B.M.

S COLLEGE OF ENGINEERING
Department of Electronics and Communication Engineering
Bull Temple Road, Bangalore - 560 019

Research Methodology and IPR – 23ES6AERMI

AAT Report

on

“Neuromorphic Computing: Bridging the Gap Between AI and

Brain”

Submitted by

Name of the Student USN

Saharsha 1BM22EC211
Sarvash singh 1BM22EC222

Dr. Rajanikanth
Assistant Professor

1|Page
 TABLE OF CONTENTS:

SI NO: CONTENT: PG NO:

01) Abstract 03

02) Introduction 04

03) Neuromorphic Hardware 05

Components

04) VLSI For Neuromorphic 08

Computing

05) Applications Of Neuromorphic 10

Computing

06) Challenges And Future Directions 11

07) Conclusion 12

08) References 13

2|Page
ABSTRACT
Neuromorphic computing is emerging as a transformative computational
paradigm that seeks to emulate the structure, dynamics, and functionality of the
human brain. In contrast to traditional von Neumann architectures—which
separate memory and processing units and operate in a sequential, clock-driven
manner—neuromorphic systems utilize massively parallel, distributed
architectures inspired by biological neural networks. These systems aim to
replicate key cognitive processes such as learning, adaptation, and perception
through brain-inspired mechanisms.

This paper provides a comprehensive overview of the current landscape of

neuromorphic computing, with a focus on the enabling hardware technologies
and computational models that make this paradigm possible. In particular, we
explore the role of memristors—non-volatile resistive devices that naturally
mimic synaptic plasticity—and Spiking Neural Networks (SNNs), which more
closely resemble the asynchronous and sparse communication of biological
neurons compared to traditional artificial neural networks.

We also discuss event-driven computation, a hallmark of neuromorphic

systems, where processing is triggered by input events rather than continuous
clock cycles, leading to significant reductions in power consumption and latency.
The implementation of such systems heavily relies on VLSI design
methodologies, which facilitate the integration of large-scale, low-power
neuromorphic circuits.

Additionally, the paper reviews recent advances in neuromorphic hardware

platforms, evaluates their performance and scalability, and highlights real-world
applications ranging from edge AI and robotics to biomedical signal processing.
Finally, we address the key challenges that remain, including the standardization
of neuromorphic architectures, training mechanisms for SNNs, and bridging the
gap between software tools and hardware design.

3|Page
INTRODUCTION:
Neuromorphic computing refers to a new class of computational systems
designed to emulate the structure, functionality, and processes of the human brain.
These systems aim to bridge the gap between traditional computing architectures
and biological intelligence by using specialized hardware such as spiking neural
networks (SNNs) and memristors. By mimicking neural activity and synaptic
behavior, neuromorphic systems are capable of offering high computational
efficiency, low power consumption, and real-time processing— attributes
intrinsic to human cognition.

The development of neuromorphic systems has been significantly driven by

progress in brain-inspired hardware design. Unlike traditional systems based on
the von Neumann architecture, which rely on clock-driven, sequential
execution, neuromorphic systems operate asynchronously and adopt an event-
driven approach. This enables them to handle complex, real-world tasks more
efficiently, especially those involving pattern recognition, sensory data
processing, and adaptive learning.

Inspired by the human brain, neuromorphic systems replicate how biological

neurons and synapses process vast amounts of data simultaneously. They are
particularly effective in applications that require real-time responsiveness and
energy efficiency, such as robotics, healthcare, and artificial intelligence (AI).
These areas benefit greatly from neuromorphic computing’s capability to adapt
and learn from the environment—just like the human brain.

4|Page
NEUROMORPHIC HARDWARE COMPONENTS:
 Memristors and Their Role in Neuromorphic Computing

Memristors are a type of resistive switching device whose resistance can change
based on the voltage applied across them. One of their key advantages is their
non-volatile memory capability, meaning they can retain stored information
even after power is turned off. This characteristic makes them ideal for mimicking
biological synapses in neuromorphic computing systems.

In a neuromorphic architecture, memristors not only store data but also act as
artificial synapses, connecting neuron-like components and facilitating the
transmission of signals. Their ability to both process and store information in
the same physical location is a major advantage over traditional memory and
logic elements like transistors and capacitors, which separate computation and
memory.

