Pooled development fund Strategic Elements (ASX: SOR) received confirmation from the University of New South Wales that its printable nanocube memory technology can be used as a printable “brain-inspired” neuromorphic computing hardware.
Strategic Elements said recently conducted research showed that its technology can “combine computing and memory in one place in a way similar to how biological neurons operate”.
This could allow the company to develop a hardware component able to store and process data without relaying information to other parts of the computer.
If successfully developed, the concept would serve as a semiconductor industry breakthrough and could potentially change the paradigm of how power, storage and processing are done in modern computing.
Work conducted at the University of New South Wales over the past few months has now been completed with early test results trickling in before a larger dataset is published in January next year.
Nanocube Memory testing involved subjecting artificial synapses to varying levels of voltage using a “memristor device” that was specifically fabricated and subjected to consecutive positive and negative current voltage “sweeps” measuring for synaptic plasticity.
“This shows the core synaptic function of synaptic plasticity of the memristor technology. Next, when electrical signals are applied to a biological synapse, the connection strength between neurons can be excited or inhibited which can be interpreted as the memorising and forgetting behaviour of the human brain,” Strategic Elements said.
With initial testing complete, the company has confirmed plans to conduct further testing at the UNSW Nanoionics laboratory in the first week of January where researchers will study features such as stability, potentiation, depression, latency and power requirements.
Mimicking the brain
Strategic Elements’ Nanocube Memory technology can purportedly store a range of values as “resistance states” rather than just traditional ones and zeroes.
This allows the hardware solution to mimic and replicate connection formed between two biological synapses, and most importantly, introduce the concept of variability to a system built on binary ones and zeroes.
The developers of nanocube memory claim that by varying the connection strength in artificial synapses, neuromorphic computing can potentially create the required architecture to achieve genuine self-learning.
However, rather than focusing on highly sophisticated artificial intelligence (AI) functionality, Strategic Elements is focusing on direct applications in the near term.
The company intends to commercialise its neuromorphic hardware technology that stores and processes data from the same component. Such a capability would greatly reduce latency and power consumption by making information transfers between the memory and processing unit redundant.
“This technology could potentially enable neuromorphic hardware to emulate high processing ability of the human brain with low energy consumption,” the company said.
Printable brain hardware
In a further advancement of its core technology, Strategic Elements is also developing what it calls “printed brain-inspired hardware” – a means of improving upon existing data processing storage based on rigid silicon semiconductors.
According to the pooled development fund, it is developing printable artificial synapses could potentially enable combined data processing and storage for printed/flexible electronics.
More specifically, Strategic Elements referenced conducting development to alleviate resource-constrained yet power-hungry IoT devices.
In a separate development project, the company is also developing a self-charging battery technology capable of producing printable memory ink.