A new computing architecture inspired by the human brain, developed by an international team of researchers, uses the functioning of a molecule to further increase ultra-fast decision-making in computers and gain flexibility, reminiscent of the plasticity of neurons in the brain .
According to a Press release from the University of Limerick, in Ireland, one of the academic centers that was part of the research, the advance creates a new type of computer architecture with considerable time and energy savings, which could have important implications in areas ranging from financial decision-making to bioinformatics.
The molecule that inspired the scientists uses the natural asymmetry in its metal-organic bonds to rapidly switch between different states, allowing it to make ultra-fast decisions. On the new device, everything is done in one place, so there is no need to keep reading or moving information
The new computer-molecular circuit is based on the fact that the computer processing unit no longer has to search for data in memory for each operation it performs: this integration and concurrency greatly saves time and energy costs. The study on which this innovation is based was recently published in the journal Nature.
Seeking to mimic brain plasticity
The researchers began their work by recognizing that the synaptic networks between neurons in the neocortex incorporate intricate logical structures that allow for extremely fast and sophisticated decision-making, far surpassing any artificial electronic analog.
Besides, the neural network in the human brain it can be dynamically reconfigured, providing flexibility and adaptability to changing environments: this feature has not yet been achieved by artificial networks and computer systems. It is not surprising, then, that many advances in these areas are made from developments inspired by the human brain.
In the new research, the scientists discovered that a simple molecule made of just 77 atoms can be used to create a computing architecture that is not only faster and cheaper: it can also achieve a level of flexibility reminiscent of the neural plasticity that characterizes our brain.
Leaving the “bottleneck” behind
In this way, innovation would be exceeding the so-called Von Neumann bottleneck. It is a limitation in computer processing as a consequence of the computer architectures that are used today. In them, the data, applications and programs that are in the memory are physically separated from the processor.
Because of this, computers must spend a significant amount of time transporting information between both systems (memory and processor). This is when a “bottleneck” is generated, since even if it has high processing speeds, the computer must spend a lot of time integrating the data in memory with the processor, remaining inactive for other purposes during that period.
Innovation based on the functioning of a molecule and inspired by the human brain overcomes this limitation, integrating memory and processing. Scientists even demonstrated experimentally that their device was capable of performing complex calculations in a single step or fragment of time, to later reprogram itself and perform another task immediately.
Researchers have explored a solution based on organic molecules, which are being used lately to create devices that can overcome the Von Neumann bottleneck, a condition that traditional systems and other options cannot achieve with any efficiency.
Now, this type of development promises the creation of ultra-fast computing devices with incredible flexibility to multitask and adapt to change, almost on the same level as the ever-amazing human brain.
Decision Trees within a Molecular Memristor. Sreetosh Goswami, Rajib Pramanick, Abhijeet Patra, Santi Prasad Rath, Martin Foltin, Ariando Ariando, Damien Thompson, T. Venkatesan, Sreebrata Goswami and R. Stanley Williams. Nature (2021).DOI:https://doi.org/10.1038/s41586-021-03748-0
Photo: Gerd Altmann and Pixabay.
Eddie is an Australian news reporter with over 9 years in the industry and has published on Forbes and tech crunch.