Category: AI

Categories
AI Assistive Technologies

Eye tracking virtual keyboard

Click2Speak is SwiftKey based software that allows users to type on a virtual keyboard using eye movements.   AI technology predicts words from texts, facebook, and twitter.  A camera tracks eye movement, and one can click with a foot mouse or by looking at a button for a few seconds.

Founder Gal Sont, who has ALS, created Click2Speak because “Your communication is the basis of everything.  To tell someone you love them, to ask for something to drink, it is the basic need of every one of us: to communicate with my family and friends and tell them all that I need, in words. So communicating is very important, and if I can do it faster and more efficiently,  I’ve won the world.”

Categories
AI Machine Learning Robotics

Context sensitive robot understands casual language

Cornell professor Ashutosh Saxenam is developing a context sensitive robot that is able to understand natural language commands, from different speakers, in colloquial English.  The goal is to help robots account for missing information when receiving instructions and adapt to their environment.

Tell Me Dave is equipped with a 3D camera for viewing its surroundings.  Machine learning has enabled it to respond to entire commands with flexibly defined actions. Its  computer “brain” has been fed simulated video simulations of actions, accompanied by voice commands from different speakers of different dialects and accents. The robot matches instructions to a range of potential actions and determines the the right one using the context of other words in the command and environment details.

At this time, Tell Me Dave can correctly follow human voice instructions 64 percent of the time.

Categories
AI

Chatbot passes Turing Test

A chatterbot named Eugene Goostman has become the first to pass the Turing Test.

“Eugene” and four other contenders participated in the Turing Test 2014 Competition at the Royal Society in London.  Each chatterbox was required to engage in a series of five-minute text-based conversations with a panel of judges.  A computer passes the test if it is mistaken for a human more than 30% of the time. Eugene convinced 33% of the judges it was human– the only machine in history to do so.

The competition was held on the 60th anniversary of the death of Alan Turing, the great British mathematician, logician, cryptanalyst , computer scientist and philosopher.

During World War II, Turing worked for the Government Code and Cypher School (GC&CS) at Bletchley Park, Britain’s code breaking center.  He led Hut 8, the section responsible for German naval cryptanalysis. He devised a number of techniques for breaking German ciphers, and improved the pre-war Polish bombe method, an electromechanical machine that could find settings for the Enigma machine.

After the war, he worked at the National Physical Laboratory, where he designed the ACE, among the first designs for a stored-program computer.

The great Alan Turing was highly influential in the development of computer science, providing a formalization of the concepts of “algorithm” and “computation”. Turing is widely considered  the “father” of theoretical computer science and artificial intelligence.

The shameful British government prosecuted Turing  for being gay, showing no respect for a man whose contributions to Britain and the world were enormous.  He accepted treatment with estrogen injections (chemical castration) as an alternative to prison, and later committed suicide.

Categories
AI Heart Pregnancy Respiratory

Human body simulation for health research

The Virtual Physiological Human project is a computer simulated replica of the human body that is being created to test drugs and treatments.  It will allow physicians to model the mechanical, physical and biochemical functions of the body as a single complex system rather than as a collection of organs.  The goal is to offer personalized treatment of disease.

The University of Sheffield is significantly contributing to the project:

Dr Paul Morris is creating personalized “virtual arteries” using images of a patient’s heart, which can accurately predict how effective an operation, such as the introduction of a stent, might be.

Professor Jim Wild believes that computer models of a patient’s lungs could allow doctors to detect signs of lung diseases such as emphysema much earlier.

Dr Xinshan Li  is researching the impact of pregnancy on women’s pelvic floor muscles.   “It’s positioned as a predictive tool,” she said. “The idea is to get the geometry of your pelvic floor either during pregnancy or before you become pregnant, and then run simulations of the birth process and see how likely it is that the muscle will be damaged. If the risk is high, then there are certain interventions we can do during the birth process to mitigate the risk.”

Categories
AI Brain

Brain based machine learning software

Vicarious FPC claims to be “building software that thinks and learns like a human.”  Their goal is to replicate the neocortex, the part of the brain that sees, controls the body, understands language and does math.

Its first project is a visual perception system that interprets contents of photographs and videos.  The next milestone is to create a computer that can understand the textures associated with shapes and objects.

Co-founder Scott Phoenix hopes that Vicarious’s computers will learn to how to cure diseases, create cheap, renewable energy, and perform the jobs that employ most human beings.    While these goals are ambitious and long-term, the recent backing of Elon Musk, Mark Zuckerberg and Ashton Kutcher will, hopefully, help the company benefit humanity through artificial intelligence.

Categories
AI

Facebook’s DeepFace achieves near human-level face verification

Following its acquisition of Tel Aviv University Professor Lior Wolf‘s face.com, Facebook is developing DeepFace, with near-human level face recognition capabilities.

The technology processes images of faces in two steps. First it corrects the angle of a face so that the person in the picture faces forward, using a 3-D model of an “average” forward-looking face. Then a simulated neural network determines a numerical description of the reoriented face. If DeepFace comes up with similar enough descriptions from two different images, it decides they must show the same face.

Categories
AI BCI Brain Robotics

Collaborative cloud system for human-serving robots

http://www.roboearth.org/what-is-roboearth

RoboEarth’s goal is to speed the development of human-serving robots. Scientists from five European universities gathered this week for its launch and demonstrated potential applications. This included a robot scanning a room’s physical layout, including the location of the patient’s bed, and the placing a carton of milk on a nearby table.

