Electronics based human body and brain activities are miraculous and a great feat by the creator. It provides a lesson for the pioneers of science and technology to learn and mimic these activities by artificially creating electronically-controlled devices.
The human brain is possibly the most complex entity in the Universe. It is absolutely remarkable and beautiful to contemplate, and the things you are capable of doing because of your brain are outstanding. The human brain is littered with some 100 billion nerve cells, together forming connections in tandem as each neuron is simultaneously engaged with another 1000 or so.
In total, some 20 million billion calculations per second are performed by the brain. Inspired by the operation and structure of the brain, engineers and scientists are now developing bio-inspired integrated circuit technology that mimics the neuron structure and operation of the human brain.
Although processors have gotten smaller and faster over time, only a few computers can compete with the speed and computing power of the human brain and none comes close to the organ’s energy efficiency. So some engineers want to develop electronics that mimic how the brain computes to build more powerful and efficient devices.
Scientists have developed a new prototype for electronic synapses to replicate the human brain, which could one day make neural networks incredibly clever. At present, neural networks built by research groups around the world consist of computers that are being trained using complex computer algorithms to solve complex problems and gain a deeper recognition and understanding of art and the world around us.
Memristors are the fourth class of electronic circuitry, alongside resistors, capacitors and inductors. Only confirmed to exist in 2008, memristors behave in a way similar to the synapses of neurons within the human brain. The resistance to current within a memristor is a product of currents that have previously flowed through it, meaning that the current flows easier as more current flows through.
Due to these properties, memristors hold potential for non-volatile memories, and can make computers better at understanding speech, images and the world around these.
In the past, it was very difficult to reproduce synapses because there are billions of neurons and thousands of synapses in the human brain, and the only way to get even remotely close to the power of a human brain using ordinary electronics is to utilise a gigantic amount of circuitry that would consume huge amounts of power. But then researchers managed to bring to life the concept of a memristor (using titanium oxide with two platinum electrodes), which means that conductivity in the memristor device changes depending on the charge passing through it, and connectivity can be changed from high to low.
This is similar to the way brain synapses work, and could make it possible for much more energy-efficient computer systems to be developed that have memories that retain information even when the power is off. So in the future, computers would not need to boot up and could be switched off and on like an electric light.
So now, researchers all over the world are trying to use memristors technology to develop electronic synapses that work just like the human brain, to power neural networks of the future, where computers are capable of understanding and processing the information as well, and as quickly as a human being can.
The brain’s neurons encode information in the patterns and timing of spikes of activity. That encoding is hard to model using electronic hardware because most electronics use binary (0 and 1) switches. However, researchers have combined memristors and capacitors in a way that allows for the creation of spiking output patterns.
Memristors are devices made of materials that behave as insulators until these are heated, at which point these act as conductors. Researchers paired a memristor and a capacitor in a parallel circuit and applied a current. As the voltage heated it, the memristor behaved as a resistor until it reached a critical temperature; then it became a conductor. That switching allowed for full release of the energy stored in the capacitor and, thus, mimicked the spiking behaviour of neurons.
The system, which the researchers termed a neuristor, is a very simplified model of neuron behaviour and produces a much more regular spiking pattern than a real neuron. They believe that using a different memristor and a more complicated circuit could allow them to more closely reproduce neuron behaviour on a computer chip.
A neuristor is the simplest possible device that can capture the essential property of a neuron, that is, the ability to generate a spike or impulse of activity when a threshold is exceeded. A neuristor can be thought of as a slightly-leaky balloon that receives inputs in the form of puffs of air. The only major difference is that more complex neuristors can repeat the process again and again, as long as spikes occur no faster than a certain recharge period known as the refractory period.
A neuristor uses a relatively simple electronic circuit to generate spikes. Incoming signals charge a capacitor that is placed in parallel with a memristor. When that happens, charge builds up on the capacitor by incoming spikes’ discharges, and we have a spiking neuron that comprises just two elementary circuit elements.
