SRM INSTITUTE OF SCIENCE AND TECHNOLOGY

FACULTY OF ENGINEERING AND TECHNOLOGY

DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING
20CSE527J - Pattern Recognition Techniques
Question Bank – Unit 1

PART – A

1. What does a Pattern Recognition System Look Like?

Pattern recognition is a data analysis method that uses machine learning algorithms to automatically
recognize patterns and regularities in data. This data can be anything from text and images to sounds
or other definable qualities. Pattern recognition systems can recognize familiar patterns quickly and
accurately.
2. Why do we use pattern recognition?
Pattern recognition is one of the four cornerstones of Computer Science. It involves finding the
similarities or patterns among small, decomposed problems that can help us solve more complex
problems more efficiently.
3. Why feature selection/extraction is required?
Feature selection/extraction is an important step in many machine-learning tasks,
including classification, regression, and clustering. It involves identifying and selecting the most
relevant features (also known as predictors or input variables) from a dataset while discarding the
irrelevant or redundant ones. This process is often used to improve the accuracy, efficiency, and
interpretability of a machine-learning model.

4. What is Classifier?

In machine learning, a classifier is an algorithm that automatically sorts or categorizes data into one

or more "classes." Targets, labels, and categories are all terms used to describe classes. One of the

most prominent instances is an email classifier, which examines emails and filters them according to
whether they are spam or not.
5. What is decision region and decision boundary in pattern recognition?
The decision regions are separated by surfaces called the decision boundaries. These separating
surfaces represent points where there are ties between two or more categories. For a minimum-
distance classifier, the decision boundaries are the points that are equally distant from two or more of
the templates.

6. What is training and learning in pattern recognition?

The data should be divided into two set i.e. training and testing set. Learning from the data can tell
how the predictions of the system are depending on the data provided as well which algorithm suits
well for specific data, this is a very important phase. As data is divided into two categories, we can
use training data to train an algorithm and testing data is used to test model, as already said the data
should be diverse training and testing data should be different.
So we divide data into two sets normally we divide data in which 70% of data is used for training the
model, algorithms extract the important patterns from the provided data and creates a model. Testing
set contains 30% of whole data and it is then used to verify the performance of the model i.e how
accurately is the model predicting the results.
 PART – B
111 What is Pattern Recognition System? Explain in detail about Learning and Adaptation?
1 Pattern Recognition is the branch of machine learning a computer science which deals with the regularities
and patterns in the data that can further be used to classify and categorize the data with the help of Pattern
Recognition System.
Pattern Recognition System
This system comprises of mainly five components namely sensing, segmentation, feature extraction,
classification and post processing. All of these together generates a System and works as follows.

