Neurotronics

Neurotronics, also known as neural electronics, is the word combination of neuron and electronics. As its name suggests, neurotronics is a branch of electronics that integrates principles of neural-inspired concepts to tackle a wide array of real-life problems. Neural systems that are incorporated into the design of the devices, circuits and system may include brain-inspired algorithms and neural network computational models (e.g. artificial neural networks (ANNs)). Principles of the brain-inspired concepts can then be applied to the construction of electronic circuits that implements these concepts for various purposes. Electronics that can be used to embody neural systems for neurotronics can include any range of devices or technologies including but not limited to: microprocessors, robots, medical systems, and wearable technologies.
Examples
Neuromorphic Chips
One primary example of neurotronics is the neuromorphic chip. First developed by Carver Mead in the 1980s, neuromorphic chips are microprocessors configured to operate based on the principle spiking neural networks. In a spiking neural network, neurons have corresponding membrane potentials that indicate the state of the neuron. When a neuron receives enough spikes to reach the threshold potential, it will then transmit spikes to other neuron(s) in the network. In addition, different connections between various neurons have their respective weight values, so the spikes transmitted may have differing effects on the target neuron. This configuration is unlike traditional computers that are based on the von Neumann architecture, where computer instructions are executed in linear sequences of calculations. Although this approach may be useful for solving numerical problems, their applications towards pattern recognition in areas such as images or sound are quite limited.
There have been multiple neuromorphic chips modeled on biological brains in this manner, including IBM’s TrueNorth, Qualcomm’s Zeroth, European Spikey and Spinnaker chip etc. Several problems that these recently-developed neuromorphic chips have been demonstrated to handle include recognizing the composer of a musical piece, classifying handwritten digits, and recognizing traffic signs. As the capabilities of traditional computers start to reach a bottleneck, neuromorphic devices are emerging as an important alternative to allow for faster computing for a diverse range of computational tasks. Furthermore, the devices have shown significant improvement in factors such as power consumption and parallelism.
Neurorobotics
Another region in the field of neurotronics, neurorobotics, incorporates biologically-inspired models in the physical implementation of artificially intelligent systems. Divided into multiple classes of neurorobotic models such as motor control, learning and memory systems, and sensory perception, neurorobotics combines neural computing power in the construction of more dynamic robots. For example one study produced a robot based on a learning algorithm involving synaptic plasticity and neuromodulation in order to simulate the behavior of owls, while another study had a herd of small robots conduct pattern recognition on the conductive property of various batteries within the surrounding area.
Applications
Neurocomputers
Neurocomputers are computers constructed to function like a biological brain. As artificial intelligence has become an increasingly popular area of scientific research over the past few decades, researchers are looking towards the human brain for inspiration in how to build a processor that can develop learning capabilities to effectively and efficiently solve real-life problems. Neurocomputers have the capacity to execute instructions in parallel and have various interconnections between processing neurons.
Retinal Prosthesis
At the USC Institute for Biomedical Therapeutics, a retinal implant known as the Argus II has been created for individuals diagnosed with retinitis pigmentosa (RP). This device, the world’s first FDA-approved artificial retina system, is a 60-electrode retinal prosthesis. It allows the wearer some degree of sight by capturing images through the mounted camera and then transmitting the signal wirelessly to the implanted chip that allows the wearer to view the image.
Research
Neural Engineering System Design (NESD)
The Neural Engineering System Design (NESD) Program by Defense Advanced Research Projects Agency (DARPA) has aims to develop an implantable neural interface to allow greater signal communication between the brain and electronics. The planned device would act as a translator that bidirectionally converts between neural signals in the brain and bits used in electronics. The program strives to create an interface that is both biocompatible and portable.
See Also
* Neuromorphic engineering
* Neural engineering
* Physical neural network
* TrueNorth
* Neurorobotics
*

Main Menu