Furthermore, memristors offer benefits such as low power consumption, high

switching speed, and scalability, making them a promising building block for
developing efficient, brain-inspired hardware systems.

5|Page
  Spiking Neural Networks (SNNs)

Spiking Neural Networks (SNNs) are a biologically-inspired model of neural

networks that represent a significant leap forward in neuromorphic computing.
Unlike conventional Artificial Neural Networks (ANNs) that use continuous
values to represent neuron activations, SNNs transmit data using discrete
electrical pulses or "spikes", closely resembling the way neurons in the human
brain communicate.

One of the key features of SNNs is their event-driven nature, meaning neurons
only become active and communicate when a specific threshold is reached. This
leads to sparse and asynchronous computation, which significantly reduces
power consumption compared to continuously active neural networks.

SNNs are particularly well-suited for real-time tasks such as adaptive learning,
pattern recognition, and decision-making under dynamic conditions. By
accurately modeling biological behaviors, SNNs offer an efficient and scalable
path for implementing intelligent computing systems.

6|Page
  Event-Driven Computation

Event-driven computation lies at the heart of neuromorphic computing. Unlike

traditional processors that operate based on a continuous clock signal (clock-
driven systems), event-driven systems perform computation only when an input
event occurs—such as a spike from a neuron.

This mechanism greatly reduces the amount of idle processing and unnecessary
data transfer, leading to higher energy efficiency and faster response times. In
neuromorphic systems, event-driven computation enables real-time
responsiveness to environmental changes, which is critical in applications like
robotics, autonomous vehicles, and sensory processing.

Moreover, this approach aligns well with how biological systems function,
where neurons are often dormant until stimulated. As a result, event-driven
neuromorphic architectures are especially suitable for embedded systems and
low-power edge devices, where power efficiency and speed are of utmost
importance.

7|Page
VLSI FOR NEUROMORPHIC COMPUTING:
 Brain-Inspired VLSI Design

Brain-inspired Very Large Scale Integration (VLSI) design focuses on

embedding neuromorphic principles directly into hardware. The goal is to
develop circuits that emulate the operation and architecture of the human brain—
particularly the behavior of spiking neurons and synapses. Leveraging VLSI
technology allows for the creation of compact, power-efficient, and high-
performance circuits capable of processing large volumes of data in parallel,
much like biological neural networks.

Recent innovations, particularly in memristor-based VLSI design, have

significantly enhanced the scalability, efficiency, and speed of neuromorphic
systems. These specialized chips are built to support the massive parallelism
found in biological systems by integrating both memory and processing units
within a single architecture. This enables complex computations to be performed
simultaneously, reducing latency and power consumption.

Furthermore, advances in emerging materials and novel circuit design techniques

are pushing the frontiers of neuromorphic VLSI. Devices such as memristors

8|Page
offer the ability to store and process data simultaneously, leading to more
brain-like behavior in hardware systems.

 Challenges in VLSI Implementation

While VLSI design is fundamental to realizing practical neuromorphic systems,
it comes with a range of challenges. One major hurdle is the integration of
memristive devices with conventional CMOS (Complementary Metal-
Oxide-Semiconductor) technology. CMOS is the backbone of current digital
circuit design, but it was not originally developed for brain-like computing
models such as spiking neural networks (SNNs) or event-driven architectures.

This mismatch creates issues in circuit compatibility, performance consistency,

and fabrication complexity. Moreover, achieving scalability—designing systems
that can grow in complexity without excessive power or area overhead—remains
a significant difficulty. Ensuring that neuromorphic VLSI circuits can scale while
preserving energy efficiency, robustness, and real-time responsiveness is
essential for the broader adoption of this technology.

Addressing these challenges requires interdisciplinary efforts in materials

science, circuit design, and computer architecture. Researchers are currently
exploring hybrid approaches that combine the reliability of CMOS with the
adaptive, energy-efficient characteristics of memristive devices.

