from the world's big
Hyperdimensional computing discovered to help AI robots create memories
New computing theory allows artificial intelligences to store memories.
- To become autonomous, robots need to perceive the world around them and move at the same time.
- Researchers create a theory of hyperdimensional computing to help store robot movement in high-dimensional vectors.
- This improvement in perception will allow artificial intelligences to create memories.
Do androids dream of electric sheep? Philip K. Dick famously wondered that in his stories that explored what it meant to be human and robot in the age of advanced and widespread artificial intelligence. We aren't quite in "Blade Runner" reality just yet, but now a team of researchers came up with a new way for robots to remember that may close the gap between robots and us for good.
For robots to be as proficient as humans in various tasks, they need to coordinate sensory data with motor capabilities. Scientists from the University of Maryland published a paper in the journal Science Robotics describing a potentially revolutionary approach to improve how AI handles sensorimotor representation using hyperdimensional computing theory.
What the researchers set out to create was a way to improve a robot's "active perception" - its ability to integrate how it perceives the world around it with how it moves in that world. As they wrote in their paper, "we find that action and perception are often kept in separated spaces," which they attribute to traditional thinking.
They proposed instead "a method of encoding actions and perceptions together into a single space that is meaningful, semantically informed, and consistent by using hyperdimensional binary vectors (HBVs). "
As their press release explains, HBVs work in very high-dimensional spaces, containing a plethora of information about different discrete items like an image or a sound or a command. These can be further grouped into sequences of discrete items and groupings of items and sequences.
By utilizing these vectors, the researchers look to keep all sensory information the robot receives in one place, essentially creating its memories. As more information gets stored, "history" vectors would be created, increasing the robot's memory content.
The scientists think that active perception and memories would make the robots better at autonomous decisions, expecting future situations and completing tasks.
The Hyperdimensional "pipeline"
This "pipeline" describes how data from a drone flight is recorded and translated into binary vectors that are integrated into memory through vector operations. This memory can then be recalled.
Credit: Perception and Robotics Group, University of Maryland.
"An active perceiver knows why it wishes to sense, then chooses what to perceive, and determines how, when and where to achieve the perception," said Aloimonos. "It selects and fixates on scenes, moments in time, and episodes. Then it aligns its mechanisms, sensors, and other components to act on what it wants to see, and selects viewpoints from which to best capture what it intends. Our hyperdimensional framework can address each of these goals."
Outside of robots, the scientists also see an application of their theories in deep learning AI methods employed in data mining and visual recognition.
To test the theory, the team employed a dynamic vision sensor (DVS) which continually captures the edges of objects in event clouds as they move by. By quickly focusing on the contours of the scene and the movement, this sensor is well-suited for autonomous navigation of robots. The data from the event clouds is stored in binary vectors, allowing the scientists to apply hyperdimensional computing.
Here’s a video of how DVS works:
The research was carried out by the computer science Ph.D. students Anton Mitrokhin and Peter Sutor, Jr., along with Cornelia Fermüller, an associate research scientist with the University of Maryland Institute for Advanced Computer Studies, as well as the computer science professor Yiannis Aloimonos. He advised Mitrokhin and Sutor.
Check out their paper "Learning sensorimotor control with neuromorphic sensors: Toward hyperdimensional active perception" in Science Robotics.
Construction of the $500 billion dollar tech city-state of the future is moving ahead.
- The futuristic megacity Neom is being built in Saudi Arabia.
- The city will be fully automated, leading in health, education and quality of life.
- It will feature an artificial moon, cloud seeding, robotic gladiators and flying taxis.
The Red Sea area where Neom will be built:
Saudi Arabia Plans Futuristic City, "Neom" (Full Promotional Video)<span style="display:block;position:relative;padding-top:56.25%;" class="rm-shortcode" data-rm-shortcode-id="c646d528d230c1bf66c75422bc4ccf6f"><iframe type="lazy-iframe" data-runner-src="https://www.youtube.com/embed/N53DzL3_BHA?rel=0" width="100%" height="auto" frameborder="0" scrolling="no" style="position:absolute;top:0;left:0;width:100%;height:100%;"></iframe></span>
A new study suggests that a century-old vaccine may reduce the severity of coronavirus cases.
- A new study finds a country's tuberculosis BCG vaccination is linked to its COVID-19 mortality rate.
- More BCG vaccinations is connected to fewer severe coronavirus cases.
- The study is preliminary and more research is needed to support the findings.
