Futuristic, SciFi, Cyberpunk, Future Tech ~ images, art, videos, info....
- Thread starter Lexx Diamond
- Start date
Halo Wars 2 Concept Art
https://www.forbes.com/sites/ibm/20...ard-universal-quantum-computing/#57e78deb5687
By Dario Gil, Ph.D., IBM Research, and Scott Crowder, Ph.D., IBM Systems
Quantum computing is the most exciting new frontier of information technology.
A universal quantum computer promises us more complete knowledge of our environment, down to the molecules that make up everything around us. And much like when the first room-sized computers were turned on in the 1940s, a quantum computer’s full potential is unknown and untapped. That’s why IBM isannouncingthe industry’s first commercial program to build a universal quantum computer: we are flipping the switch on something that doesn’t exist today.
Source: IBM Research
IBM Research's Four Qubit Square Circuit
Quantum computing is a term that’s becoming more mainstream, but its complex nature means it’s not well understood beyond quantum physicists. At its core, quantum is a radical new computing model that harnesses the power and rules of nature to address problems that are impractical to solve with today's systems.
The universal quantum computers IBM has worked toward in more than 35 years of research are the most powerful and general class of quantum computers. Their mojo comes from the quantum bit, or qubit, which leverages quantum effects that are not visible in our daily lives, to explore an exponentially more-powerful computational space, to help solve certain problems that we could not otherwise solve.
Today, the six founding members of theIBM Research Frontiers Institute-- Samsung, JSR, Honda, Hitachi Metals, Canon, and Nagase -- have set out with us to explore quantum computing applications for their industries. Our new goal is to deliver a universal quantum computer to select early access partners within the next few years.
Because of their exponential power, a universal quantum system with just 50 qubits may be able to perform certain complex calculations at a rate that today’s top multi-Petaflop supercomputers can’t yet emulate, nor can any other class of quantum computers. Our goal is to provide businesses and organizations with access to a new realm of computational power, before unachievable, to solve real-world and societal problems.
Quantum through computer science and integration
Access to theIBM Quantum Experience’sfive qubit processor was just the opening volley toward creating a universal quantum computer. The quantum experience was recently upgraded with a simulator capable of handling up to 20 qubits, and anew APIthat gives users a way to run batches of commands using scripting languages. We see this as the beginning ofquantum computer sciencewhen engineers and scientists without deep quantum physics expertise can start to develop quantum applications.
And similar to how the IBM Quantum Experience operates on thecloud,solving these problems will be possible because quantum computers will work in concert with classical computers. A classical computer will still do “traditional” things, like crunchingbig data.But partners in our early access commercial program will also be able to tap into a quantum computer to tackle complex, exponential problems in chemistry, machine learning, and optimization, to address applications ranging from chemistry, to drug discovery, to financial portfolio optimization.
Part of our universal quantum computing roadmap’s success relies on advancing high-performance computing systems and semiconductor technologies. Quantum computers, the end of Moore’s Law … these things are not the end of classical computing.
Instead, the continued innovations in system design for workload acceleration and software-defined storage enable classical systems to manage vast quantities of data, and to extract meaningful insights from that data. With cognitive infrastructure and quantum systems, we can pair insight from the data we humans have created and collected, with the possibility of exploring the world we don’t have data for, a world which quantum computers can reveal.
Think of it this way:Watsonprocesses the data we’ve collected on existing drugs to help us make better drugs. A quantum computer closely maps how molecules interact to help us form entirely new drugs.
Quantum computers ready to solve the unknown
At a conference in 1981 Nobel Laureate Richard Feynman said, in paraphrase, “Nature is quantum, and if you want to simulate nature, you better build a quantum computer.” He’s right. As this new program successfully develops increasingly capable universal quantum computers, it will open a new understanding of chemistry to help us develop new building materials, new drugs, and even optimize shipping logistics.
But perhaps even more exciting than what we know quantum computers will do is what we don’t know. With the advent of any very new technology often comes unexpected applications. No one is entirely sure of the full breadth of what a universal quantum computer is capable of. That’s exciting. And that’s what we want to find out.
