Singapore researchers patent

Two lecturers from Singapore’s Ngee Ann Polytechnic-Alexei Mikhailov and Pok Yang Ming-are the first to secure a patent for a “digital gray matter” cortical chip that has the potential to replicate the ability of the human brain to retrieve, classify and process information.

With this chip, personal computers and other digital devices are able to behave like pets and think like the brain, said Mikhailov.

In November 2000, Ngee Ann Polytechnic filed for a patent application for “Digital Gray Matter” under the Patent Cooperation Treaty (PCT) at the World Intellectual Property Organization at Geneva. It secured the PCT and Singapore patents last May.

The digital neural cortex technology does not rely on template matching to learn. Instead, it instantly memorizes patterns by developing links between the columns of neurons through indexing.

By avoiding template storing and matching, the digital gray matter can accommodate a vast number of artificial neurons as it reduces the needed neural network resources.

In the past, scientists and engineers applied artificial intelligence in an effort to replicate human intelligence and other attributes such as creativity.

Artificial neural networks are similar to behavioural patterns in the way they work. Digitized data is fed into neural networks, which then learn by storing the prototype patterns and adjusting the data until a stable pattern is established. Artificial neural networks classify unknown patterns by matching them to pre-stored templates or prototypes.

With Digital Gray Matter, the possibilities are limitless as we are no longer limited to a small number of artificial neurons,” said Pok. Neural cortex technology can be used in the space industry, resource exploration, defence and communications systems. One possible application is in aviation where the technology can be used to predict pilot mental workload. Current research uses features such as electrocardiogram, cardiac, ocular and respiratory measure in an attempt to identify mental workload levels. Existing classification methods successfully classify pilot workload using flight data for a single pilot on a given day but are unsuccessful across different pilots or days.

Mikhailov said that Ngee Ann is attempting to miniaturize the cortical chip so that it can be compacted into nano-sized chips that can be incorporated into PCs and other personal devices. Pok said that Ngee Ann is currently looking for business partners to develop commercial and industrial applications based on the digital neural cortex technology.