, thus creating a tree.
The researchers tested the I2As on the puzzle game Sokoban and a spaceship navigation game, both of which require planning and reasoning. You can watch the agent playing Sokoban in the video below. For both tasks, the I2As performed better than agents without future reasoning abilities, were able to learn with less experience and were able to handle imperfect environments.
DeepMind AI has been taught how to navigate a parkour course and recall past knowledge and researchers have used it to explore how AI agents might cooperate or conflict with each other. When it comes to planning ability and future reasoning, there’s still a lot of work to be done, but this first look is a promising step towards imaginative AI.
from Engadget http://engt.co/2uUE8hd
Reason users – don’t forget that VST plugins are now compatible with Reason (9.5), so now Circle² can be loaded into the rack as you would any other inbuilt instrument. You can even still patch audio and CV as normal!
from Weblog – Future Audio Workshop http://bit.ly/2uuQFH9
Lamassu is a bitcoin ATM manufacturer that started in 2013 with the goal of making it “as simple as possible to get bitcoin.” There are now hundreds of these bitcoin ATMs around the world, and one of Lamassu’s co-founders, Zach Harvey, recently shared some data related to how these machines are used at a Bitcoin meetup in Milan.
While a recent article in the New York Post warned bodega owners against putting bitcoin ATMs in their stores due to potential use by darknet market customers, Harvey stated his belief that most of the bitcoin ATM users are using the devices for speculation on the bitcoin price, which he sees as the biggest use case for the digital asset right now.
According to Harvey, Lamassu set up their business in a decentralized, distributed manner where they manufacture the machines in Portugal and then sell them to operators around the world.
“The reason we did this in a way that is more distributed is because we felt the system would be a lot more robust if every one of these individual operators ran the machines themselves, had their own bitcoin wallets from which to send to the end user, had full control over their user data, there would be no single point of failure, and it would also be more in the spirit of Bitcoin,” said Harvey at the recent meetup.
Harvey went on to discuss data related to the use of bitcoin ATMs, problems associated with increased congestion on the Bitcoin network, and a specific example of why he thinks bitcoin is going mainstream.
Who Uses Bitcoin ATMs?
According to Harvey, the people who use bitcoin ATMs are mostly non-tech savvy users who want to get their first taste of digital currency. Harvey added that bitcoin ATMs tend to attract these types of users due to a focus on convenience and user experience. The Lamassu co-founder also claimed that a new user can complete a transaction in 20 to 30 seconds, while someone familiar with the machine can be done in less than ten seconds.
Two of the key selling points of bitcoin ATMs mentioned by Harvey were that users don’t have to go through the process of connecting a bank account with an exchange and the machines can feel like a safer option than meeting up with a random person found on a P2P bitcoin trading marketplace like LocalBitcoins.
According to Harvey, the selling points of bitcoin ATMs are so strong that many people are willing to pay the 10 to 15 percent exchange fees that come with them. He claimed that the more popular bitcoin ATMs get around 50 transactions per day and “sometimes you’ll even have queue at some of the machines.”
“If you’re around a bitcoin ATM, there really is no easier way of [getting some bitcoin],” Harvey later added.
Having said that, Harvey indicated that many of the bitcoin ATMs in the United States include some sort of identity verification due to Know Your Customer and anti-money laundering regulations. The level of identity verification required tends to vary, depending on the amounts involved and where the bitcoin ATM is located.
Harvey also shared data from one of their operators, who owns 14 machines, that indicated these bitcoin ATMs tend to be used for low-value transactions.
“These are people that just want to get the first experience — see what it’s like to get into bitcoin,” said Harvey. “If they really just want to see what it’s about — feel a little bit of the magic of bitcoin — they’re going to start with a low amount.”
“If you look at machine number 13, there’s almost 90 percent of transactions that are under $100,” Harvey added.
According to the data shared by Harvey, 20 to 30 percent of the transactions at these particular bitcoin ATMs are for less than $10.
Network Transaction Fees Have an Effect on Bitcoin ATMs
During his appearance at the Bitcoin meetup in Milan, Harvey also discussed the effect that increased congestion on the Bitcoin network has had on bitcoin ATMs. He noted that operators asked for the functionality to add a flat fee or the complete removal of the $5 and $10 transaction amounts when on-chain transaction fees get into the $1–$2 range.