Tweet this:Read about "Paving the path toward universal quantum computing" by Dario Gil and Scott Crowder @Forbes #IBMQ
Dario Gil, Ph.D., is Vice President of Science and Solutions at IBM Research. Scott Crowder, Ph.D., is Vice President and Chief Technology Officer of Quantum Computing, Technical Strategy and Transformation at IBM Systems.
Learn about the new era of business atwww.ibm/cognitive,and experience the IBM Cloud atInterConnect 20017.
A version of this blog appeared on IBM'sTHINK blog.
Appreciate this thread! Its turned into one of my faves.. The artwork is FIRE! It's inspired me to play with the drawing app on my tablet more.
That last story about the quantum computing blew me away! I'm not even a computer guy, but I see how they're working on something...and they don't even know where it's gonna take em!!
That last story about the quantum computing blew me away! I'm not even a computer guy, but I see how they're working on something...and they don't even know where it's gonna take em!!
Would YOU take this Italian UFO for a spin? Concept drone seats two people and hits speeds of 120mph
PUBLISHED: 07:39 EDT, 27 March 2017 | UPDATED: 11:00 EDT, 27 March 2017
This strange vehicle might look like a flying saucer, but it's not for extraterrestrials.
Humans looking for a trip that's truly 'out of this world' can now fly in style with the I.F.O or 'Identified Flying Object', a proposed two-seater drone concept that looks like a UFO.
The I.F.O is powered by eight electric rotary engines that push the saucer to an estimated top speed of 120mph (190kph) and can lift it to the average helicopter altitude.
People looking for a trip that's truly 'out of this world' can now fly in style with the I.F.O or 'Identified Flying Object' (pictured), a proposed two-seater drone/copter vehicle that looks like a UFO
VEHICLE SPECS
- 3 metres (10 feet) tall and two meters (6.5 feet) wide
- Seats two passengers
- Eight electric rotary engines
- Top speed of 120mph (190kph)
- Detachable cockpit with emergency parachute
- Six folding and extending feet
- Accessed via elevator or detachable foot bridges
The cockpit (pictured) is surrounded by a carbon fibre disk to protect passengers while keeping the drone's weight down
The I.F.O's futuristic cockpit can be accessed through an elevator that descends from the spherical body and then lifts passengers into the drone
Users can also enter and exit the capsule through two removable bridges (black slabs pictured above) that can be attached to a small ladder or staircase
For a smooth landing, the vehicle has six folding and extending feet with built-in cushioning suspension.
The I.F.O's futuristic cockpit can be accessed through an elevator that descends from the spherical body.
Users can also reach the capsule through two removable bridges that can be attached to a small ladder or staircase.
'The vehicle does not change much with respect to the technology that is at the base of a standard drone,' said lead designer Pierpaolo Lazzarini.
'But it is driven from the inside and can have the same elevation of a helicopter.
'I'm looking for investors to make it happen - the technology is on the market.'
It is not known what the final cost of the I.F.O drone will be.
The I.F.O is powered by eight electric rotary engines that push the flying saucer to an estimated top speed of 120mph (190kph) and can lift it to the average helicopter altitude, its designers said
Two doors on either side of the small cockpit allow users to access the I.F.O's engines and removable footbridges that can be stored in the vehicle
For a smooth landing, the vehicle has six folding and extending feet that come with cushioning suspension
The vehicle's cockpit can detach in the event of an emergency - a parachute attached to the back will guide passengers to safety
The flying vehicle stands three metres (10 feet) tall and has a capsule cockpit that measures two meters (6.5 feet) in diameter
The I.F.O's cockpit elevator gives passengers a smooth entry route into their vehicle
A series of rechargeable batteries located in the vehicle's outer disk power each engine. Two charging ports can be seen pictured at the back of the vehicle in this image
With a backup battery pack installed in the centre of the capsule body, the vehicle's estimated flight time ranges between sixty and seventy minutes
Designed by Pierpaolo Lazzarini from Italian company Jet Capsule, the drone's cockpit seats two passengers
It is not yet known what the final cost of the drone will be, nor where it will be legal to fly it
The I.F.O vehicle works much like modern drones but on a larger scale, its designers said
The technology is now on the market and vehicle designers Jet Capsule are waiting for investors
Read more: http://www.dailymail.co.uk/sciencet...gers-reaches-120mph-190kph.html#ixzz4cf6zm4yE
- Designed by Pierpaolo Lazzarini from Italian company Jet Capsule, the drone seats two people
- It is powered by eight electric engines that have a battery life of up to 70 minutes
- The cockpit capsule has a carbon fibre casing to protect users and is installed with an emergency parachute
- The vehicle's futuristic cockpit can be accessed through an elevator that descends from the vehicle's main body or via detachable foot bridges
PUBLISHED: 07:39 EDT, 27 March 2017 | UPDATED: 11:00 EDT, 27 March 2017
This strange vehicle might look like a flying saucer, but it's not for extraterrestrials.