Pattern Recognition System

1. Sensing and Data Acquisition: It includes, various properties that describes the object, such as its
entities and attributes which are captured using sensing device.
2. Segmentation: Data objects are segmented into smaller segments in this step.
3. Feature Extraction: In this step, certain features of data objects such as weight, colors, dimension etc.
are extracted.
4. Classification: Based on the extracted features, data objects are classified.
5. Post Processing & Decision: Certain refinements and adjustments are done as per the changes in
features of the data objects which are in the process of recognition. Thus, decision making can be done
once, post processing is completed.
Need: Pattern Recognition System
 ✓ Pattern Recognition System is responsible for generating patterns and similarities among given
problem/data space, that can further be used to generate solutions to complex problems effectively
and efficiently.
✓ Certain problems that can be solved by humans, can also be made to be solved by machine by
using this process.
Applications Of Pattern Recognition
1. Character Recognition.
2. Weather Prediction.
3. Sonar Detection.
4. Image Processing.
5. Medical Diagnosis.
6. Speech Recognition.
7. Information Management Systems.
Learning and Adaptation
Learning and Adaptation can be collectively called as machine learning which can be defined as the branch
of computer science which enables computer systems to learn and respond to queries on the basis of
experience and knowledge rather than from predefined programs. Also, it can be classified into
supervised, unsupervised and reinforcement learning.
1. Learning is a process in which the acquisition of knowledge or skills through study, experience,
or being taught.
2. Adaptation refers to the act or process of adapting and adjustment to environmental conditions.
1. Learning and Adaptation: Supervised Learning
• When learning of a function can be done from its inputs and outputs, it is called as supervised learning.
• One of the example of supervised learning is "Classification".
• It classifies the data on the basis of training set available and uses that data for classifying new data.
• The class labels on the training data is known in advance which further helps in data
classification.
Issues: Supervised Learning
Data Cleaning: In data cleaning, noise and missing values are handled.
Feature Selection: Abundant an irrelevant attributes are removed while feature selection is done.
Data Transformation: Data normalization and data generalization is included in data
transformation.
Classification Methods
Decision Trees.
Bayesian Classification.
Rule Based Classification.
Classification by back propagation.
Associative Classification.
2. Learning and Adaptation: Unsupervised Learning
• When learning can be used to draw inference from some data set containing input data, it is called as
unsupervised learning.
• It clusters the data on the basis of similarities according to the characteristics found in the data and
grouping similar objects into clusters.
• The class labels on the training data is not known in advance i.e. no predefined class.
• The problem of unsupervised learning involves learning patterns from the inputs when specific
output values are supplied.
• Clustering is an example of unsupervised learning which can further be used on the basis of
different methods as per requirements.
Clustering Methods
• Hierarchical.
• Partitioning.
• Density Based.
• Grid Based.
• Model Based.
3. Learning and Adaptation: Reinforcement Learning
Reinforcement in general is, the action or process of establishing a pattern of behavior.
 Hence, Reinforcement learning is the ability of software agents to learn and get reinforced by acting in
environment i.e. learning from rewards.
In reinforcement learning, the software agents acts upon the environment and gets rewarded for its
action after evaluation but is not told, of which action was correct and helped it to achieve the goal.
For Example: Game Playing, Statistics.
Applications: Reinforcement Learning
• Manufacturing.
• Financial Sector.
• Delivery Management.
• Inventory Management.
• Robotics.
• Delivery Management.
• Inventory Management.
• Robotics.
2. Explain about Training and Learning in Pattern Recognition Systems
Training and Learning in Pattern Recognition:
Learning is a phenomenon through which a system gets trained and becomes adaptable to give
results in an accurate manner. Learning is the most important phase as to how well the system
performs on the data provided to the system depends on which algorithms are used on the data.
The entire dataset is divided into two categories, one which is used in training the model i.e.
Training set, and the other that is used in testing the model after training, i.e. Testing set.
• Training Set:The training set is used to build a model. It consists of the set of images that are
used to train the system. Training rules and algorithms are used to give relevant information on
how to associate input data with output decisions. The system is trained by applying these
algorithms to the dataset, all the relevant information is extracted from the data, and results are
obtained. Generally, 80% of the data of the dataset is taken for training data.
• Testing set: Testing data is used to test the system. It is the set of data that is used to verify
whether the system is producing the correct output after being trained or not. Generally, 20% of the
data of the dataset is used for testing. Testing data is used to measure the accuracy of the system.
For example, a system that identifies which category a particular flower belongs to is able to identify
seven categories of flowers correctly out of ten and the rest of others wrong, then the accuracy is
70 %
Real-time Examples and Explanations: A pattern is a physical object or an abstract notion. While
talking about the classes of animals, a description of an animal would be a pattern. While talking about
various types of balls, then a description of a ball is a pattern. In the case balls considered as pattern,
the classes could be football, cricket ball, table tennis ball, etc. Given a new pattern, the class of the
pattern is to be determined. The choice of attributes and representation of patterns is a very important
step in pattern classification. A good representation is one that makes use of discriminating attributes
and also reduces the computational burden in pattern classification.
An obvious representation of a pattern will be a vector. Each element of the vector can represent one
attribute of the pattern. The first element of the vector will contain the value of the first attribute for
the pattern being considered.
Example: While representing spherical objects, (25, 1) may be represented as a spherical object with
25 units of weight and 1 unit diameter. The class label can form a part of the vector. If spherical objects
belong to class 1, the vector would be (25, 1, 1), where the first element represents the weight of the
object, the second element, the diameter of the object and the third element represents the class of the
object.
Advantages:
• Pattern recognition solves classification problems
• Pattern recognition solves the problem of fake biometric detection.
• It is useful for cloth pattern recognition for visually impaired blind people.
• It helps in speaker diarization.
• We can recognize particular objects from different angles.
Disadvantages:
• The syntactic pattern recognition approach is complex to implement and it is a very slow process.
• Sometimes to get better accuracy, a larger dataset is required.
• It cannot explain why a particular object is recognized.
Example: my face vs my friend’s face.
Applications:
• Image processing, segmentation, and analysis
Pattern recognition is used to give human recognition intelligence to machines that are required in
image processing.
• Computer vision
Pattern recognition is used to extract meaningful features from given image/video samples and is
used in computer vision for various applications like biological and biomedical imaging.
• Seismic analysis
The pattern recognition approach is used for the discovery, imaging, and interpretation of
temporal patterns in seismic array recordings. Statistical pattern recognition is implemented and
used in different types of seismic analysis models.
• Radar signal classification/analysis
Pattern recognition and signal processing methods are used in various applications of radar signal
classifications like AP mine detection and identification.
• Speech recognition: The greatest success in speech recognition has been obtained using pattern
recognition paradigms. It is used in various algorithms of speech recognition which tries to avoid
the problems of using a phoneme level of description and treats larger units such as words as pattern
• Fingerprint identification: Fingerprint recognition technology is a dominant technology in the
biometric market. A number of recognition methods have been used to perform fingerprint matching
out of which pattern recognition approaches are widely used.
Imagine we have a dataset containing information about apples and oranges. The features of each fruit
are its color (red or yellow) and its shape (round or oval). We can represent each fruit using a list of
strings, e.g. [‘red’, ’round’] for a red, round fruit.

