Neuromorphic Computing: A Revolutionary Approach to Computing

This article discusses the concept of Neuromorphic Computing, which is a revolutionary approach to computing that aims to mimic the structure and functionality of the human brain. It explores the history and concept of Neuromorphic Computing, the differences between Von Neumann and Neuromorphic architectures, the development of Neuromorphic chips, and the applications and benefits of Neuromorphic Computing in various fields. The article also provides insights into the future perspectives of this technology.

Running head: NEUROMORPHIC COMPUTING
Neuromorphic Computing
Name of the Student
Name of the University
Authors note

1
NEUROMORPHIC COMPUTING
Table of Contents
Introduction................................................................................................................................2
Discussion..................................................................................................................................3
History and Concept:..............................................................................................................3
Von Neumann versus Neuromorphic.....................................................................................4
Neuromorphic Chip:...............................................................................................................5
Neuromorphic and Artificial Intelligence..................................................................................5
Neuromorphic computing in Space:...........................................................................................6
Future Perspectives....................................................................................................................7
Application and benefits of Neuromorphic computing..............................................................8
Conclusion................................................................................................................................10
References................................................................................................................................11

2
NEUROMORPHIC COMPUTING
Introduction
Neuromorphic computing is also referred as Neuromorphic engineering.
Neuromorphic computing has been a different approach than normal conventional
computing. The model of Neuromorphic was essential in the industry in order to consume
less energy and something that is has an enhanced version of computing. The industry
demanded a kind of technology that is more versatile and adaptable. Hence led the
development of Neuromorphic models (Adam et al., 2017). The aim of Neuromorphic
engineering was to encourage high and vibrant results that other traditional architectures
failed to. With the help of Neuromorphic computing the introduction of Neuromorphic chips
is possible. It is considered to be a great replacement for microprocessors. The Neuromorphic
chips are complex because the architecture is derived from the neurons of human brain. The
idea behind the creation of this chip was to easily achieve the trending technical goals. This
paper outlines the concept of Neuromorphic engineering and the development of
Neuromorphic models (Boybat et al., 2018). It also explains the application and benefits in
context of modern era and technologies. Neuromorphic computing is the enhanced version of
computing that can replace the numerical methods of computing. Hence to develop a unique
system it requires new architectures. The researchers started exploring the design of
Neuromorphic computing in 1990. They found that the analog computers which has been
working with the idea of nervous system of human was more powerful and efficient. The
traditional system lacked certain benefits like speech and image recognition (Brunner et al.,
2017). Therefore, the concept of using biological unit in computing technology has led to
additional improvement in the world of technologies.

3
NEUROMORPHIC COMPUTING
Discussion
History and Concept:
The idea of Neuromorphic computing was initiated by Carver Mead in 1990. He
wanted to introduce a system that is faster in computation but requires less power (Grübl &
Baumbach, 2017). It specifies of an input neuron which generates pulse electric signals and
output Synapse that passes state to other neuron. Memristor generates neuroplasticity for
Neuromorphic computing. Neuromorphic computing was developed to compete with the
tradition architecture. The architecture of Neuromorphic computing was based upon human
brain. It was modeled with neurons and synapses (Burr et al., 2017). Neuromorphic
computing is one of the most outstanding invention oin the digital era. The concept can be
used to develop intelligent computers with high quality performance and mantainence. The
tool was a combination of machine learning and artificial intelligence with certain idea from
neuroscience. It includes components such as hardware, software, neuroscience and
computational science. It indicates the neural structure of human brain. The concept was
together inspired by biology, physics, mathematics and computer science. They are modeled
on the neurons which signals through nodes that delivers pulses in which the information is
carried by the frequency similar to the functioning of a nervous system in a human body. The
cell neural networking is implemented in Neuromorphic computing. It enables synapses to
get information and to identify patterns. It recognizes the texts without performing any
programming. It focuses on matching the traits of human brain. Carver Mead wanted to
create a computing model that works exactly like a human brain. Therefore he created
Neuromorphic computing which was flexible and could understand the unstructured stimuli.
The use of Spiking neural networks (SNNs) helps in generating original neural network of the
human brain. This concept has been inspired by the human brain (Esser et al., 2015). The
Neuromorphic computing is different from an artificial neuron network which is programmed