Humans looking for a trip that's truly 'out of this world' can now fly in style with the I.F.O or 'Identified Flying Object', a proposed two-seater drone concept that looks like a UFO.
The I.F.O is powered by eight electric rotary engines that push the saucer to an estimated top speed of 120mph (190kph) and can lift it to the average helicopter altitude.
People looking for a trip that's truly 'out of this world' can now fly in style with the I.F.O or 'Identified Flying Object' (pictured), a proposed two-seater drone/copter vehicle that looks like a UFO
VEHICLE SPECS
- 3 metres (10 feet) tall and two meters (6.5 feet) wide
- Seats two passengers
- Eight electric rotary engines
- Top speed of 120mph (190kph)
- Detachable cockpit with emergency parachute
- Six folding and extending feet
- Accessed via elevator or detachable foot bridges
The cockpit (pictured) is surrounded by a carbon fibre disk to protect passengers while keeping the drone's weight down
The I.F.O's futuristic cockpit can be accessed through an elevator that descends from the spherical body and then lifts passengers into the drone
Users can also enter and exit the capsule through two removable bridges (black slabs pictured above) that can be attached to a small ladder or staircase
For a smooth landing, the vehicle has six folding and extending feet with built-in cushioning suspension.
The I.F.O's futuristic cockpit can be accessed through an elevator that descends from the spherical body.
Users can also reach the capsule through two removable bridges that can be attached to a small ladder or staircase.
'The vehicle does not change much with respect to the technology that is at the base of a standard drone,' said lead designer Pierpaolo Lazzarini.
'But it is driven from the inside and can have the same elevation of a helicopter.
'I'm looking for investors to make it happen - the technology is on the market.'
It is not known what the final cost of the I.F.O drone will be.
The I.F.O is powered by eight electric rotary engines that push the flying saucer to an estimated top speed of 120mph (190kph) and can lift it to the average helicopter altitude, its designers said
Two doors on either side of the small cockpit allow users to access the I.F.O's engines and removable footbridges that can be stored in the vehicle
For a smooth landing, the vehicle has six folding and extending feet that come with cushioning suspension
The vehicle's cockpit can detach in the event of an emergency - a parachute attached to the back will guide passengers to safety
The flying vehicle stands three metres (10 feet) tall and has a capsule cockpit that measures two meters (6.5 feet) in diameter
The I.F.O's cockpit elevator gives passengers a smooth entry route into their vehicle
A series of rechargeable batteries located in the vehicle's outer disk power each engine. Two charging ports can be seen pictured at the back of the vehicle in this image
With a backup battery pack installed in the centre of the capsule body, the vehicle's estimated flight time ranges between sixty and seventy minutes
Designed by Pierpaolo Lazzarini from Italian company Jet Capsule, the drone's cockpit seats two passengers
It is not yet known what the final cost of the drone will be, nor where it will be legal to fly it
The I.F.O vehicle works much like modern drones but on a larger scale, its designers said
The technology is now on the market and vehicle designers Jet Capsule are waiting for investors
Read more: http://www.dailymail.co.uk/sciencet...gers-reaches-120mph-190kph.html#ixzz4cf6zm4yE
Eve "vision car" opens new door on autonomous car design
C.C. Weiss March 27, 2017
The Eve does away with front and rear doors in favor of a single forward-sliding door on each side
While we were preoccupied with the Geneva Motor Show, electrifying automotive startup NIO was in the United States revealing its vision for the autonomous car of the future. NIO plans to branch out from its Nürburgring-scorching, 1-megawatt electric supercar and bring autonomous electric vehicles to US roads by 2020. The Eve "vision car" puts NIO's stamp on the world of autonomous car design with intriguing features like forward-sliding doors, a roomy interior layout and an active glass canopy that comes to life with information and entertainment.