Details of trying to build neuristors have been largely unglamorous. In fact, most people trying to create artificial brains with neuristor-like elements have been content to do so by simulating these on regular old digital computers. A group has come up with an improved method of fabricating a neuristor made from niobium-dioxide (NbO2). For now, their prototypes are too inefficient to be used in large numbers on a single chip, but the group is developing ways to make these more compatible with current fabrication methods.
This new study does not deny long-standing efforts of the neuromorphic community to build analogue neurons into chips. It is simply the next step towards getting people on board with neuristor hardware instead of neuristor simulation.
In order to mimic the amazing human brain, we need a chip and its constituting elements behaving like the constituting elements of the brain—transistors for neurons. A transistor in some ways behaves like a synapse, acting as a signal gate. When two neurons are in connection, electrochemical reactions through neurotransmitters relay specific signals. In a real synapse, calcium ions induce chemical signaling. Transistors use oxygen ions instead, engulfed in an 80-nanometre-thick layer of samarium-nickelate crystal, which is the analogue to the synapse channel.
When a voltage is applied to the crystal, oxygen ions slip through, changing the conductive properties of the lattice and altering signal-relaying capabilities. Strength of the connection is based on the time delay in the electric signal fed into it. It is the same way for real neurons that get stronger as these relay more signals. Exploiting unusual properties in modern materials, the synaptic transistor could mark the beginning of a new kind of artificial intelligence—one embedded not in smart algorithms but in the very architecture of a computer.
Each time a neuron initiates an action and another neuron reacts, the synapse between these increases the strength of its connection; the faster the neurons spike each time, the stronger is the synaptic connection. Essentially, it memorises the action between neurons. So it does, in fact, run a bit like a neuron, in the sense that it adapts, strengthens and weakens connections according to external stimuli.
Also, opposed to traditional transistors, the creation is not restricted to the binary system of 0s and 1s and, interestingly enough, runs on non-volatile memory, which means that, even when power is interrupted, the device remembers its state. Still, it cannot form new connections like a human neuron can. By exploiting the extreme sensitivity of this material, a very small excitation allows to get a large signal, so the input energy required to drive this switching is potentially very small. That could translate into a large boost for energy efficiency. It does have a significant advantage over the human brain—these transistors can run at high temperatures exceeding 160 degree Celsius. This kind of heat typically boils the human brain.
The idea of building bio-inspired cognitive adaptive solid-state devices has been around for decades. It forms the basis for synaptic electronics, a field of research that aims to build artificial synaptic devices to emulate the computation performed by biological synapses.
Synapses dominate the architecture of the brain and are responsible for massive parallelism, structural plasticity and robustness of the brain. These are also crucial for biological computations that are needed for perception and learning. Therefore a compact nano-electronic device emulating the functions and plasticity of biological synapses will be the most important building block of brain-inspired computational systems.
Scientific interest has expanded towards building electronic systems that mimic the ability of the human brain in performing energy-efficient and fault-tolerant computation in a compact space, and research activities in this field have been growing rapidly.
Various efforts in neuromorphic engineering—a new interdisciplinary field that includes nanotechnologies—are being carried out, and the goal is to design artificial neural systems with physical architectures similar to biological nervous systems. In research findings, scientists have developed a carbon nanotube synapse with the elementary dynamic logic, learning and memory functions of a biological synapse.
Research efforts on brain-inspired computing and synaptic electronics can be understood considering the inefficiency of conventional computational systems in solving complex problems. Researchers are focusing on developing synapse-like electronic devices with directed attention to material systems that have been investigated for non-volatile memory technologies.
Interconnection scheme of phase-change memory (PCM) synapses to reach ultra-high density and compactness of brain is being developed. In the crossbar array architecture, PCM synapses lie between post-spike and pre-spike electrodes, inspired by biological synapses formed between pre-synaptic and post-synaptic neurons. A broad spectrum of device systems with programmable conductance inspired by already existing device technologies, such as PCM, resistive-change memory, conductive bridge-type memory, ferroelectric switches, carbon nanotube devices and three terminal devices or field effect transistors based devices, have been explored.