Our goal is to write a function that can predict whether a given fruit is an apple or an orange. To do
this, we will use a simple pattern recognition algorithm called k-nearest neighbors (k-NN).

Here is the function in Python:

• Python3

\from collections import Counter

def predict(fruit):

# Count the number of apples and oranges in the training data

num_apples = sum([1 for f in training_data if f[-1] == 'apple'])

num_oranges = sum([1 for f in training_data if f[-1] == 'orange'])

# Find the k nearest neighbors of the fruit

nearest_neighbors = find_nearest_neighbors(fruit, training_data, k=5)

# Count the number of apples and oranges among the nearest neighbors

num_apples_nn = sum([1 for nn in nearest_neighbors if nn[-1] == 'apple'])

num_oranges_nn = sum([1 for nn in nearest_neighbors if nn[-1] == 'orange'])

# Predict the label of the fruit based on the majority class among the nearest neighbors

if num_apples_nn > num_oranges_nn:

return 'apple'

else:

return 'orange'

# Create a training dataset

training_data = [

['red', 'round', 'apple'],

['red', 'oval', 'apple'],

['yellow', 'round', 'orange'],

 ['yellow', 'oval', 'orange']

# Create a test fruit

test_fruit = ['red', 'round']

# Predict the label of the test fruit

prediction = predict(test_fruit)

print(prediction)

Output:
‘apple’
3 Explain about Stages of Pattern Recognition and its Types?

Pattern recognition is a branch of machine learning that emphasizes the recognition of data patterns. There are
two main parts to it:

1. Explorative Pattern Recognition—aims to identify data patterns in general. The algorithms focus on
looking for hidden patterns or clusters of features in the data. They mainly rely on the unsupervised
classification that assigns the input pattern as an undefined class.

2. Descriptive Pattern Recognition—aims to categorize the detected patterns and identify them as part
of a predefined class. It mainly uses supervised classification.

Pattern recognition is often a classification or categorization task. The classes are either defined by the system
(in the case of supervised classification) or learned based on the similarity between patterns (as in unsupervised
classification).