It's amazing how quickly design can move from groundbreaking to formulaic. Just a few years ago, we began seeing some of the very first visions of autonomous car cabins – and they were awesome, completely disrupting the forward-facing, driver's-seat layout that has dominated for generations.
But as automaker after automaker after design house after automotive think tank has presented its own take on the autonomous car interior, elements like retractable steering wheels, front swivel seats, large digital displays and loads of panoramic glass have started to feel less radical and revolutionary and more utterly predictable. It seems like autonomous design opens up a lot more freedom for car designers to experiment with, but we're still just getting reshuffled versions of the same ideas.
The Eve has some of these predictable features, but NIO at least dangles its toes outside the box with some different ideas. The exterior is familiar enough, a stretched hatchback with a very concept-car presence. Faraday Future's research development VP Nick Sampson apparently thinks it's too familiar ... to the FF91 revealed (in drawn-out, torturous fashion) at CES 2017. Sampson took to social media after the Eve's reveal, implying it a clone and citing the design language and wheels as too close to what Faraday is doing.
Since the Eve is billed as nothing more than a vision, we'll let Faraday worry itself about those similarities. We're more interested in the differences, the first being the huge, forward-sliding doors that provide access to both the rear and front seats. Kind of like large, reverse minivan doors, these sliding slabs create wide, generous entry. The design requires the driver and front passenger to enter through the rear and make their way to their seats, assisted by a central aisle on the flat floor.
The interior houses a layout very different from what you'd expect to find in a six-seater. The front is simple enough, with driver and passenger seats wrapping their way into the side panels and dashboard to create a cozy little cocoon. Things gets more interesting in back, where the cushion on the back of the driver's seat accommodates a passenger facing the rearmost passenger. A fold-out table between them provides space for playing games, doing some work on a laptop, or diving into a snack.
Then there's the seat that everyone will fight over. Much like Volvo's extra-comfy rear seat, the Eve's passenger-side rear seat is sort of the VIP position, complete with a recline function and wraparound headrest. The good news for the other passengers is that the combination of autonomous driving technology and open layout should allow for switching seats while the car is in motion, giving everyone the chance for a fully reclined nap. The sixth (and seemingly least comfortable) seat is across from this VIP recliner, a fold-out cushion built into the front passenger seat back.
So there are no RV-style swivel seats, but there is the ubiquitous retractable steering wheel in perhaps the roughest iteration we've ever seen due to its weird rectangular shape. There are also deployable pedals so that "manual driving can still be experienced." How quaint.
When the driver isn't living like the ancients and exerting themselves steering, accelerating and braking, they can enjoy an explosion of content on the digital glass canopy, which can display basic information, augmented reality, entertainment content and more. For instance, NIO's digital assistant NOMI can use that glass to bring local points of interest to life with accompanying information, or highlight specific stars and constellations against the sparkling night sky. There's also a more traditional infotainment display in the "transparent plane" that stands in for the dashboard. Touch panel controllers help occupants navigate through settings and content.
Like previous autonomous car concepts, such as the Chrysler Portal andRinspeed Oasis, the Eve's design incorporates a high percentage of transparent material. Its sides feature active darkening, allowing passengers to switch from transparent to opaque, depending upon whether they want panoramic views or shade from the sun.
"NIO's vision is to give people time back so that they can be everything they want to be," NIO CEO Padmasree Warrior wrote in a Linkedin Pulse post a few days after the car's reveal. "With NIO autonomous electric vehicles, you will be as productive as you would be at your desk, or as relaxed as you would be on your couch. As the car drives, you can conduct a video conference call for work or catch up on last night's episode of Limitless orSupergirl. The vehicle's interior can be set up for a commute, a road trip or even a nap. The car's AI system will know where you are going, what's on your calendar and it will adjust to your needs."
NIO intends to support that autonomous ride with technologies like vehicle-to-vehicle communications, pedestrian detection and multi-car "platooning."