Memristors can provide a bridge for interfacing electronic circuits with nervous systems, moving us closer to the realisation of a double-layer perceptron, an element that can perform classification functions after an appropriate learning procedure. The main difficulty the research team faced was in understanding the complex electrochemical interplay that is the basis for memristive behaviour, which would give them the means to control it.
Researchers addressed this challenge by using commercial polymers and modifying their electrochemical properties at the macroscopic level. The most surprising result was that it was possible to check the electrochemical response of the device by changing the formulation of gels acting as polyelectrolytes, allowing for the study of ionic exchanges related to the biological object, which activates the electrochemical response of the conductive polymer. These developments open the way to make compatible polyaniline based devices with an interface that should be naturally-, biologically- and electrochemically-compatible and functional.
Pectin is a key ingredient for making delicious jellies and jams, not as a component for a complex hybrid device that links biological and electronic systems. But a team of scientists has built on previous work in this field using pectin with a high degree of methylation as the medium to create a new architecture of a hybrid device with a double-layered polyelectrolyte that alone drives memristive behaviour.
The team of researchers has explained the creation of the hybrid device. In this research, they applied materials generally used in the pharmaceutical and food industries in electrochemical devices. The idea of using the buffering capability of these bio-compatible materials as solid polyelectrolyte is completely innovative, and it is the first time that these bio-polymers have been used in devices based on organic polymers and in a memristive device.
The next steps are interfacing the memristor network with other living beings, for example, plants and, ultimately, the realisation of hybrid systems that can learn and perform logic/classification functions.
Using phase-change material.
A nano-sized device made from a chalcogenide performs neuron-like calculations. Electrical pulses convert a doped chalcogenide from an amorphous phase into a crystalline one. Once a certain amount has been converted, conductance of the material suddenly jumps, mimicking the firing of a neuron.
A research team has reported that nano-sized devices made from phase-change materials can mimic how neurons fire to perform certain calculations. This report shows quite concretely that we can make simple but effective hardware mimics of neurons, which could be made really small and therefore have low operating powers. The device imitates how an individual neuron integrates incoming signals from other neurons to determine when it should fire. These input signals change the electrical potential across the neuron’s membrane—some increase it, others decrease it. Once that potential passes a certain threshold, the neuron fires.
Previously, engineers have mimicked this process using combinations of capacitors and silicon transistors, which can be complex and difficult to scale down. The new work demonstrates a potentially simpler system that uses a phase-change material to play the part of a neuron’s membrane potential.
The doped chalcogenide Ge2Sb2Te5, which has been tested in conventional memory devices, can exist in two phases: glassy amorphous and crystalline. Electrical pulses slowly convert the material from amorphous to crystalline, which, in turn, changes its conductance. At a certain level of phase change, the material’s conductance suddenly jumps, and the device fires like a neuron.
Different material systems or devices have strengths in various characteristics and required characteristics for a synaptic device that will enable the development of brain-inspired systems technology that scales to biological levels of device density, parallelism and functionality. The synaptic device is a simple two-terminal nano-scale device that can reach brain-level parallelism and compactness.
Main characteristics are the ones that are directly related to parallelism, energy efficiency and fault tolerance. The synaptic device should emulate plasticity by implementing an analogue-like transition between different conductance states with very low energy consumption per synaptic event. Exact performance metrics depend on the target application and the scale of the system design.
Although most of the work focuses on some specific applications for synaptic devices, the field has the potential to be utilised in a broad range of computing applications, especially the ones at the intersection of sensing and computation, where real-time and parallel processing of large-scale data is crucial.
There is an enormous opportunity to completely rethink the design of computational systems in order to gain orders of magnitude of improvement in computational efficiency through inspiration from the biological brain. More interactions among different research disciplines—devices, circuits, architecture and computing—can further cultivate the synaptic electronics field and help define more targeted research paths for the future.
Parallel Distributed Processing, Volume 1
Explorations in the Microstructure of Cognition: What is new here!