Scientists had previously laid the foundations of a memory circuit element, a memristor. A new research says that this circuit element can be used in neurocomputer in the future thanks to its structure that can forget memories like the human brain.
Even the most advanced computers human brain Not the same level with. Our brain, in a sense, represents the point we want to reach in computer science. As a matter of fact, researchers often take the human brain as an example to find inspiration for their work and imitate It works.
A system developed by physicists is one of the most basic features of humans. To be able to forget ability to transfer to computers. memristor The so-called circuit elements can mimic the information that human synapses carry, the ability to delete when information is not used.
It may be important for the neurocomputer of the future:
There is not much space for us to use the memristor right now. Nevertheless, this technology is not entirely useless, but rather will lead to newly developed neuro-computers. Thus, more advanced and realistic artificial intelligence can be created. Nowadays analogue neurocomputers are only in theory.
Technically, computers Unut Forget unused information in this time ” is easy to say. Human memory imitation The reason for the difficulty in experiencing is due to the structure of synapses. A memory, the stronger and more often used synapses related stronger, unused, weakens and examines. So the memory of a song is due to the very strong synapses associated with that song.
A new study envisages more useful mem- bers:
Similar studies have been done before. The problem in the first memristor systems, Moscow Institute of Physics and TechnologyAccording to Anastasia Chouprik, the behavior of the devices change over time and cause malfunctions. According to physicist own systems, 100 billion it is able to work normally even after a change in behavior.
To improve this technology, researchers use a self-polarizing material. Material used hafnium oxide is now preferred by many companies in chip production.
The real forgetting part is with hafnium oxide silicon can be achieved by tracking the interaction between. For now, these memorists are very primitive and researchers know that they need to make a lot of progress. On the other hand, the work is a new step as a very promising development. Research results ACS Applied Materials & Interfaces was published in the magazine.