The system is sometimes called a “Wikipedia” for robots, allowing them, or their programmers,  to share and retrieve information.

Researchers claim that the robots’ technically sophisticated capabilities are comparable to those of high-end robots in car factories. They, however,  look clumsier because robots that interact with humans are not performing repetitive tasks in the controlled, sanitized and predictable surroundings of a factory.

RoboEarth’s system of networked computers enable it to perform intensive computing tasks that smaller computers (or simpler robots) cannot.  It also allows individual robots to communicate between themselves through its “RoboCloud” and robot database. This could facilitate light and cheap robots that use the cloud for computing.

Categories
AI BCI Brain

K supercomputer runs largest neural simulation to date

http://www.telegraph.co.uk/technology/10567942/Supercomputer-models-one-second-of-human-brain-activity.html

RIKEN, Okinawa Institute of Science and Technology, and Forschungszentrum Jülich researchers have used Japan’s K computer to run the longest brain simulation to date.With 705,024 processor cores, running at speeds of over 10 petaflops, it is ranked the world’s fourth-most powerful computer. Using Neural Simulation Technology software and 92,944 of its processors, the computer replicated one second of brain activity across 1.73 billion nerve cells and 10.4 trillion synapses.  This represents one per cent of the brain’s neuronal network and took the K computer 40 minutes.

“The new result paves the way for combined simulations of the brain and the musculoskeletal system using the K computer,” said Kenji Doya of the Okinawa Institute of Science of Technology,. “These results demonstrate that neuroscience can make full use of the existing peta-scale supercomputers.”

“If peta-scale computers like the K computer are capable of representing one per cent of the network of a human brain today, then we know that simulating the whole brain at the level of the individual nerve cell and its synapses will be possible with exa-scale computers hopefully available within the next decade,” said project leader Markus Diesmann.

Categories
AI Assistive Technologies BCI Brain

Johns Hopkins develops thought controlled prosthetic arm and “targeted innervation” technique

http://hub.jhu.edu/2013/01/02/prosthetic-arm-60-minutes

The number of researchers developing advanced prosthetics, particularly thought controlled limbs, is increasing rapidly. This can significantly impact the lives of many.  In Johns Hopkins Universty’s Applied Physics Lab, a motorized arm with a five fingered hand that operates much like human hand is nearing completion.

Professor Michael McLoughlin and trauma surgeon Albert Chi have developed a technique known as targeted innervations—in which nerves can be rerouted through spare muscle, allowing amputees to operate motorized prosthetics using motor commands.  In a recent surgery, Dr. Chi successfully combined this technique with the aforementioned prosthetic.

“The body is amazing in terms of its will to return to normal function,” Chi says. “When you have a missing limb, all the information is there, but the body has no way to get it out. So we rerouted the pathway for that information so that it can be expressed.”

Categories
AI BCI Brain Crowdfunding

Crowdfunded 3-D augmented reality glasses aim to compete with Google Glass

https://www.spaceglasses.com/

For several months, after a successful Kickstarter campaign, Meta has been developing augmented reality glasses “that combine the power of a laptop and smartphone in a pair of thick Ray-Bans and a small pocket computer.”

The Meta Pro will have an i5 CPU, 4GB of RAM, 128 GB of storage, Wi-Fi 802.11n and Bluetooth 4.0 connectivity. It will cost $3,000, and the company hopes to ship by June.

Meta is hoping to compete with Google Glass, although it is not currently wireless.  The glasses have 15x the display of Google Glass, and runs 3-D instead of 2-D. Its optics are thinner, at 2mm vs 5mm, and its sensors recognize hand gestures, which makes control easier than touching the side of your face.

Categories
AI BCI Brain

Year end review of “neuromorphic” chip prototypes

http://www.technologyreview.com/featuredstory/522476/thinking-in-silicon/

MIT Technology Review today features an overview of processors that they claim are “about to narrow the gulf between artificial and natural computation—between circuits that crunch through logical operations at blistering speed and a mechanism honed by evolution to process and act on sensory input from the real world.”

Caltech’s Carver Mead pioneered “brain inspired” computing in the 1980’s, based on theoretical math and logic.  ApplySci has featured related research from The University of Zurich/ETH, DARPA, Intel, IBM, Qualcomm, and others, as well as the “deep learning” initiatives of Google and Facebook.  We anticipate and will report on advances in this space in the coming year.

Categories
AI Assistive Technologies BCI Brain

Nerve interface simulates touch in prosthetic hand

http://www.technologyreview.com/news/522086/an-artificial-hand-with-real-feelings/

Cleveland Veterans Affairs Medical Center and Case Western Reserve University researchers have developed an interface that can convey a sense of touch from 20 spots on a prosthetic hand. It directly stimulates nerve bundles, known as peripheral nerves, in the arms of patients.   Two people have been fitted with the interface to date. The implants continue to work after 18 months, which is notable because electrical interfaces to nerve tissue can gradually degrade in performance.

According to Case Western Professor Dustin Miller, who is leadning the project:  “The work opens up the possibility that prosthetic limbs could one day provide enduring and nuanced feedback to humans.”