9|Page
APPLICATIONS OF NEUROMORPHIC COMPUTING:
 Autonomous Systems

Neuromorphic computing plays a critical role in advancing autonomous

technologies such as self-driving vehicles, drones, and intelligent robots. These
systems require real-time sensory processing, rapid decision-making, and the
ability to adapt to changing environments—all of which are core strengths of
neuromorphic architectures.

By imitating the parallel and asynchronous data processing capabilities of the

human brain, neuromorphic systems excel in performing complex tasks like
obstacle detection, path planning, and environmental perception. For
instance, a neuromorphic robot can perceive its surroundings through sensors,
process visual or spatial information almost instantaneously, and make intelligent
decisions on how to navigate—all with minimal power usage.

The adaptability and learning capabilities of neuromorphic chips also allow

autonomous machines to improve their performance over time, making them
suitable for long-term deployment in dynamic and unpredictable real-world
environments.

 Healthcare Applications

Healthcare is another domain where neuromorphic computing shows significant

promise. One of the most impactful uses is in brain-machine interfaces (BMIs),
which enable direct communication between the brain and external devices. With
neuromorphic hardware, these systems can provide more natural, real-time
interaction by interpreting neural signals more efficiently and with lower latency.

Neuromorphic systems also have applications in medical diagnostics, where they

can perform fast and energy-efficient pattern recognition on complex data such
as MRI scans, electroencephalograms (EEGs), or genomic sequences. This
can assist doctors in making quicker and more accurate decisions.

Moreover, next-generation prosthetic devices can benefit from neuromorphic

architectures by responding dynamically to real-time neural inputs from the user.
This results in prosthetics that are more responsive, intuitive, and energy-
efficient—potentially transforming the quality of life for individuals with
disabilities.

10 | P a g e
CHALLENGES AND FUTURE DIRECTIONS:
 Future Trends

Neuromorphic computing is poised to significantly shape the future of computing

and artificial intelligence. One of the most promising directions is the
development of hybrid systems that combine quantum computing with
neuromorphic architectures. These systems aim to harness the strengths of both
paradigms—leveraging the probabilistic and entangled nature of quantum
systems alongside the brain-inspired, event-driven efficiency of neuromorphic
models.

There is also a strong push towards integrating neuromorphic processors into

supercomputing platforms, enabling them to tackle complex computational
problems while minimizing energy consumption. Additionally, advancements in
spiking neural network (SNN) algorithms are unlocking new levels of
performance in adaptive learning and pattern recognition.

As research progresses, neuromorphic computing is expected to become an

essential technology for energy-efficient AI, particularly in edge computing,
autonomous systems, and real-time data processing applications. Its biologically
inspired design aligns perfectly with future needs for intelligent systems that can
operate in a low-power, decentralized, and highly responsive manner.

 Challenges

Despite its enormous potential, neuromorphic computing faces several technical

and practical challenges. A major obstacle is the integration of memristor
technology with existing CMOS-based systems. While memristors offer unique
advantages like non-volatility and analog computation, ensuring compatibility
with traditional silicon technology remains a hurdle.

Scalability is another concern—neuromorphic systems must maintain low power

consumption even as they grow in complexity and size. Moreover, there is a need
for specialized algorithms that can fully exploit the capabilities of spiking neural
networks and event-driven hardware.

Overcoming these challenges will be crucial for the mainstream adoption of

neuromorphic systems in real-world applications. Continued research in materials
science, circuit design, and algorithm development will play a vital role in
unlocking the full potential of this transformative computing paradigm.

11 | P a g e
CONCLUSION:
Neuromorphic computing represents a transformative approach to artificial
intelligence by adopting brain-inspired models of computation. Unlike traditional
computing systems, neuromorphic architectures leverage key components such
as memristors, spiking neural networks (SNNs), and event-driven
computation to mimic the structure and functioning of the human brain.

These advancements have enabled the creation of highly efficient, low-power

hardware systems capable of parallel and real-time processing—traits that are
essential for intelligent systems operating in dynamic environments. From
autonomous machines to healthcare applications, neuromorphic computing
demonstrates immense potential in solving complex, real-world problems.

While technical and integration challenges remain, continued progress in VLSI

design, neuromorphic algorithms, and device technology is paving the way for
widespread adoption. As research advances, neuromorphic computing is expected
to become a cornerstone technology in the next generation of artificial
intelligence systems.