A few weeks before revealing the Eve show car at SXSW earlier this month, NIO raced an autonomous version of its EP9 supercar around the Circuit of the Americas track in Austin, Texas, setting a track record for an autonomous car with a 2:40.33 run. It says the car completed the lap without any interventions, topping out at 160 mph (257.5 km/h).
NIO will work with partners like Mobileye, Nvidia and NXP as it moves toward launching a production autonomous vehicle. Get more of a feel for the Eve in the video below.
C.C. Weiss March 27, 2017
The Eve does away with front and rear doors in favor of a single forward-sliding door on each side
While we were preoccupied with the Geneva Motor Show, electrifying automotive startup NIO was in the United States revealing its vision for the autonomous car of the future. NIO plans to branch out from its Nürburgring-scorching, 1-megawatt electric supercar and bring autonomous electric vehicles to US roads by 2020. The Eve "vision car" puts NIO's stamp on the world of autonomous car design with intriguing features like forward-sliding doors, a roomy interior layout and an active glass canopy that comes to life with information and entertainment.
It's amazing how quickly design can move from groundbreaking to formulaic. Just a few years ago, we began seeing some of the very first visions of autonomous car cabins – and they were awesome, completely disrupting the forward-facing, driver's-seat layout that has dominated for generations.
But as automaker after automaker after design house after automotive think tank has presented its own take on the autonomous car interior, elements like retractable steering wheels, front swivel seats, large digital displays and loads of panoramic glass have started to feel less radical and revolutionary and more utterly predictable. It seems like autonomous design opens up a lot more freedom for car designers to experiment with, but we're still just getting reshuffled versions of the same ideas.
The Eve has some of these predictable features, but NIO at least dangles its toes outside the box with some different ideas. The exterior is familiar enough, a stretched hatchback with a very concept-car presence. Faraday Future's research development VP Nick Sampson apparently thinks it's too familiar ... to the FF91 revealed (in drawn-out, torturous fashion) at CES 2017. Sampson took to social media after the Eve's reveal, implying it a clone and citing the design language and wheels as too close to what Faraday is doing.
Since the Eve is billed as nothing more than a vision, we'll let Faraday worry itself about those similarities. We're more interested in the differences, the first being the huge, forward-sliding doors that provide access to both the rear and front seats. Kind of like large, reverse minivan doors, these sliding slabs create wide, generous entry. The design requires the driver and front passenger to enter through the rear and make their way to their seats, assisted by a central aisle on the flat floor.
The interior houses a layout very different from what you'd expect to find in a six-seater. The front is simple enough, with driver and passenger seats wrapping their way into the side panels and dashboard to create a cozy little cocoon. Things gets more interesting in back, where the cushion on the back of the driver's seat accommodates a passenger facing the rearmost passenger. A fold-out table between them provides space for playing games, doing some work on a laptop, or diving into a snack.
Then there's the seat that everyone will fight over. Much like Volvo's extra-comfy rear seat, the Eve's passenger-side rear seat is sort of the VIP position, complete with a recline function and wraparound headrest. The good news for the other passengers is that the combination of autonomous driving technology and open layout should allow for switching seats while the car is in motion, giving everyone the chance for a fully reclined nap. The sixth (and seemingly least comfortable) seat is across from this VIP recliner, a fold-out cushion built into the front passenger seat back.
So there are no RV-style swivel seats, but there is the ubiquitous retractable steering wheel in perhaps the roughest iteration we've ever seen due to its weird rectangular shape. There are also deployable pedals so that "manual driving can still be experienced." How quaint.
When the driver isn't living like the ancients and exerting themselves steering, accelerating and braking, they can enjoy an explosion of content on the digital glass canopy, which can display basic information, augmented reality, entertainment content and more. For instance, NIO's digital assistant NOMI can use that glass to bring local points of interest to life with accompanying information, or highlight specific stars and constellations against the sparkling night sky. There's also a more traditional infotainment display in the "transparent plane" that stands in for the dashboard. Touch panel controllers help occupants navigate through settings and content.
Like previous autonomous car concepts, such as the Chrysler Portal andRinspeed Oasis, the Eve's design incorporates a high percentage of transparent material. Its sides feature active darkening, allowing passengers to switch from transparent to opaque, depending upon whether they want panoramic views or shade from the sun.