Types of Pattern Recognition Models

The major approaches to pattern recognition define the different types of models that are used in common:
 Types of pattern recognition models

In terms of the approaches to pattern recognition, there are four main ones:
Check out the comparative chart below that focuses on Statistical Pattern Recognition (StatPR) vs.
Syntactic Pattern Recognition (SyntPR) vs. Neural Pattern Recognition (NeurPR):

Statistical Pattern Recognition

In Statistical Pattern Recognition (StatPR), each pattern is described with the help of d features or
measurements and is viewed as a point in a d-dimensional space. The goal of StatPR is to choose the
features that allow pattern vectors to belong to different categories in this d-dimensional feature space

As the name suggests, this model relies on historical data points and statistical techniques to learn the features
and patterns from the data. This includes collecting observations, studying and analysing them to infer general
rules or concepts that can be applied to new, unseen observations.

The pattern is grouped based on its features, and the number of features determines how the pattern is viewed as
a point in a d-dimensional space. It follows this simple technique to recognize patterns:

1. Representation. Identifying the way the objects are related maybe by a n-dimensional vector space.

2. Generalization. The rules and concepts that can be derived from a given representation of the set of
examples, and can be therefore applied to unseen data points.

3. Evaluation. Accuracy and trustworthiness estimates of the model to assess the system's performance.

For example, this model is handy when developing AI for the financial sector to predict stock prices based on
past market trends and extrapolate market conditions from past data.
Syntactic Pattern Recognition
Syntactic Pattern Recognition (SyntPR) relies on the elementary/simplest subpatterns that are called
primitives (for example, letters of the alphabet). The pattern then is described in terms of the primitives’
interrelation, when they are assembled into words and sentences. There are grammatical rules that
command the model, which have to be inferred from the available training samples. Patterns that include
structural or relational information are hard to quantify as feature vectors due to their underlying complexity. In
such cases, statistical pattern recognition may prove insufficient. However, you can classify data based on
structural similarities in data patterns. That’s why syntactic pattern recognition techniques work for picture
recognition and scene analysis, where patterns are complex and the number of features is extensive. Breaking
complex patterns into simpler hierarchical sub-patterns comes in handy for recognizing roads, rivers, or humans
in images, or syntax and grammar in text.

Neural Pattern Recognition

Neural Pattern Recognition (NeurPR) are sizeable parallel computing systems that consist of a vast
number of simple processors and many interconnections between them. The main characteristics of NN
are that they can learn complex nonlinear input-output relations, use sequential training procedures, and
adapt themselves to the data. By far, it’s the most popular technique for detecting patterns, thanks to the amount
of complexity it can handle. An Artificial Neural Network is a computational system modeled after the neural
network architecture, similar to how the human brain processes complex signals. It can learn to recognize patterns
in various data types (e.g., textual, visual, or audio), and works efficiently with unknown data.
Template Matching

Object identification using template matching

 This technique matches the object's features with the predefined template and identifies the object by proxy. It’s
commonly used in computer vision for object detection in such fields as robotics, vehicle tracking, and others to
find small sections of an image that match the template. In medical imaging, for example, nodules are detected
via template matching techniques.

4. What Does Pattern Recognition Mean and explain the basic components of Pattern Recognition
System?

Pattern recognition is a cognitive process that happens in our brain when we match some information that
we encounter with data stored in our memory. For example, when a mom teaches her kid to count, she
says, “One, two, three.” After multiple repetitions, when mom says, “One, two…”, the child can respond
with “Three.” As we can see, the child recognized the pattern.

There are also pattern recognition receptors (PRR) in our body – macrophages, monocytes, etc. – cells
that have a specific mission to identify and tackle pathogenic molecular patterns and damage-associated
molecular patterns. But that’s biology and not technology.
What is pattern recognition in computer science? In computer science and machine learning, pattern
recognition is a technology that matches the information stored in the database with the incoming data.
Sometimes people ask, “What is the difference between pattern recognition and machine learning?” The
answer is simple: pattern recognition is a type of machine learning.