12 | P a g e
REFERENCES:
1. Brunner, D., Reitzenstein, S., & Fischer, I. (2016). All-Optical
Neuromorphic Computing in Optical Networks of Semiconductor Lasers.
Institut FEMTO-ST UMR 6174, University of Bourgogne Franche-
Comté, Besançon, France.
2. Bhanja, T., Abuibaid, K., Nußbaum, J., Legler, T., Wagner, A., &
Ruskowski, M. (2024). The Need for Neuromorphic Computing in
Industrial Robotics. Department of Mechanical and Process Engineering,
University of Kaiserslautern-Landau, Germany.
3. Su, T.-K., Cheng, W.-K., Yang, L., Shi, Z.-M., Chen, L.-W., Wang, W.-
C., & Lin, H.-W. (2024). Halide Perovskite and Metal Halide Memristors
and Artificial Synapses for Neuromorphic Computing. Department of
Materials Science and Engineering, National Tsing Hua University,
Taiwan.
4. Shooshtari, M., Serrano-Gotarredona, T., Través, M. J., Linares-
Barranco, B., Pahlavan, S., & Ruskowski, M. (2024). Applying Hodgkin-
Huxley Neuron Model for Perovskite Memristor in Circuit Simulation.
Neuromorphic Group, IMSE-CNM, University of Seville, Spain.
5. An, H., Ehsan, M., Zhou, Z., & Yi, Y. (2024). Electrical Modeling and
Analysis of 3D Neuromorphic IC with Monolithic Inter-tier Vias.
Department of Electrical Engineering, University of Kansas, USA.
6. Schuman, C.D., Young, S.R., Mitchell, J.P., Johnston, T., Rose, D.,
Maldonado, B.P., & Kaul, B.C. (2024). Low Size, Weight, and Power
Neuromorphic Computing to Improve Combustion Engine Efficiency.
Oak Ridge National Laboratory, USA.
7. Theilman, B.H., Wang, Y., Parekh, O., Severa, W., Smith, J.D., &
Aimone, J.B. (2024). Stochastic Neuromorphic Circuits for Solving
MAXCUT. Neural Exploration and Research Lab, Sandia National
Laboratories, USA.
8. Wang, R., Zhang, X., Shi, T., Wei, J., Lv, H., Liu, Q., & Liu, M. (2024).
Non-filamentary Pd/Al2O3/Ta Memristor as Artificial Synapse for
Neuromorphic Computing. Institute of Microelectronics, Chinese
Academy of Sciences, Beijing, China.
9. Rahman, M., Khasanvis, S., Shi, J., & Moritz, C.A. (2024). Wave
Interference Functions for Neuromorphic Computing. Neuro-Biomorphic
Engineering Lab, The Open University of Israel, Israel.

