The brain, an extraordinary asset of the human being, performs mind boggling functions which are not possible to be achieved technologically. It has the capacity of performing functions in a jiffy (one-tenth of a second), transmitting information faster than light and is so good with numbers – calculating the distance between each step in a flight of stairs, focusing the eye to see objects at varied distances etc – that you hardly have to concentrate doing your regular work. Brain, being a marvelous creation, has been underestimated and misused by us humans. The full potential of the brain is yet to be explored and understood by us. Our brain is equivalent to 100 modern computers, and yet our brain focuses on creating newer, upgraded and exalted computers for making our lives easier and brains idler.
This new computer focuses on mimicking the human brain – its way of perceiving and transmitting information, its structure and function. The human brain transmits information throughout the body with the help of neurons which are specialized nerve cells, that aid in transmitting impulses from and to the brain. The brain is made up of about 100 billion neurons which provides a quadrillion (1 million billion) synapses. A synapse is a junction between two neurons which allows the transfer of chemical or electrical signals from one neuron to another. This is the underlying principle of this brain based computer where it uses a memristor, an electrical component that mimics the flow of calcium ions at the synapse in the brain as its resistance depends upon the flow of charge through it in the past. This new device – memristor – has led to significant advances in these neuromorphic computers (brain-inspired). The research on developing such computers trail back to many years and the research began with the help of capacitors and transistors. The disadvantages with them were that they had very little similarity with the biological system, enormous consumption of power and required a large area.
The main aim of creating neuromorphic computers was to mimic the exorbitant computing capacity and efficiency of the brain. Hence, the synaptic dynamics was revolutionized by the invention of the memristor which carried out a much real synaptic function. These memristors, like the synapses in the brain, are capable of carrying out multiple synaptic functions at a time.
THE ACTUAL MIMIC:
In our body, when an impulse reaches a synaptic junction, it causes the calcium ion channels to open up and thus there is a flood of the ions in the synapse. Once the synapse is flooded, it triggers the release of neurotransmitters (brain chemicals) which fills the space of the synapse (space between two neurons) and thus helps in transferring the impulse to the post-synapse neuron. In the neuromorphic computers which use the diffusive memristor, clusters of silver nano-particles are embedded in silicon oxynitride film which are sandwiched between electrodes. The nano-particles act as calcium ions and the film the ion channels. By combining electrical forces and heating during the application of a voltage pulse, the clusters of nano-particles break leading to their diffusion through the film which acts as an insulator and eventually forms a conductive bridge between the electrodes. Once the voltage supply is stopped, the temperature drops and the nano-particles coalesce into clusters. This device imitates short-term neuro plasticity like in biological systems. The researchers have also combined a drift memristor along with diffusion memristor for spiking-timing-dependent plasticity (long-term plasticity), which are capable of adjusting the connection strengths based on the impulse duration. These memristors are based on physical processes and hence are different from biological synapses. But using them in combination with diffusion memristors make them behave like a real synapse. Reproducing synaptic plasticity is the key to obtain brain like functioning computers. This is aimed as the brain is highly efficient, compact and best at learning and recognizing.
Though neuromorphic computers are under research, no computer can match up to the ability of the brain. Yet there are so many wonders to this creation of man. This device can be made in a size that is smaller than human synapse as its key component measures just 4 nm across whereas the average human hair has a diameter of 100,000 nm. This might make the device extremely efficient when compared to other traditional electrical components for building a neuromorphic computer. Studies say that the approach is scalable and single unit systems will be able to get to the scale of the biological synapses. But for multiple units, the device would be much bigger considering practical needs to make it work.