The Basic Components of Pattern Recognition System

As you can see from the chart above, the result of the pattern recognition can be either class assignment,
cluster assignment, or predicted variables. Therefore, there is no point in asking “what is the difference
between pattern recognition and classification” – classification algorithm is a part of the supervised
machine learning problems, where the target value is a finite set of classes.

We also have to distinguish between pattern recognition and computer vision. While these two
technologies seem similar, computer vision technology mostly focuses on processing and analyzing
images and visual information, such as object detection, visual-based learning, and segmentation. Pattern
recognition, on the other hand, is aimed at the automated discovery of patterns in all kinds of data – visual
as well as others.

There is also a term called “curse of dimensionality.” What is the curse of dimensionality in pattern
recognition?

This phenomenon can be found in areas such as sampling, combinatorics, data mining, and numerical
analysis (among many others.) The issue here is that when the dimensionality increases, the space volume
increases fast as well, and available data becomes sparse. Why is it called the dimensionality curse?
Because, for statistically sound and valid results, you need a decent amount of information.
 Pattern recognition is the technology that enables the learning process. Therefore it is an integral part of
the entire technique of machine learning. It empowers the algorithms to discover regularities within vast
amounts of data and helps to classify them into various categories.

5. Expain the types of Pattern Recognition and how does pattern recognition work?

There are various types of algorithms used in the field. Here are some commonly used ML models
along with brief descriptions:

1. k-Nearest Neighbours (k-NN): The k-NN algorithm classifies patterns by comparing them to the
labeled patterns in the training set. It assigns the label of the majority of the k nearest neighbors to
the pattern being classified.
2. Support Vector Machines (SVM): SVM is a powerful algorithm used for both classification and
regression tasks. It constructs hyperplanes to separate patterns of different classes with the
maximum margin. SVM can handle high-dimensional data and is effective in dealing with
complex patterns.
3. Decision Trees: Decision trees use a hierarchical structure of decision nodes and leaf nodes to
classify patterns. The nodes apply conditions on features to split the data into different classes
recursively until reaching the leaf nodes, which represent the final class labels.
4. Random Forest: Random Forest is an ensemble learning method that combines multiple decision
trees to improve accuracy and reduce overfitting. It constructs a forest of decision trees and uses
their collective predictions to make the final classification decision.
5. Neural Networks: Neural networks, particularly deep learning architectures, have gained
popularity in pattern recognition. They consist of interconnected layers of artificial neurons that
learn complex patterns by adjusting the weights between the neurons through a process called
training. Convolutional Neural Networks (CNNs) are particularly effective for image recognition
tasks.
6. Naive Bayes: Naive Bayes is a probabilistic algorithm based on Bayes' theorem. It assumes
independence among features and calculates the probability of a pattern belonging to a particular
class based on the probabilities of its individual features. Despite its simplicity, Naive Bayes can
perform well in many pattern recognition tasks.
7. Principal Component Analysis (PCA): PCA is a dimensionality reduction technique used to
transform high-dimensional data into a lower-dimensional space. It finds the orthogonal axes that
capture the maximum variance in the data, allowing for efficient representation and visualization
of patterns.
Features of Pattern Recognition
Pattern recognition offers several technical features that contribute to the analysis and understanding
of data. Here are some key technical benefits and features of recognizing patterns:
1. Automation: Pattern recognition enables the automation of tasks that require the identification or
classification of patterns within data. By developing algorithms that can recognize patterns
automatically, repetitive, or time-consuming tasks can be performed efficiently and consistently
without manual intervention.
2. Data analysis and Interpretation: It helps in analyzing large and complex datasets by extracting
meaningful patterns and structures. By identifying and understanding patterns within the data,
valuable insights can be gained, leading to better decision-making, and understanding of
underlying relationships.
3. Prediction and Forecasting: It can be used for predictive modeling and forecasting. By analyzing
historical patterns and trends in the data, algorithms can learn and make predictions about future
outcomes or events. This is particularly useful in domains such as finance, weather forecasting,
stock market analysis, and demand forecasting.
4. Anomaly Detection: Anomaly detection is an important application in this field. By establishing
patterns of normal behavior within the data, algorithms can identify unusual or anomalous patterns
that deviate significantly from the expected patterns. Anomaly detection is valuable in detecting
fraud, network intrusion, equipment failure, or any other abnormal behavior.
5. Pattern Matching and Search: Techniques involving the recognition of patterns enable efficient
matching and searching of patterns within large datasets. By defining a pattern of interest,
algorithms can quickly search and locate occurrences of that pattern, even in vast amounts of data.
This is particularly useful in text mining, image recognition, and DNA sequence analysis.
6. Pattern Visualization: Visualization techniques help in representing and interpreting patterns
visually. Visualizing patterns can provide intuitive insights into the data, facilitate pattern
exploration, and aid in communicating the results effectively to stakeholders.
7. Feature Engineering and Selection: Pattern recognition often involves feature engineering,
where relevant features are extracted or constructed from the data. These features help capture the
essential characteristics of the patterns and improve the performance of recognition algorithms.
Additionally, these techniques can assist in feature selection, identifying the most informative
features for efficient and accurate recognition of patterns.
8. Real-time Pattern Recognition: Advances in recognizing patterns have enabled real-time or near
real-time processing of data streams. This is particularly valuable in applications such as video
surveillance, sensor networks, and industrial monitoring, where timely recognition of patterns is
crucial for decision-making and response.
These technical benefits contribute to the advancement of numerous domains, enabling automation,
prediction, anomaly detection, and efficient analysis of complex data, ultimately leading to improved
efficiency, accuracy, and understanding.