"NIO's vision is to give people time back so that they can be everything they want to be," NIO CEO Padmasree Warrior wrote in a Linkedin Pulse post a few days after the car's reveal. "With NIO autonomous electric vehicles, you will be as productive as you would be at your desk, or as relaxed as you would be on your couch. As the car drives, you can conduct a video conference call for work or catch up on last night's episode of Limitless orSupergirl. The vehicle's interior can be set up for a commute, a road trip or even a nap. The car's AI system will know where you are going, what's on your calendar and it will adjust to your needs."
NIO intends to support that autonomous ride with technologies like vehicle-to-vehicle communications, pedestrian detection and multi-car "platooning."
A few weeks before revealing the Eve show car at SXSW earlier this month, NIO raced an autonomous version of its EP9 supercar around the Circuit of the Americas track in Austin, Texas, setting a track record for an autonomous car with a 2:40.33 run. It says the car completed the lap without any interventions, topping out at 160 mph (257.5 km/h).
NIO will work with partners like Mobileye, Nvidia and NXP as it moves toward launching a production autonomous vehicle. Get more of a feel for the Eve in the video below.
Facebook teaches machines to negotiate with humans
Posted 1 hour ago by John Mannes
Though Facebook is rarely mentioned alongside Apple, Microsoft and Amazon in discussions about conversational AI, the company has published a hoard of papers that underscore a deep interest in dialog systems. As has become clear with Siri, Cortana and Alexa, dialog is hard — it requires more than just good speech recognition to deliver a killer experience to users. From the sidelines Facebook has been tinkering with big challenges like natural language understanding and text generation. And today the Facebook AI Research team added to its portfolio with a paper bringing negotiation into the conversation (all puns intended).
Facebook’s team smashed game theory together with deep learning to equip machines to negotiate with humans. By applying rollout techniques more commonly used in game-playing AIs to a dialog scenario, Facebook was able to create machines capable of complex bargaining.
To start, Facebook dreamed up an imaginary negotiation scenario. Humans on Amazon’s Mechanical Turk were given an explicit value function and told to negotiate in natural language to maximize reward by splitting up a pot of random objects — five books, three hats and two balls. The game was capped at ten rounds of dialog, the rules stated that nobody would receive any reward if that limit was exceeded.
Because each agent had distinct hidden preferences, the two had to engage in dialog to sort out which objects should be given to which agent. Over the course of the interactions, machines naturally adopted many common negotiation tactics — like placing false emphasis on a low-value item in an attempt to use it as a more valuable bargaining chip later.
Under the hood, Facebook’s rollout technique takes the form of a decision tree. Decision trees are a critical component of many intelligent systems. They allow us to model future states from the present to make decisions. Imagine a game of tic-tac-toe, at any given point of the game, there is a finite option set (places you can place your “X” on the board.
In that scenario, each move has an expected value. Humans don’t usually consider this value in an explicit way but if you decompose your decision process when playing the game, you are effectively short-handing this math in your head.
Games like Tic Tac Toe are simple enough that they can be completely solved in a decision tree. More complex games like Go and Chess require strategies and heuristics to reduce the total number of states (it’s an almost unimaginable number of possible states). But even Chess and Go are relatively simple compared to dialog.
Dialog doesn’t draw from a finite set of outcomes. This means that for any question, there is an infinite number of possible human responses. To model a conversation, researchers have to take extra effort to bound the uncertainty problem into a reasonable size and scope. Opting to model a negotiation scheme makes this possible. The language itself can exist in an infinite number of states but its intent generally clusters around simple outcomes (I’ll take the deal or reject it).
But even in a bounded world, it’s still difficult to get machines to interact with humans in a believable way. To this avail, Facebook trained its models on negotiations between pairs of people. Once this was done, the machines were set up to negotiate with each other using reinforcement learning. At the end of each round of conversation, agents received rewards to guide improvement.
FAIR researchers Michael Lewis and Dhruv Batra explained to me that their algorithms were better at preventing individuals from making bad decisions than ensuring individuals made the best decisions. This is still important — the team told me to imagine a calendar application that doesn’t try to schedule meetings for the best time for everyone but instead tries to just ensure the meeting actually happens.