13 | P a g e
 10. Sharma, S., Mittal, V., & Bhushan, B. (2024). Nanowires Based
Artificial Synaptic Networks for Building Neuromorphic Artificial
Intelligence. HMRITM, GGSIPU, New Delhi, India
11. Suri, M. (2023). CMOS Compatible Phase Change Materials and
Applications in Neuromorphic Computing. IBM Research, Zurich,
Switzerland.
12. Wang, Z., Rao, M., Midya, R., Joshi, S., Jiang, H., Lin, P., & Yang, J.J.
(2024). State-of-the-Art Memristor-Based Artificial Synapses and
Neurons for Neuromorphic Computing. Hewlett Packard Labs, USA.
13. Goel, S., & Kaushik, B.K. (2022). Graphene Nanoribbon FET Based
Ternary Logic Circuits for Neuromorphic Applications. Department of
Electronics and Communication Engineering, IIT Roorkee, India.
14. Sajid, M., Hasan, M., & Masud, M. (2022). 3-D Neuromorphic
Computing with Carbon Nanotube FETs. Department of Electrical
Engineering, IIT Delhi, India.
15. Shaban, H., & Kaushik, B.K. (2022). Performance Analysis of Hybrid
CMOS–Memristor-Based SRAM Cell for Neuromorphic Applications.
Department of Electronics and Communication Engineering, IIT
Roorkee, India.
16. Tang, J., Wu, Z., Chen, P., & Wang, Y. (2023). Implementing a Spiking
Neural Network Model on a Low-Power SpiNNaker Neuromorphic
Platform. College of Artificial Intelligence, Jilin University, China.
17. Krithivasan, S., Jaiswal, A., Gopalakrishnan, S., Agrawal, A., & Roy,
K. (2024). In-Memory Computation with Memristors for Neuromorphic
Applications. Department of Electrical and Computer Engineering,
Purdue University, USA.
18. Burr, G.W., Shelby, R.M., Sebastian, A., Kim, S., Kim, S., Sidler, S.,
Virwani, K., Ishii, M., Narayanan, P., Fumarola, A., Sanches, L.L.,
Boybat, I., Le Gallo, M., Moon, K., & Leblebici, Y. (2017).
Neuromorphic Computing Using Non-Volatile Memory. IBM Research –
Almaden, USA.
19. Hu, M., Li, H., Li, Q., Li, G., Ge, N., Montgomery, E., Davila, N., Jiang,
H., Lin, P., Wang, Z., Strachan, J.P., Wu, H., & Williams, R.S. (2016).
Dot-Product Engine for Neuromorphic Computing: Programming 1T1M
Crossbar to Accelerate Matrix-Vector Multiplication. Hewlett Packard
Labs, USA.
20. Upadhyay, N.K., Hasan, M.M., & Mahmoodi, M.R. (2020). Emerging
Memory Devices for Neuromorphic Computing. Department of Electrical
Engineering and Computer Science, University of Tennessee, USA.
21. Naous, R., Fan, D., & Ramasubramanian, K. (2016). Memristor-Based
Neural Networks: Synaptic Design and Operation. Department of
Electrical and Computer Engineering, George Mason University, USA.

14 | P a g e
 22. Sebastian, A., Le Gallo, M., & Burr, G.W. (2020). Tutorial: Brain-
Inspired Computing Using Phase-Change Memory Devices. IBM
Research – Zurich, Switzerland.
23. Wang, Z., Joshi, S., Savel’ev, S.E., Song, W., Midya, R., Li, Y., Rao, M.,
Yan, P., Asapu, S., Zhuo, Y., Jiang, H., Lin, P., Xia, Q., & Yang, J.J.
(2018). Fully Memristive Neural Networks for Pattern Classification with
Unsupervised Learning. Hewlett Packard Labs, USA.
24. Shao, T., Zhang, H., Wen, Y., Zhou, Y., Zhang, Z., & Li, X. (2020). An
Overview of In-Memory Computing With Emerging Non-Volatile
Memory for Neuromorphic Architectures. State Key Laboratory of ASIC
and System, Fudan University, Shanghai, China.
25. Zidan, M.A., Strachan, J.P., & Lu, W.D. (2018). The Future of
Electronics Based on Memristive Systems. Department of Electrical
Engineering and Computer Science, University of Michigan, USA.
26. Chen, A., & Lin, M.R. (2016). Variability of Resistive Switching
Memories and Its Impact on Crossbar Array Performance. Taiwan
Semiconductor Manufacturing Company (TSMC), Taiwan.
27. Xiang, C., Zhu, J., Liang, X., Song, C., Zhang, H., Li, X., & Chen, C.
(2020). Functional Oxide-Based RRAM for Neuromorphic Computing:
Advances and Perspectives. Institute of Microelectronics, Chinese
Academy of Sciences, Beijing, China.
28. Suri, M., Garbin, D., Sousa, V., Sousa, R., Drouin, D., Querlioz, D.,
Bichler, O., Gamrat, C., & DeSalvo, B. (2015). Phase Change Memory
as Synapse for Ultra-Dense Neuromorphic Systems: Application to
Pattern Recognition. CEA-Leti, France.
29. Boybat, I., Le Gallo, M., Nandakumar, S.R., Moraitis, T., Papandreou,
N., Tang, J., Marangoni, M., Eleftheriou, E., & Sebastian, A. (2018).
Neuromorphic Computing with Multi-Memristive Synapses. IBM
Research – Zurich, Switzerland.
30. Choi, S., Yang, Y., & Lu, W.D. (2017). Flexible Memristive Synapse for
Neuromorphic Computing Systems. Department of Electrical Engineering
and Computer Science, University of Michigan, USA.
31. Prezioso, M., Merrikh-Bayat, F., Hoskins, B.D., Adam, G.C., Likharev,
K.K., & Strukov, D.B. (2015). Training and Operation of an Integrated
Neuromorphic Network Based on Metal-Oxide Memristors. University of
California, Santa Barbara, USA.
32. Burr, G.W., Kurdi, B., Shenoy, R.S., Narayanan, P., Virwani, K.,
Hwang, H., & Kim, K. (2015). Overview of Candidate Device
Technologies for Storage-Class Memory. IBM Almaden Research
Center, USA.
33. Guo, T., Li, Y., & Wang, H. (2018). Spintronic Devices for
Neuromorphic Computing: Principle, Progress and Perspective. Institute
of Semiconductors, Chinese Academy of Sciences, Beijing, China.