How Does Pattern Recognition Work?

Pattern recognition is the process of identifying and classifying patterns or structures within data. It
involves extracting meaningful features from the data and using them to make informed decisions or
predictions. The steps involved in it can be summarized as follows:

1. Data Collection and Pre-processing

The first step is to collect the relevant data that contains the patterns of interest. This data may come
from various sources, such as sensors, databases, text documents, or images. Before proceeding with
recognition, it is often necessary to pre-process the data by cleaning, normalizing, or transforming it
to ensure its quality and compatibility with the analysis techniques.

2. Feature Extraction

Feature extraction is a crucial step in pattern recognition. It involves selecting and representing the
most relevant information or attributes from the raw data. These features should capture the essential
characteristics of the patterns while reducing the dimensionality of the data. Feature extraction
techniques can vary depending on the type of data and the specific problem at hand. Common
techniques include statistical measures, frequency analysis, wavelet transforms, or image descriptors.

3. Feature Selection

In some cases, the raw data may contain many features that are not all equally informative or
discriminative for pattern recognition. Feature selection aims to identify the most relevant subset of
features that contribute to accurate recognition while reducing noise and redundancy. This step helps
improve computational efficiency, mitigate the curse of dimensionality, and enhance the performance
of the recognition system.

4. Training Phase

In pattern recognition, training involves providing the algorithm with a labeled dataset, where the
patterns of interest are already identified or labeled. During the training phase, the algorithm analyses
the provided examples and learns the patterns and relationships between the features and their
corresponding labels. The learning process can be supervised, where the labels are provided, or
unsupervised, where the algorithm discovers patterns and structures on its own.

5. Classification or Recognition Phase

Once the training is complete, the pattern recognition system is ready to classify or recognize new,
unseen patterns. In this phase, the algorithm applies the learned model to the feature representations of
the new data and assigns labels or categories to them. Classification can be based on various techniques
such as statistical models, decision trees, support vector machines, neural networks, or deep learning
architectures.

6. Performance Evaluation

The systems associated with recognizing patterns need to be evaluated to assess their accuracy and
effectiveness. Performance evaluation involves measuring the system's ability to correctly recognize
patterns. Common evaluation metrics include accuracy, precision, recall, F1 score, or receiver
operating characteristic (ROC) curve analysis. Cross-validation techniques, such as training-testing
splits or k-fold cross-validation, are often employed to obtain reliable performance estimates.