4
NEUROMORPHIC COMPUTING
to run on a normal computer that emulates the logic of how a human brain thinks. The use of
silicon makes it hard to control the current between artificial neurons. Therefore, a new
design from an MIT team used different material such as single crystalline silicon and silicon
germanium layered on the top of one another. The Korean team is still investigating on other
materials. They used tantalum oxide to give them control over the flow of ions which is also
more durable (Hassabis et al., 2017). The Colorado team is implementing magnets to control
the way computer neurons communicate.
Von Neumann versus Neuromorphic
The system architecture of Von Neumann was based on resistors, inductors, capacitors and
transistors. Though they have characteristics like speed and size, but they lack certain
functionality of less energy consumption and fault tolerance. While the Neuromorphic
architectures are made up of neurons, synapses, and axons of a nervous system. The design of
Neuromorphic is complex and complicated as compared to that of Neumann’s devices. The
system of Von Neumann is divided into various which performs specific functions of the
system such as Memory unit, Control unit, central processing unit and arithmetic logical unit
(Lee et al., 2018). The limitation of Neumann Bottleneck has made it difficult to evolve.
Whereas the Neumann architecture uses brain as a working system which benefits in
generating efficient energy. It is able to cope up the environment and performs complicated
programming. The traditional systems used to perform complex arithmetic by multiplying the
inputs on the basis of their arrival and weight. The result is then performed under
normalization technique and then to the network. While the Neuromorphic nodes used to
perform pulse modulated function using spike function. It has several inputs but single
output. The input is known as dendrites and output is known as axon. The Neuromorphic
computing and neural networks can work together because the progress is made in both the
fields. The Neuromorphic hardware will probably be the best option to rum the neural

5
NEUROMORPHIC COMPUTING
networks. Traditional computers used to think in binary which makes it a very rigid structure
whereas the works in a flexible manner. Instead of using electrical signals, designer if these
new chips intended to make the computer’s neurons communicate to each other in the same
way as the biological neurons do. The ability to transmit a gradient of understanding from
neuron to neuron and to make them work together simultaneously has eventually made the
Neuromorphic chips more energy efficient than the normal computers especially for complex
tasks (Malhotra, 2018).
Neuromorphic Chip:
The design of Neuromorphic chip is inspired by biological brain. It the electronic
model of neural networks in a biological brain. The chip functions by interpreting sensory
data and responding. The chips can be used for tasks which requires to encode large data
within less time. It carries huge amount of information in a single burst. The machine
learning and artificial intelligence can use these chips to effectively communicate with the
world exactly like a human being. The continuous development of the Neuromorphic chip
will help in generating excellent result in the field if medical science. It can provide warning
signs by early detection of any issue in the patient’s health and gives the required treatment.
The Neuromorphic chips are modeled after brain of human which deliberately performs tasks
effectively which Von Neumann system would require great time and effort. It can process
multiple tasks (Pantazi et al., 2016)
Though the Neuromorphic chip has not been used massively in the present era but it is
estimated that an explosive increase in this technology will take place in the next decade. The
estimated cost is $1.2 billion at present which will explode to become $1.4 billion in 2022.

6
NEUROMORPHIC COMPUTING
Neuromorphic and Artificial Intelligence
The rapid utilization of artificial intelligence and machine learning systems has
increased the technological advancement in computers as well as mobile phones. Machine
learning helps in evaluating large samples of input data without being programmed to
perform those tasks. The use of artificial neural network has successfully created a
benchmark for the other technological network. The main concern of Neuromorphic chips
that it consumes less power to execute the artificial intelligence algorithm had made it a
powerful and robust tool (Rajendran & Alibart, 2016). A single Neuromorphic chip consists
of transistors that is five times greater than normal Intel processor but still uses only 70 MW
of power. The use of ANNs has proved to be helpful in performing number of tasks such as
pattern recognition, speech recognition, image recognition, medical diagnosis and computer
vision. The one the most happening benefits of ANN is cyber security. The artificial
intelligence uses ANN to automatically defend against any cyber-attacks. Artificial
intelligence potentially identifies the hidden patterns even if it is in the form of encrypted
packets, and detects any malicious activity inside the network. Some of the devices such as
mobile phones are unable to bear enough memory space and power to operate the neural
models, so specific applications has been implemented for the same. Moreover, the
introduction of SNN has helped in producing accurate real time learning environment. It can
be used to render images because of its threshold functionality in spite of using mathematical
logics (Schuman et al., 2017). The number of methods of machine learning technique such as
logical programing, decision tree and artificial neural network has gain complete success in
the market. The execution of ANN requires to imitate the exact procedure of how a neuron
functions inside the human brain. (Shanmuganathan, 2016).With the help of artificial
intelligence it builds an interconnected network of artificial neuron. Back propagation helps
to feed the result back to system.

