A new computer prototype called a "memcomputer" works by mimicking the human brain, and could one day perform notoriously complex tasks like breaking codes, scientists say.

These new, brain-inspired computing devices also could help neuroscientists better understand the workings of the human brain, researchers say.

In a conventional microchip, the processor, which executes computations, and the memory, which stores data, are separate components. This constant relaying of data between the processor and the memory consumes time and energy, thus limiting the performance of standard computers.

In contrast, Massimiliano Di Ventra, a theoretical physicist at the University of California, San Diego, and his colleagues are building "memcomputers," made up of "memprocessors," that both process and store data. This setup mimics the neurons that make up the human brain, with each neuron serving as both the processor and the memory. The building blocks of memcomputers were first theoretically predicted in the 1970s, but they were manufactured for the first time in 2008. [Super-Intelligent Machines: 7 Robotic Futures]

Now, Di Ventra and his colleagues have built a prototype memcomputer they say can efficiently solve one type of notoriously difficult computational problem. Moreover, they built their memcomputer from standard microelectronics.

"These machines can be built with available technology," Di Ventra told Live Science.

The scientists investigated a class of problems known as NP-complete. With this type of problem, a person may be able to quickly confirm whether any given solution may or may not work but can't quickly find the best solution to it.

One example of such a conundrum is the "traveling salesman problem," in which someone is given a list of cities and is asked to find the shortest possible route from a city that visits every other city exactly once and returns to the starting city. Although someone may be able to quickly find out whether a route gets to all of the cities and does not go to any city more than once, verifying whether this route is the shortest involves trying every single combination — a brute-force strategy that grows vastly more complex as the number of cities increases.

The memprocessors in a memcomputer can work collectively and simultaneously to find every possible solution to such conundrums.

The new memcomputer solves the NP-complete version of what is called the subset sum problem. In this problem, one is given a set of integers — whole numbers such as 1 and negative 1, but not fractions such as 1/2 — and must find if there is a subset of those integers whose sum is zero.

"If we work with a different paradigm of computation, those problems that are notoriously difficult to solve with current computers can be solved more efficiently with memcomputers," Di Ventra said.

But solving this type of problem is just one advantage these computers have over traditional computers. "In addition, we would like to understand if what we learn from memcomputing could teach us something about the operation of the brain," Di Ventra said.

To solve NP-complete problems, scientists are also pursuing a different strategy involving quantum computers, which use components known as qubits to investigate every possible solution to a problem simultaneously. However, quantum computers have limitations — for instance, they usually operate at extremely low temperatures.

In contrast, memcomputers "can be built with standard technology and operate at room temperature," Di Ventra said. In addition, memcomputers could tackle problems that scientists are exploring with quantum computers, such as code breaking.

However, the new memcomputer does have a major limitation: It is difficult to scale this proof-of-concept version up to a multitude of memprocessors, Di Ventra said. The way the system encodes data makes it vulnerable to random fluctuations that can introduce errors, and a large-scale version would require error-correcting codes that would make this system more complex and potentially too cumbersome to work quickly, he added.

Still, Di Ventra said it should be possible to build memcomputers that encode data in a different way. This would make them less susceptible to such problems, and hence scalable to a very large number of memprocessors.

The scientists detailed their findings online July 3 in the journal Science Advances.

Source: http://m.livescience.com/51447-brain-computer-may-solve-complex-math-problems.html

These new, brain-inspired computing devices also could help neuroscientists better understand the workings of the human brain, researchers say.

In a conventional microchip, the processor, which executes computations, and the memory, which stores data, are separate components. This constant relaying of data between the processor and the memory consumes time and energy, thus limiting the performance of standard computers.

In contrast, Massimiliano Di Ventra, a theoretical physicist at the University of California, San Diego, and his colleagues are building "memcomputers," made up of "memprocessors," that both process and store data. This setup mimics the neurons that make up the human brain, with each neuron serving as both the processor and the memory. The building blocks of memcomputers were first theoretically predicted in the 1970s, but they were manufactured for the first time in 2008. [Super-Intelligent Machines: 7 Robotic Futures]

Now, Di Ventra and his colleagues have built a prototype memcomputer they say can efficiently solve one type of notoriously difficult computational problem. Moreover, they built their memcomputer from standard microelectronics.

"These machines can be built with available technology," Di Ventra told Live Science.

The scientists investigated a class of problems known as NP-complete. With this type of problem, a person may be able to quickly confirm whether any given solution may or may not work but can't quickly find the best solution to it.

One example of such a conundrum is the "traveling salesman problem," in which someone is given a list of cities and is asked to find the shortest possible route from a city that visits every other city exactly once and returns to the starting city. Although someone may be able to quickly find out whether a route gets to all of the cities and does not go to any city more than once, verifying whether this route is the shortest involves trying every single combination — a brute-force strategy that grows vastly more complex as the number of cities increases.

The memprocessors in a memcomputer can work collectively and simultaneously to find every possible solution to such conundrums.

The new memcomputer solves the NP-complete version of what is called the subset sum problem. In this problem, one is given a set of integers — whole numbers such as 1 and negative 1, but not fractions such as 1/2 — and must find if there is a subset of those integers whose sum is zero.

"If we work with a different paradigm of computation, those problems that are notoriously difficult to solve with current computers can be solved more efficiently with memcomputers," Di Ventra said.

But solving this type of problem is just one advantage these computers have over traditional computers. "In addition, we would like to understand if what we learn from memcomputing could teach us something about the operation of the brain," Di Ventra said.

**Quantum computing**To solve NP-complete problems, scientists are also pursuing a different strategy involving quantum computers, which use components known as qubits to investigate every possible solution to a problem simultaneously. However, quantum computers have limitations — for instance, they usually operate at extremely low temperatures.

In contrast, memcomputers "can be built with standard technology and operate at room temperature," Di Ventra said. In addition, memcomputers could tackle problems that scientists are exploring with quantum computers, such as code breaking.

However, the new memcomputer does have a major limitation: It is difficult to scale this proof-of-concept version up to a multitude of memprocessors, Di Ventra said. The way the system encodes data makes it vulnerable to random fluctuations that can introduce errors, and a large-scale version would require error-correcting codes that would make this system more complex and potentially too cumbersome to work quickly, he added.

Still, Di Ventra said it should be possible to build memcomputers that encode data in a different way. This would make them less susceptible to such problems, and hence scalable to a very large number of memprocessors.

The scientists detailed their findings online July 3 in the journal Science Advances.

Source: http://m.livescience.com/51447-brain-computer-may-solve-complex-math-problems.html

## No comments:

Post a Comment