Harvey also noted that roughly 90 percent of the transactions that are usually processed by bitcoin ATMs would become uneconomical if on-chain bitcoin transaction fees reached $10.
Due to demand from their operators, Lamassu plans to add altcoins, such as ether and zcash, to their machines in the coming weeks. These altcoins feature shorter confirmation times, which can be helpful in situations where a user wants to trade their cryptocurrency for physical cash.
“It’s not as secure as a bitcoin confirmation, but it’s more secure than a zero-confirmation of bitcoin,” Harvey said of confirmations on other cryptocurrency networks.
Due to Bitcoin network congestion and poorly implemented fee estimation software on users’ mobile wallets, Harvey noted that some users have had to wait over a day to get their cash out of a bitcoin ATM.
In terms of unconfirmed bitcoin transactions, Harvey stated that not many operators have had issues with accepting them.
Bitcoin Going Mainstream?
Although Lamassu also has machines in North America, Europe, Asia, Australia and New Zealand, Canada has become their most active userbase. According to data shared by Harvey, the Toronto area alone has around 50 Lamassu bitcoin ATMs.
“Several years ago, [this] would have seemed like way too many, and now it’s starting to be something that’s the norm as bitcoin goes more mainstream,” said Harvey of the density of bitcoin ATMs in Toronto.
When sharing data related to the proliferation of Lamassu’s bitcoin ATMs around the world, Harvey showed a screenshot of an email he received from a convenience store owner in Toronto.
“Customers are coming into my store, and they’re telling me, ‘Why don’t you have a bitcoin ATM?’” Harvey paraphrased from the email. “Can I please get one placed here?”
This is the same area that already has roughly 50 bitcoin ATMs around it.
Watch the full presentation here.
The post Lamassu’s Zach Harvey Shares Data on the Growing Use of Bitcoin ATMs appeared first on Bitcoin Magazine.
from Bitcoin Magazine http://bit.ly/2vXYlzm
The design team created two versions: One is made the make the wearer invisible, allowing them to blend into the ocean. The other is a wetsuit that has blue and white stripes, a pattern that is found in nature and tells sharks that the prey is not safe to eat.
from Mashable! http://on.mash.to/2vXSHx2
Tel Aviv-based startup Prospera has raised a $15 million Series B to expand the scope of its technology, which uses computer vision and artificial intelligence to help farmers analyze data gathered from their fields. The round was led by Qualcomm Ventures, with participation from Cisco Investments, ICV and returning investor Bessemer Venture Partners, and brings Prospera’s total funding to $22 million (its Series A was covered by TechCrunch in July 2016).
The startup’s new capital will be used enter more global markets, add people to its delivery and customer-facing teams and broaden its services to cover more crops, “including making a key shift from indoor farms to outdoor farms, which has huge implications given that 40 percent of U.S. land is farmland,” says co-founder and CEO Daniel Koppel.
Since its Series A, Prospera has added new customers in Europe, Mexico, and the U.S. and now claims thousands of users, including produce growers for Walmart, Tesco, Sainbury’s, and Aldi.
Its technology has also evolved from its previous focus on automatically detecting pests and diseases to “every aspect of farm production,” says Koppel. This includes agronomy (the science of soil management and crop production), operations and managing a farm’s labor to increase its bottom line.
Some of the most notable companies also looking at agtech include drone makers like DJI and Agribotix (interestingly, Prospera’s new investors Qualcomm and Cisco are both working tech to support the development and manufacturing of drones).
Koppel doesn’t see those companies as competitors, but as potential partners. “Drones will provide another valuable data stream for our analyses, enriching our database and potentially helping us provide even more value to our clients,” he says.
Featured Image: Pgiam/Getty Images
from TechCrunch http://tcrn.ch/2tHE6ZY
Thanks to all those Star Wars movies we know the Jedi can barely survive an attack from just a pair of evil Sith lords. But pit a small battalion of 300 lightsaber-wielding Jedi knights against a giant army of 60,000 medieval soldiers armed with only swords, and it’s not even a challenge.