7
NEUROMORPHIC COMPUTING
Neuromorphic computing in Space:
Neuromorphic computing can be a game changer implementation for space
applications. The success of any mission depends upon the quickness and amount of
information received from sources. The motive of every application is to increase the speed
and decrease the consumption of power and energy. To execute any space mission, a quality
performance and a reliable computing device is necessary. A device that can run on extreme
temperate, low power, bad connection and high radiation is need. Hence the Neuromorphic
computing devices has the capability to fulfil all these types of necessities (Sharbati et al.,
2018). The potential to manage and monitor the security of spacecraft is included in
Neuromorphic chips. It can detect any intrusion in the spacecraft.
Future Perspectives
The implementation of Neuromorphic computing will have certain positive affect in the
future of the computing world. The advances in the architecture of Neuromorphic systems are
all working towards to place where the neurons on the chips can learn as they compute.
Software neural network have been able to do this for a while but it is a new advancement for
physical Neuromorphic devices and the experiments are showing promising results. It can be
used in laptops and mobile phones with the help of Neuromorphic chips in them which
consume ultra-low power to work (Suri, 2017). It can be used in medical science to note vital
responses and signs to the medical treatment. The smartphones will be able to detect what is
to be done by the owner and will generate alerts accordingly. The smart watches will work
and give result with the blood flow in human system. It will use human body as an interface
to give potential result. Hence the Neuromorphic computing has a promising future ahead
that will enhance the capacity to learn and reason.

8
NEUROMORPHIC COMPUTING
 The Neuromorphic chip will create a glass for the blind people which will help in
identifying the objects with the help of the sensory nerves in the glasses.
 The smart phones will adapt itself according to its user. It will automatically enter into
do not disturb mode as soon as the user goes to the bed.
Application and benefits of Neuromorphic computing
The impact of this technology will probably help the neuroscientists to fill the gaps in
understanding brain. Neuromorphic computers can also alter the communication of other
devices with intelligent machines. (van De Burgt et al., 2017).The introduction of
Neuromorphic sensors in cars, smartphones and robots will help the people top get a sense of
direction. The machines will be able to work as efficient as the human brain. The devices will
respond fast and quick exactly like a brain. This could also help in creating machines that will
enable it to learn, recall and mitigate the problems of humans by providing better decisions.
In High Blood Pressure (HBP), the use of Neuromorphic computing will help in executing a
large scale exclusive Neuromorphic machine based on 2 principle. The SpiNNaker machine
situated in Manchester and BrainScaleS machine situated in Heidelberg are integrated
together to create a HBP software which performs data analysis of the system and operations.
SpiNNaker machine is built from numerical models with the help of ARM architecture.
BrainScaleS machine is build by emulating neurons and synapse of the human brain. One of
the exclusive feature offered by Neuromorphic machines is speed (Wicker & Pashmakov,
2017). The speed with which SpiNNaker and BrainScaleS is executed is ten thousand times
the real time. HPB has been the only system that has helped in development and learning of
neuroscience and cognitive computing. Pattern Recognition technology (PRT) is an
organizational industry that is facilitating load, wind generation and price services for energy

9
NEUROMORPHIC COMPUTING
industry. This forecasting service is based on artificial neurons and genetic algorithm (van De
Burgt et al., 2018). It has been using the idea of Neuromorphic computing that has made the
industry earn good reputation and customer support. Brainschip AI technology requires less
power and can work efficiently in difficult environment. It can detect any anomaly and helps
in providing cyber security. Silicon Brain is the technology that is introduced by Intel after
studying Neuromorphic chips. It has helped humans to understand their functionality if brain
more clearly. This technology has emerged out to be increasingly growing technology in the
recent market. This chip created by Intel, could learn and adapt thing because of the self
learning concept. The other machine learning system required excessive training and data
usage which was complex and complicated. The chips reduces workloads.
Benefits:
 Economical: the use of Neuromorphic computers in economy will give positive
impact to the financial marketing and forecasting. It will help in monitoring the
agriculture business.
 Environmental: the efficient use of this system will enhance the data mining process
and to monitor process and pattern recognition.
 Social: the positive impact in the medical science and will improve the health issues
of humans. The concept of neurons in this system will create therapies and artificial
retina that will help in diagnosing the health issues (Wang et al., 2017)
 Domestic: The implementation of intelligent cameras with the capability to analyze
real time video will help in serving domestic politics.
 Science: In science, the Neuromorphic computers will help in implementing self-
learning machines that will help the humans to perform their job and research more
efficiently.

10
NEUROMORPHIC COMPUTING
 Military: the Neuromorphic chips that are introduced in robots can be sent to combat
zones.
Conclusion
Therefore, the use of Neuromorphic processors will probably create a new era of
mobile devices and sensory objects that will be able to perform any task intelligently and
independently. Thus, this technology will help to deploy faster and smarter satellites which
will provide significant advantages to the researches. However the concept of Neuromorphic
computing is not so popular but it will emerge out to be greatest invention in the field of
Artificial intelligence. Hence, the paper concludes that since this technology is inspired by
the theory of neural networking, the Neuromorphic computing will become a brain that will
never die.