As with a lot of research, the application of this technology isn’t necessarily as explicit as the scenario simulated for the paper. Engineers often employ adversarial relationships between machines to improve outcomes — think using generative adversarial networks to generate training data by having a machine generate data looking to fool another “gatekeeper” machine.
Semi-cooperative, semi adversarial relationships, like the relationship between a coach and an athlete, could be an interesting next frontier — further connecting game theory and machine learning.
Facebook has open sourced its code from this research project. If you’re interested, you can read additional details about the work in the full paper here.
Posted 1 hour ago by John Mannes
Though Facebook is rarely mentioned alongside Apple, Microsoft and Amazon in discussions about conversational AI, the company has published a hoard of papers that underscore a deep interest in dialog systems. As has become clear with Siri, Cortana and Alexa, dialog is hard — it requires more than just good speech recognition to deliver a killer experience to users. From the sidelines Facebook has been tinkering with big challenges like natural language understanding and text generation. And today the Facebook AI Research team added to its portfolio with a paper bringing negotiation into the conversation (all puns intended).
Facebook’s team smashed game theory together with deep learning to equip machines to negotiate with humans. By applying rollout techniques more commonly used in game-playing AIs to a dialog scenario, Facebook was able to create machines capable of complex bargaining.
To start, Facebook dreamed up an imaginary negotiation scenario. Humans on Amazon’s Mechanical Turk were given an explicit value function and told to negotiate in natural language to maximize reward by splitting up a pot of random objects — five books, three hats and two balls. The game was capped at ten rounds of dialog, the rules stated that nobody would receive any reward if that limit was exceeded.
Because each agent had distinct hidden preferences, the two had to engage in dialog to sort out which objects should be given to which agent. Over the course of the interactions, machines naturally adopted many common negotiation tactics — like placing false emphasis on a low-value item in an attempt to use it as a more valuable bargaining chip later.
Under the hood, Facebook’s rollout technique takes the form of a decision tree. Decision trees are a critical component of many intelligent systems. They allow us to model future states from the present to make decisions. Imagine a game of tic-tac-toe, at any given point of the game, there is a finite option set (places you can place your “X” on the board.
In that scenario, each move has an expected value. Humans don’t usually consider this value in an explicit way but if you decompose your decision process when playing the game, you are effectively short-handing this math in your head.
Games like Tic Tac Toe are simple enough that they can be completely solved in a decision tree. More complex games like Go and Chess require strategies and heuristics to reduce the total number of states (it’s an almost unimaginable number of possible states). But even Chess and Go are relatively simple compared to dialog.
Dialog doesn’t draw from a finite set of outcomes. This means that for any question, there is an infinite number of possible human responses. To model a conversation, researchers have to take extra effort to bound the uncertainty problem into a reasonable size and scope. Opting to model a negotiation scheme makes this possible. The language itself can exist in an infinite number of states but its intent generally clusters around simple outcomes (I’ll take the deal or reject it).
But even in a bounded world, it’s still difficult to get machines to interact with humans in a believable way. To this avail, Facebook trained its models on negotiations between pairs of people. Once this was done, the machines were set up to negotiate with each other using reinforcement learning. At the end of each round of conversation, agents received rewards to guide improvement.
FAIR researchers Michael Lewis and Dhruv Batra explained to me that their algorithms were better at preventing individuals from making bad decisions than ensuring individuals made the best decisions. This is still important — the team told me to imagine a calendar application that doesn’t try to schedule meetings for the best time for everyone but instead tries to just ensure the meeting actually happens.
As with a lot of research, the application of this technology isn’t necessarily as explicit as the scenario simulated for the paper. Engineers often employ adversarial relationships between machines to improve outcomes — think using generative adversarial networks to generate training data by having a machine generate data looking to fool another “gatekeeper” machine.
Semi-cooperative, semi adversarial relationships, like the relationship between a coach and an athlete, could be an interesting next frontier — further connecting game theory and machine learning.
Facebook has open sourced its code from this research project. If you’re interested, you can read additional details about the work in the full paper here.
This chick does some cool artwork,i follow her on FB
https://hybridgothica.artstation.com/
Not bad-looking herself,either..