This latest simulation, courtesy of YouTuber SergiuHellDragoonHQ and a PC game called Ultimate Epic Battle Simulator, is yet another Star Wars sequel waiting to happen. The Jedi spend a few years shoring up their numbers again, and then find a time machine which takes them back (or forward?) to medieval England to right history’s wrongs. It has summer blockbuster written all over it!
from Gizmodo http://bit.ly/2vY8wnK
Manila, the capital of the Philippines, is one of the most crowded cities on Earth. An estimated 200,000 people live in a single square mile in some neighborhoods — nearly three times the density of Manhattan.
In 2017, German photographer Bernhard Lang set out to capture the overpopulation crisis from above. The images show the residents’ living conditions as they are rarely seen. Lang shared some photos from his series with us. You can check out more of his work on Instagram and Facebook.
About one billion people around the world live in slums, the BBC reports. About 2% of slum dwellers worldwide can be found in the Philippines, an island nation.
The capital city of Manila sits on the northwest coast of the Philippines. It contains a booming business district, residential areas, and slums and shanty towns on the fringes.
In the poorest regions, multiple families cram into makeshift homes along the rivers. The units are built on stilts as a precaution against frequent flooding.
from SAI http://read.bi/2v5eaYA
Like a baseball player running to make a catch, dragonflies are also capable of predicting the trajectory of a moving object, typically its next meal. New research is revealing the mechanisms behind this complex cognitive task, which was once thought to be exclusive to mammals. It’s hoped that these insights will lead to innovations in robot vision.
As humans, we take it for granted that we can track and predict the trajectory an object moving through time and space. We can focus on a moving object when there’s a lot going on in the background, and even temporarily take our eyes off the object, knowing where it’ll be a few moments later. We can do this because our brains are equipped to make these complex calculations in real time. As a new study published in eLife points out, similar abilities are also part of the dragonfly’s cognitive arsenal.
Back in 2012, Steven Wiederman from the University of Adelaide, with the help of Lund University biologist David O’Carroll, discovered that dragonflies can do something scientists didn’t think invertebrates were capable of: “selective attention.” Dragonflies use this skill when hunting for insects, many of which take refuge in swarms. Their observations showed that dragonflies can lock onto a single target, and ignore any distractions as they dives in for the attack. Trouble is, the researchers had no idea how dragonflies were capable of knowing where moving objects would be in the future, mostly because insects weren’t thought to have the requisite brain architecture to allow for such an advanced skill.
To figure out how dragonflies were such effective hunters, Wiederman and O’Carroll decided to conduct a new study. They brought 63 male dragonflies back to their lab and immobilized them with a wax-rosin mixture. With electrodes attached to their brains, the dragonflies were shown a series of black squares intended to mimic prey. As they were shown simulated prey, the researchers watched as the activity of Small Target Motion Detectors (or STMD neurons for short) fired up in the dragonfly brains.
Instead of just following the target through time and space along a straight line (as most insects are thought to do), the STMD neurons actually worked to predict the future location of the black dots, according to the new study. These specialized neurons increased their activity in a tiny “focus” area just in front of the moving object being tracked; individual neurons were more sensitive to movements just ahead of the black dot’s current position. Also, when the black dot suddenly disappeared (as a target might do when, say, it flies behind a tree), the dragonfly brain still worked to predict where the dot was most likely to reappear. Insects without this trajectory-crunching capacity would just give up the chase once an object disappears from its vision.
“Insects and mammals last shared a common ancestor more than 500 million years ago, and, in many respects, mammalian brains are substantially more complex than insect brains,” conclude the authors in the study. “Nevertheless, [our findings] show that the insect brain can perform visual tasks that were previously associated only with mammals.”
This is an exciting finding, and not just because of what it tells us about insectoid brains. In the coming years, scientists can use these insights to develop safer and more capable autonomous robots, whether they be self-driving vehicles (which, like dragonflies, are moving objects that need to predict the trajectory of other moving objects), or flying robots used for reconnaissance purposes or as artificial pollinators.
Indeed, there’s no need to reinvent the wheel; nature has already performed plenty of research and development on our behalf over the millennia.
from Gizmodo http://bit.ly/2v5Ko6a
Cell phones are being used as solar computers to save the rain forests
Cell phones have powerful computers inside them. This company is using old ones to save the rain forests.
Mashable’s new series, A Cleaner Future, highlights the best and brightest innovators working to change our world for the better.
Subscribe for new episodes of A Cleaner Future and more: http://on.mash.to/subscribe
from Mashable! http://on.mash.to/2v5oP5E