Researchers from the Sandia National lab have shown that neuromorphic computers, which synthetically replicate the brain’s logic, can solve more complex problems than artificial intelligence can.
In a paper published in the journal Nature Electronics, the researchers found that neuromorphic simulations using a statistical method called “random progression” can be used in all kinds of ways, such as tracing X-rays passing through bone and soft tissue, progression of infectious diseases, passing information flowing over social networks. He detailed his findings showing that he was able to perform advanced computation.
Theoretical neuroscientist and principal investigator James Bradley Aimone of Sandia says neuromorphic computers can solve problems faster in optimal situations, using less energy than traditional computation. This new technology is of particular interest for high-performance computing, as statistical problems are not easily solved by GPUs and CPUs.
Sandia researchers used the 50-million-chip Loihi platform from Intel a year and a half ago to run their tests. While neomorphic computing isn’t meant to challenge other computational methods, its computing speed and low energy cost make it a better choice in some areas.
In addition, chips with artificial neurons are inexpensive and easy to install, unlike the challenges posed by adding qubits to quantum computers. However, transporting data on or off neurochip processors can become expensive as the amount of data increases, as the system slows down.
To overcome this hurdle, Sandia’s researchers use a small set of neurons that calculate extracted summary statistics rather than raw data.
62034
