GAYATRI VIDYA PARISHAD

COLLEGE FOR DEGREE AND PG

Department of
COURSES (A)
Computer Science

Neuromorph
ic
Under the Computing By :
D.PRASANTH KUMAR
Guidance of : PG232406003
Mr.B.DIVAKAR MSc(CS)
SEM - !V
Assistant
CONTENT
S
INTRODUCTION
BACKGROUND AND
MOTIVATION
BASIC CONCEPTS
HISTORY
RELATED WORK
METHODOLOGY/TECHNOLOGY
FEATURES
ADVANTAGES
DISADVANTAGES
CONCLUSION
 INTRODUCTIO
Neuromorphic computing is an approachN
to designing computer systems that mimic
the structure and functionality of the human brain. It uses artificial neurons and
synapses to process information efficiently, similar to biological neural networks.
Objective:
The goal of neuromorphic computing is to develop energy-efficient and adaptive
computing systems capable of real-time processing and learning, overcoming the
limitations of traditional computing architectures.
Why is it Important?
•Traditional computers struggle with power efficiency and real-time learning.
•Neuromorphic systems process data in parallel, reducing energy consumption.
•They are ideal for AI applications, robotics, edge computing, and intelligent
decision-making.
 BACKGROUND AND
MOTIVATION
Background :
•Traditional computers use the Von Neumann
architecture, where memory and processing units are
separate, leading to high energy consumption and
latency.
•The human brain, in contrast, is highly efficient,
consuming only 20W of power while performing
complex computations.
•Inspired by the brain, neuromorphic computing was
introduced to create energy-efficient, parallel-
processing systems.
 BACKGROUND AND
MOTIVATION
Motivation :
•Growing Demand for Energy-Efficient AI :-
•AI models require massive computation power, which is expensive and power-
hungry.
•Neuromorphic chips offer a low-power alternative with real-time learning.
Limitations of Traditional Computing :-
•Bottleneck in data transfer between CPU and memory.
•Inefficiency in parallel processing compared to brain-like computing.
Advancements in AI and Neuroscience :-
•Progress in Spiking Neural Networks (SNNs) and memristor-based memory
enables neuromorphic chips to process information like the brain.
 BASIC CONCEPTS