11
NEUROMORPHIC COMPUTING
References
Adam, G. C., Hoskins, B. D., Prezioso, M., Merrikh-Bayat, F., Chakrabarti, B., & Strukov,
D. B. (2017). 3-D memristor crossbars for analog and neuromorphic computing
applications. IEEE Transactions on Electron Devices, 64(1), 312-318.
Boybat, I., Le Gallo, M., Nandakumar, S. R., Moraitis, T., Parnell, T., Tuma, T., ... &
Eleftheriou, E. (2018). Neuromorphic computing with multi-memristive
synapses. Nature communications, 9(1), 2514.
Brunner, D., Jacquot, M., Fischer, I., & Lager, L. (2017, September). Photonic networks for
Neuromorphic Computing. In Frontiers in Optics (pp. FTh2E-3). Optical Society of
America.
Burr, G. W., Shelby, R. M., Sebastian, A., Kim, S., Kim, S., Sidler, S., ... & Sanches, L. L.
(2017). Neuromorphic computing using non-volatile memory. Advances in Physics:
X, 2(1), 89-124.
Esser, S. K., Appuswamy, R., Merolla, P., Arthur, J. V., & Modha, D. S. (2015).
Backpropagation for energy-efficient neuromorphic computing. In Advances in
Neural Information Processing Systems (pp. 1117-1125).
Grübl, A., & Baumbach, A. (2017). F09/F10 Neuromorphic Computing.
Hassabis, D., Kumaran, D., Summerfield, C., & Botvinick, M. (2017). Neuroscience-inspired
artificial intelligence. Neuron, 95(2), 245-258.
Lee, H. E., Park, J. H., Kim, T. J., Im, D., Shin, J. H., Kim, D. H., ... & Lee, K. J. (2018).
Novel electronics for flexible and neuromorphic computing. Advanced Functional
Materials, 28(32), 1801690.

12
NEUROMORPHIC COMPUTING
Malhotra, Y. (2018). Cognitive Computing for Anticipatory Risk Analytics in Intelligence,
Surveillance, & Reconnaissance (ISR): Model Risk Management in Artificial
Intelligence & Machine Learning (Presentation Slides). Surveillance, &
Reconnaissance (ISR): Model Risk Management in Artificial Intelligence & Machine
Learning (Presentation Slides)(January 28, 2018).
Pantazi, A., Woźniak, S., Tuma, T., & Eleftheriou, E. (2016). All-memristive neuromorphic
computing with level-tuned neurons. Nanotechnology, 27(35), 355205.
Rajendran, B., & Alibart, F. (2016). Neuromorphic computing based on emerging memory
technologies. IEEE Journal on Emerging and Selected Topics in Circuits and
Systems, 6(2), 198-211.
Schuman, C. D., Potok, T. E., Patton, R. M., Birdwell, J. D., Dean, M. E., Rose, G. S., &
Plank, J. S. (2017). A survey of neuromorphic computing and neural networks in
hardware. arXiv preprint arXiv:1705.06963.
Shanmuganathan, S. (2016). Artificial neural network modelling: An introduction.
In Artificial neural network modelling (pp. 1-14). Springer, Cham.
Sharbati, M. T., Du, Y., Torres, J., Ardolino, N. D., Yun, M., & Xiong, F. (2018). Low‐
Power, Electrochemically Tunable Graphene Synapses for Neuromorphic
Computing. Advanced Materials, 30(36), 1802353.
Suri, M. (Ed.). (2017). Advances in Neuromorphic Hardware Exploiting Emerging
Nanoscale Devices (Vol. 31). New York: Springer.
van de Burgt, Y., Lubberman, E., Fuller, E. J., Keene, S. T., Faria, G. C., Agarwal, S., ... &
Salleo, A. (2017). A non-volatile organic electrochemical device as a low-voltage
artificial synapse for neuromorphic computing. Nature materials, 16(4), 414.

13
NEUROMORPHIC COMPUTING
van De Burgt, Y., Melianas, A., Keene, S. T., Malliaras, G., & Salleo, A. (2018). Organic
electronics for neuromorphic computing. Nature Electronics, 1.
Wang, Z., Joshi, S., Savel’ev, S. E., Jiang, H., Midya, R., Lin, P., ... & Wu, Q. (2017).
Memristors with diffusive dynamics as synaptic emulators for neuromorphic
computing. Nature materials, 16(1), 101.
Wicker, G., & Pashmakov, B. (2017, August). Potential application of time dependent
threshold switching in neuromorphic computing. In 2017 17th Non-Volatile Memory
Technology Symposium (NVMTS) (pp. 1-2). IEEE.