In a new study, engineers have created a new method to design neuromorphic chips that mimic work
With the new method, the authors want to createa memory microcircuit that will work as efficiently as the brain, namely: consume less energy, learn more easily, adapt to the environment and learn new things. All of this was previously inaccessible to computing systems.
To collect information about the brain, the arrayNanoelectrodes were connected to a large number of neurons and their electrical signals were recorded with high sensitivity. Based on these recordings, the researchers created a map of neural connections, which shows where neurons connect to each other, as well as how strong that connection is.
Further, this information is transferred to the non-volatile memory network - this can be a regular flash memory or a new generation RRAM memory.
The article also explained how to quickly insertmaps of neural connections to the electronic memory network. A network of specially designed non-volatile memory elements shows a map of neural connections, and it starts working with the help of fixed intercellular signals.
Astronomers accidentally find two galaxies at the edge of space and time
People first descended to the bottom of the "Hell's Well": what did they find there
Chinese jet drone flies a record 20 hours at a speed of 800 km / h