1. Neurons & Synapses: Brain-Inspired

Computing :-
•The human brain processes information using neurons
and synapses.
•Neuromorphic systems replicate this with artificial
neurons and synapses, enabling efficient, parallel data
processing.
2. Spiking Neural Networks (SNNs) :-
•Unlike traditional Artificial Neural Networks (ANNs),
SNNs use electrical pulses (spikes) for communication,
mimicking biological neurons.
•Event-driven processing ensures efficient computation
 BASIC CONCEPTS
3. Asynchronous & Parallel
Processing :-
•Traditional computers process data sequentially,
causing delays and inefficiencies.
•Neuromorphic chips execute multiple tasks in
parallel, reducing energy use and overcoming the von
Neumann bottleneck.
4. Neuromorphic Hardware: Chips &
Memristors :-
•Specialized chips (e.g., Intel Loihi, IBM TrueNorth)
are designed to perform brain-like computations.
•Memristors function like biological synapses,
 HISTOR
• Neuromorphic computing was introduced in the 1980s by Carver Mead, who
Y
proposed brain-inspired circuit design
• Early research focused on building analog circuits that mimicked neuron behavior
• In the 2000s, digital neuromorphic chips began to emerge with advances in
technology
• IBM developed the TrueNorth chip in 2014 with one million artificial neurons
• Intel introduced the Loihi chip in 2017 with self-learning capabilities
• The SpiNNaker project in the UK aimed to simulate large-scale brain activity
• The Human Brain Project in the EU focused on brain simulation and neuromorphic
platforms
• Today, neuromorphic computing is being explored for applications in AI, robotics,
and edge computing
 TrueNorth by Loihi by Intel
IBM
www.reallygreatsite.com
 RELATED
1. IBM TrueNorth (2014) :-
WORK
•A neuromorphic chip developed by IBM.
•It mimics the human brain using 1 million artificial neurons and 256 million synapses.
•Uses Spiking neural networks (SNNs) for brain-like computations.
•Highly energy-efficient (~70 milliwatts), making it ideal for AI applications like pattern
recognition and image processing.
2. Intel Loihi (2017) :-
•A self-learning neuromorphic processor created by Intel.
•Supports on-chip learning, meaning it can adapt and improve performance without
external training (unlike traditional AI models).
•Designed for low-power AI applications, robotics, and real-time decision-making.
•Used in edge computing, where devices process data locally instead of relying on cloud
computing.
 RELATED
WORK
3. SpiNNaker (Spiking Neural Network Architecture) :-
•A research project led by the University of Manchester.
•Designed to simulate large-scale neural networks, supporting up to a billion neurons.
•Uses biological spiking behavior, making it useful for neuroscience research and brain simulations.
•Helps scientists understand how the brain processes information and can be used in brain-machine
interfaces.
4. Human Brain Project (HBP) & DARPA Research :-
•Human Brain Project (HBP) is a European research initiative aiming to digitally reconstruct the human
brain. It uses neuromorphic computing to study brain functions.
•DARPA (Defense Advanced Research Projects Agency) is a U.S. government agency funding
neuromorphic AI research for defense, robotics, and cybersecurity applications.
•Both focus on developing adaptive, energy-efficient, and real-time AI systems inspired by the brain.
These projects demonstrate the real-world impact of neuromorphic computing, advancing AI,
neuroscience, and intelligent decision-making.
Spiking Neural Network
 METHODOLOGY/
Neuromorphic Chips :- TECHNOLOGY
Neuromorphic chips are designed to process
information like the human brain. Unlike
traditional processors, which follow a
sequential execution model, these chips use
event-driven processing, activating only
when necessary. This approach significantly
reduces power consumption while enabling
real-time decision-making. Examples
include Intel Loihi, IBM TrueNorth, and
SpiNNaker.
 METHODOLOGY/
TECHNOLOGY
Spiking Neural Networks (SNNs) :-
SNNs replicate the way biological neurons
communicate using spikes (electrical pulses)
instead of continuous signals. Unlike traditional
artificial neural networks (ANNs), which process
data at fixed time intervals, SNNs operate
asynchronously, improving efficiency,
adaptability, and real-time learning. This makes
them ideal for energy-efficient AI applications.
 METHODOLOGY/
TECHNOLOGY
Memristors (Memory Resistors) :-
Memristors are brain-inspired memory devices that store and process data in a way
similar to biological synapses. Unlike conventional memory (RAM and storage),
memristors can retain information even when power is turned off. This property
enhances self-learning capabilities in neuromorphic computing systems, making
them more efficient for AI-driven tasks.
Parallel & Local Processing :-
Neuromorphic computing eliminates the von Neumann bottleneck by integrating
memory and processing into a single framework. Unlike traditional architectures
that rely on sequential execution, neuromorphic systems enable massive
parallelism, allowing multiple computations to occur simultaneously. This leads to
faster processing, reduced latency, and lower energy consumption.
FEATURES
•Ultra-Low Power Consumption :-
Operates on event-driven processing, reducing energy use to milliwatts, making it ideal for AI at
the edge.
•Real-Time Learning & Adaptation :-
Learns on-the-fly without retraining, enabling autonomous decision-making in robotics and AI.
•Massive Parallel Processing :-
Mimics the brain’s ability to process multiple tasks simultaneously, improving speed and
efficiency.
•Brain-Inspired Architecture :-
Uses neurons and synapses for dynamic, adaptive, and self-organizing computing.
•Reduced Latency :-
Processes data locally without relying on cloud computing, enabling instant decision-making.
•Event-Driven Computation :-
www.reallygreatsite.com
Activates only when needed, minimizing unnecessary computations and improving efficiency.
ADVANTAGES
Energy-Efficient – Consumes far less power than traditional
CPUs/GPUs.
Better Pattern Recognition – Excels in AI applications
like image and speech recognition.
Real-Time Decision-Making – Processes data instantly for
fast, adaptive responses.
Scalability – Can handle complex computations efficiently as
networks grow.
www.reallygreatsite.com
DISADVANTAGE
S
Hardware Complexity – Still evolving, requiring
specialized components.
Difficult to Program – Needs new algorithms and
unconventional training methods.
Limited Adoption – Not yet widely used compared to
CPUs and GPUs.
www.reallygreatsite.com
 CONCLUSIO
N
Neuromorphic computing is a new and
powerful approach that mimics the brain to
make computing faster, smarter, and more
energy-efficient. It has great potential in AI,
robotics, and real-time decision-making.
While there are challenges like complex
hardware and programming, continued
research will help make it more widely used
and impactful in the future.
Thank You
www.reallygreatsite.com