15 | P a g e
 34. Sengupta, A., & Roy, K. (2017). Spin-Transfer Torque Magnetic
Neurons for Energy-Efficient Non-Boolean Computation. Purdue
University, USA.
35. Wu, H., Yu, S., Wong, H.S.P., & Wong, S. (2017). CIM–Computing-In-
Memory for Neuromorphic Systems. Stanford University, USA.
36. Soudry, D., Keren, H., & Meir, R. (2014). Memristor-Based Synapses
for Spiking Neural Networks. Technion – Israel Institute of Technology,
Israel.
37. Indiveri, G., & Liu, S.C. (2015). Memory and Information Processing in
Neuromorphic Systems. Institute of Neuroinformatics, University of
Zurich and ETH Zurich, Switzerland.
38. Wan, C., Xie, L., Wu, H., Zhang, Y., & Gao, B. (2020). A Ternary
Content-Addressable Memory with RRAM-based Cell for Neuromorphic
Computing. Institute of Microelectronics, Chinese Academy of Sciences,
Beijing, China.
39. N. Mathews and K. David. "Neuromorphic Systems for Speech
Recognition." MIT Media Lab, Cambridge, USA, 2024.
40. B. Zhang, C. Yuan, and T. Yang. "Neuromorphic Computing: Toward
Hybrid VLSI and Photonic Architectures." Shanghai Jiao Tong
University, Shanghai, China, 2024.
41. K. Zeng, X. Zhou, and M. Wang. "Reinforcement Learning for
Neuromorphic Computing." Peking University, Beijing, China, 2024.
42. S. Tuck, P. Jackson, and M. Moore. "Energy-Efficient and High-
Performance Neuromorphic Circuits for Autonomous Systems."
University of Cambridge, Cambridge, UK, 2024.
43. A. Liu, H. Zhou, and Z. Yang. "Neuromorphic Computing Based on
Two-Terminal Devices." University of Science and Technology of China,
Hefei, China, 2024.
44. M. Liu and F. Tan. "Neuromorphic Chip Design for Vision-based
Applications." Beijing Institute of Technology, Beijing, China, 2024.
45. C. Zhang, J. Zhang, and L. Wang. "Ultra-low Power Neuromorphic
Computing." Northwestern Polytechnical University, Xi’an, China, 2024.
46. D. Sharma and A. Jain. "Optimizing Neuromorphic Circuits for Faster
Neural Networks." University of Hyderabad, Hyderabad, India, 2024.
47. P. Vasan and V. Kiran. "Neuromorphic Chips for Real-time Audio
Processing." Indian Institute of Technology, Madras, India, 2024.
48. T. Naidu and S. Gupta. "Neuromorphic Computing on Chip for Real-
Time Processing." IIT Guwahati, Assam, India, 2024.
49. A. Tripathi and M. Saini. "Neuromorphic Algorithm Design for
Intelligent Sensing Systems." University of Pune, Pune, India, 2024.
50. N. Desai and T. V. Subramanian. "Neuromorphic Computing
Applications in Robotics." Indian Institute of Robotics, Bangalore, India,
2024.

16 | P a g e