Professor Luis Escobar.
Credit: Virginia Tech
A study of the manner in which memory works turns up a surprising thing.
- Researchers have found that some basic words appear to be more memorable than others.
- Some faces are also easier to commit to memory.
- Scientists suggest that these words serve as semantic bridges when the brain is searching for a memory.
Cognitive psychologist Weizhen Xie (Zane) of the NIH's National Institute of Neurological Disorders and Stroke (NINDS) works with people who have intractable epilepsy, a form of the disorder that can't be controlled with medications. During research into the brain activity of patients, he and his colleagues discovered something odd about human memory: It appears that certain basic words are consistently more memorable than other basic words.
The research is published in Nature Human Behaviour.
An odd find
Image source: Tsekhmister/Shutterstock
Xie's team was re-analyzing memory tests of 30 epilepsy patients undertaken by Kareem Zaghloul of NINDS.
"Our goal is to find and eliminate the source of these harmful and debilitating seizures," Zaghloul said. "The monitoring period also provides a rare opportunity to record the neural activity that controls other parts of our lives. With the help of these patient volunteers we have been able to uncover some of the blueprints behind our memories."
Specifically, the participants were shown word pairs, such as "hand" and "apple." To better understand how the brain might remember such pairings, after a brief interval, participants were supplied one of the two words and asked to recall the other. Of the 300 words used in the tests, five of them proved to be five times more likely to be recalled: pig, tank, doll, pond, and door.
The scientists were perplexed that these words were so much more memorable than words like "cat," "street," "stair," "couch," and "cloud."
Intrigued, the researchers looked at a second data source from a word test taken by 2,623 healthy individuals via Amazon's Mechanical Turk and found essentially the same thing.
"We saw that some things — in this case, words — may be inherently easier for our brains to recall than others," Zaghloul said. That the Mechanical Turk results were so similar may "provide the strongest evidence to date that what we discovered about how the brain controls memory in this set of patients may also be true for people outside of the study."
Why understanding memory matters
Image source: Orawan Pattarawimonchai/Shutterstock
"Our memories play a fundamental role in who we are and how our brains work," Xie said. "However, one of the biggest challenges of studying memory is that people often remember the same things in different ways, making it difficult for researchers to compare people's performances on memory tests." He added that the search for some kind of unified theory of memory has been going on for over a century.
If a comprehensive understanding of the way memory works can be developed, the researchers say that "we can predict what people should remember in advance and understand how our brains do this, then we might be able to develop better ways to evaluate someone's overall brain health."
Image source: joob_in/Shutterstock
Xie's interest in this was piqued during a conversation with Wilma Bainbridge of University of Chicago at a Christmas party a couple of years ago. Bainbridge was, at the time, wrapping up a study of 1,000 volunteers that suggested certain faces are universally more memorable than others.
Bainbridge recalls, "Our exciting finding is that there are some images of people or places that are inherently memorable for all people, even though we have each seen different things in our lives. And if image memorability is so powerful, this means we can know in advance what people are likely to remember or forget."
Image source: Anatomography/Wikimedia
At first, the scientists suspected that the memorable words and faces were simply recalled more frequently and were thus easier to recall. They envisioned them as being akin to "highly trafficked spots connected to smaller spots representing the less memorable words." They developed a modeling program based on word frequencies found in books, new articles, and Wikipedia pages. Unfortunately, the model was unable to predict or duplicate the results they saw in their clinical experiments.
Eventually, the researchers came to suspect that the memorability of certain words was linked to the frequency with which the brain used them as semantic links between other memories, making them often-visited hubs in individuals's memory networks, and therefore places the brain jumped to early and often when retrieving memories. This idea was supported by observed activity in participants' anterior temporal lobe, a language center.
In epilepsy patients, these words were so frequently recalled that subjects often shouted them out even when they were incorrect responses to word-pair inquiries.
Modern search engines no longer simply look for raw words when resolving an inquiry: They also look for semantic — contextual and meaning — connections so that the results they present may better anticipate what it is you're looking for. Xie suggests something similar may be happening in the brain: "You know when you type words into a search engine, and it shows you a list of highly relevant guesses? It feels like the search engine is reading your mind. Well, our results suggest that the brains of the subjects in this study did something similar when they tried to recall a paired word, and we think that this may happen when we remember many of our past experiences."
He also notes that it may one day be possible to leverage individuals' apparently wired-in knowledge of their language as a fixed point against which to assess the health of their memory and brain.