Ori robotic furniture transforms studio apartments into so much more
Alexa’s involved of course
BY JENNY XIE@CANONIND
MAY 31, 2017, 9:31AM EDT TWEET
All photos courtesy Ori
Big news in the transforming furniture world: Ori, an MIT spin-off designed with Yves Behar is officially launching preorders today, aiming at developers trying to make urban studio apartments much more flexible and appealing.
First announced last year, Ori is the most sophisticated “all-in-one” furniture system we’ve seen yet. In a nutshell: It’s a multi-functional module that serves different functions with different configurations, which can be controlled via a physical controller attached to the system, mobile app, or, because this is 2017, voice commands for the Amazon Echo. (Google Home and Apple Home integrations, as well as additional skills for the Alexa, are currently under development.)
“Hey Alexa, can you ask Ori to make the bed?” This might just be the line to catch everyone’s attention in the promo video below, which demonstrates an enticing array of spaces an Ori system can create in a tiny apartment—from a standard living area to a yoga studio to a “walk-in closet.”
The Ori prototype has been tested by Airbnb guests in Boston for over a year. In an interview with Curbed, Ori CEO and Founder Hasier Larrea says adjustments have been made based on their feedback, including raising the height of the bed and fine-tuning reliability and safety features (such as the bed knowing to stop expanding when it hits something.)
Made of poplar plywood, the Ori system comes in two colors (light and dark), two sizes (full and queen), and two orientations (one that expands to one side only, and another that expands on both sides to create two separate spaces in longer apartments.)
The system is flatpack and connects to the room itself in only two places: the outlet and a gliding track system essentially taped to the bottom edge of the room. This non-intrusive assembly makes it easy for developers to transform a regular rental into a smart one.
The whole system costs some $10,000 to put together (with most of the expense falling on the furniture rather than the technology). According to Maria Masi of Brookfield Property Partners, whose New York City property The Eugene already has an Ori system installed in a model studio, an Ori-equipped apartment could command hundreds more dollars in rent since it would be positioned as a studio that functions more like a furnished one-bedroom.
Model Ori systems are currently installed in modern apartment complexes like The Eugene in ten cities around the U.S. and Canada. Preorders are expected to be delivered later this year.
Alexa’s involved of course
BY JENNY XIE@CANONIND
MAY 31, 2017, 9:31AM EDT TWEET
Big news in the transforming furniture world: Ori, an MIT spin-off designed with Yves Behar is officially launching preorders today, aiming at developers trying to make urban studio apartments much more flexible and appealing.
First announced last year, Ori is the most sophisticated “all-in-one” furniture system we’ve seen yet. In a nutshell: It’s a multi-functional module that serves different functions with different configurations, which can be controlled via a physical controller attached to the system, mobile app, or, because this is 2017, voice commands for the Amazon Echo. (Google Home and Apple Home integrations, as well as additional skills for the Alexa, are currently under development.)
“Hey Alexa, can you ask Ori to make the bed?” This might just be the line to catch everyone’s attention in the promo video below, which demonstrates an enticing array of spaces an Ori system can create in a tiny apartment—from a standard living area to a yoga studio to a “walk-in closet.”
The Ori prototype has been tested by Airbnb guests in Boston for over a year. In an interview with Curbed, Ori CEO and Founder Hasier Larrea says adjustments have been made based on their feedback, including raising the height of the bed and fine-tuning reliability and safety features (such as the bed knowing to stop expanding when it hits something.)
Made of poplar plywood, the Ori system comes in two colors (light and dark), two sizes (full and queen), and two orientations (one that expands to one side only, and another that expands on both sides to create two separate spaces in longer apartments.)
The system is flatpack and connects to the room itself in only two places: the outlet and a gliding track system essentially taped to the bottom edge of the room. This non-intrusive assembly makes it easy for developers to transform a regular rental into a smart one.
The whole system costs some $10,000 to put together (with most of the expense falling on the furniture rather than the technology). According to Maria Masi of Brookfield Property Partners, whose New York City property The Eugene already has an Ori system installed in a model studio, an Ori-equipped apartment could command hundreds more dollars in rent since it would be positioned as a studio that functions more like a furnished one-bedroom.
Model Ori systems are currently installed in modern apartment complexes like The Eugene in ten cities around the U.S. and Canada. Preorders are expected to be delivered later this year.
