Neuromorphics - Morphing Biology on Silicon

Neuromorphic systems are inspired by the structure,transistor.
function and plasticity of biological nervous systems.Artificial neuromorphic systems are applied in the
They are artificial neural systems that mimicareas of vision, hearing, olfaction, touch, learning,
algorithmic behavior of the biological animal systemsdecision-making, pattern recognition among others to
through efficient adaptive and intelligent controldevelop autonomous systems in robotics, vehicle
techniques.guidance and traffic control, pattern recognizers etc.
They are designed to adapt, learn from theirAs the systems mature, human parts replacements
environments, and make decisions like biologicalwould become a major application area. The
systems and not to perform better than them.fundamental principle is by observing how biological
There are no efforts to eliminate deficiencies inherentsystems perform these functions robust artificial
in biological systems.systems are designed.
This field, called neuromorphic engineering, is evolvingSo the philosophy of neuromorphic engineering is to
a new era in computing with a great promise forutilize algorithmic inspiration of biological systems to
future medicine, healthcare delivery and industry. Itengineer artificial systems. It is a kind of technology
relies on plenty of experiences which nature offerstransfer from biology to engineering that involves the
to develop functional, reliable and effective artificialunderstanding of the functions and forms of the
systems. Neuromorphic computational circuits,biological systems and consequent morphing into
designed to mimic biological neurons, are primitivessilicon chips.
based on the optical and electronic properties ofFor instance, the study of the structure of the
semiconductor materials.muscle in an animal inspires the creation of locomotive
Dr. Carver Mead, professor emeritus of Californiarobots that do not rely on heavy and power hungry
Institute of Technology (Caltech), Pasadenaservo motors. The fundamental thing is to
pioneered this field. He reasoned that biologicalunderstand how biological nerve tissues represent,
evolutionary trends over millions of years havecommunicate and process information. That would
produced organisms that engineers can study tobecome the prelude to engineer electronic devices.
develop better artificial systems. By giving senses andUnderstanding the biologically algorithms of animals
sensory-based behavior to machines, these systemsare vital and fundamental to reverse engineer the
can possibly compete with human senses and bringsbiological systems information representations and
an intersection between biology, computer sciencethen develop systems that use these
and electrical engineering.representations in their operations.
Neuromorphic systems depend on parallel collectiveThe fundamental biological unit mimicked in the design
computation, adaptation, learning and memoryof neuromorphic systems is the neurons. Animal brain
implemented locally at each stage of processingis composed of these individual units of computation,
within the artificial neurons (the computationalcalled neurons and the neurons are the elementary
elements).signaling parts of the nervous systems. Neurons,
Analog circuits, electrical circuits operated withwhich have common shape, produce electricity or
continuous varying signals, are used to implementchemical signals to communicate with other
these algorithmic processes with transistors operatedneighboring ones.
in the sub-threshold or weak inversion region (aThough these neurons are similar in shape, different
region of operation in which transistors are designedconnections with each other, muscles and receptors
to conduct current though the gate voltage is slightlyproduce different computational results in biological
lower than the minimum voltage, called thresholdsystems: locomotive control, perception, sensory
voltage, required for normal conduction to take place)processing, auditory processing etc. Neuron is made
where they exhibit exponential current-voltageof made up of input area (the dendrite) and output
characteristics and low currents.area (the axion) and is connected with other neurons
This circuit paradigm produces high density and lowby synapses.
power implementations of some functions that areSince neurons are the basic cells of the nervous
computationally intensive when compared with othersystems of all kinds of animals, building silicon neurons
paradigms (triode and saturation operational regions).(or neuromorphs) endowed with fundamental life-like
A triode region is operating transistor with gatecharacteristics, could enable the emulation or modeling
voltage above the threshold voltage but with theof the neural networks in biological nervous systems.
drain-source voltage lower than the differenceBy examining the retina for instance, artificial neurons
between the gate-source voltage and thresholdthat mimic the retinal neurons and chemistry are
voltage. For saturation region, the gate voltage is stillfabricated on silicon (most common material), gallium
above the threshold voltage but with thearsenide (GaAs) or possibly prospective organic
drain-source voltage above the difference betweensemiconductor materials.
the gate-source voltage and threshold voltage.In conclusion, it may not have changed the world, but
Transistor has four terminals: drain, gate, source andthe prospects of neuromorphics in medicine are many
bulk. Current flows between the drain and the sourceand could possibly herald the era of bio-grade artificial
when enough voltage is applied through the gate thatelectronics human organs.
enables conduction. The bulk is the body of the