7. Refinement and Iteration

It is an iterative process. If the performance of the recognition system is not satisfactory, it may require
refinement and iteration. This can involve adjusting the feature extraction techniques, modifying the
learning algorithms, or collecting additional or more representative data to improve the system's
performance.

It is important to note that the specific techniques and algorithms used in pattern recognition can vary
depending on the type of data, the complexity of the patterns, and the domain of application.
Additionally, advancements in machine learning, deep learning, and artificial intelligence have
significantly influenced this field, allowing for more sophisticated and accurate recognition systems
across various domains.

Pattern recognition is a process that looks at the available data and tries to see whether there are any
regularities within it. There are two main parts:

1. Explorative part, where the algorithms are looking for patterns in general
2. Descriptive part, where the algorithms start to categorize the found patterns
 Unlike with computer vision (which we discussed above), the datum for pattern recognition can be
anything:

1. Texts or words
2. Images
3. Sentiments (emotions)
4. Sounds
5. Other elements and information
The information that is gleaned from this pattern-searching process can be used for data analytics
systems. This feature is especially vital for big data analytics, where the users cannot process such vast
amounts of data by themselves or with the help of Excel or other similar tools.

Here’s a simplified process of how pattern recognition works:

As you can see from the chart above, the design of the PR system includes these three aspects:

• Data acquisition and preprocessing

• Data representation
• Decision making
6 Explain about Neural Approach

There are quite a few approaches for pattern recognition like Statistical, Syntactical, and Neural. The

statistical approach is nothing but to collect historical data and based on the observations and analyses

from those data new patterns are recognized. The syntactical approach is also known as the structural

approach as it mainly relies upon sub-patterns called primitives like words.

The pattern recognition approaches discussed so far are based on direct computation through machines.

Direct computations are based on math and stats related techniques. Other than those techniques another

one is the neural approach, neural networks related topics are discussed here to recognize the patterns. As
it is known to all neuron is the basic unit of brain cells and together these neurons create networks to

control the specific tasks. This neural network is implemented in systems. The outcome of this effort is

the invention of artificial neural networks. [3]

An artificial neural network is a computing system that tries to stimulate the working function of a

biological neural network of human brains. In this network, all the neurons are well connected and that

helps to achieve massive parallel distributing. The input units receive various forms and structures of

information based on an internal weighting system and the neural network attempts to learn about the

information presented to produce one output report [4].

The advantages of neural networks are their adaptive-learning, self-organization, and fault-tolerance

capabilities. For these outstanding capabilities, neural networks are used for pattern recognition

applications. An ANN initially goes through a training phase where it learns to recognize patterns in data,

whether visually, aurally, or textually [4]. Some of the best neural models are back-propagation, high-

order nets, time-delay neural networks, and recurrent nets.

Fig (3): Basic structure of a feed-forward neural network

Normally, only feed-forward networks are used for pattern recognition. Feed-forward means that there is

no feedback to the input. Similar to the way that human beings learn from mistakes, neural networks also

could learn from their mistakes by giving feedback to the input patterns. This kind of feedback would be

used to reconstruct the input patterns and make them free from error; thus increasing the performance of
the neural networks. Of course, it is very complex to construct such types of neural networks. These kinds

of networks are called auto-associative neural networks [3]. This complexity of constructing the network

can be avoided by using back-propagation algorithms. During this supervised phase, the network

compares its actual output produced with what it was meant to produce—the desired output.

The difference between both outcomes is adjusted using backpropagation. This means that the network

works backward, going from the output unit to the input units to adjust the weight of its connections

between the units until the difference between the actual and desired outcome produces the lowest possible

error [4]. Local minima is one of the main problems associated with back-propagation algorithms. In

addition, neural networks have issues associated with hyper-parameters like learning rate, architecture

selection, feature representation, modularity, and scaling. Though there are problems and obstacles, the

application of neural networks has spread everywhere.