Experience Seamless Full Ebook Downloads for Every Genre at textbookfull.

com

Computational Intelligence and Machine Learning

Proceedings of the 7th International Conference on
Advanced Computing Networking and Informatics
ICACNI 2019 Jyotsna Kumar Mandal
https://textbookfull.com/product/computational-intelligence-
and-machine-learning-proceedings-of-the-7th-international-
conference-on-advanced-computing-networking-and-informatics-
icacni-2019-jyotsna-kumar-mandal/

OR CLICK BUTTON

DOWNLOAD NOW

Explore and download more ebook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Advanced Computing and Communication Technologies Jyotsna

Kumar Mandal

https://textbookfull.com/product/advanced-computing-and-communication-
technologies-jyotsna-kumar-mandal/

textboxfull.com

Methodologies and Application Issues of Contemporary

Computing Framework Jyotsna Kumar Mandal

https://textbookfull.com/product/methodologies-and-application-issues-
of-contemporary-computing-framework-jyotsna-kumar-mandal/

textboxfull.com

Advanced Information Networking and Applications:

Proceedings of the 33rd International Conference on
Advanced Information Networking and Applications
(AINA-2019) Leonard Barolli
https://textbookfull.com/product/advanced-information-networking-and-
applications-proceedings-of-the-33rd-international-conference-on-
advanced-information-networking-and-applications-aina-2019-leonard-
barolli/
textboxfull.com

Proceedings of the International Conference on Advanced

Intelligent Systems and Informatics 2019 Aboul Ella
Hassanien
https://textbookfull.com/product/proceedings-of-the-international-
conference-on-advanced-intelligent-systems-and-informatics-2019-aboul-
ella-hassanien/
textboxfull.com
Advances in Computational Intelligence Proceedings of
Second International Conference on Computational
Intelligence 2018 Sudip Kumar Sahana
https://textbookfull.com/product/advances-in-computational-
intelligence-proceedings-of-second-international-conference-on-
computational-intelligence-2018-sudip-kumar-sahana/
textboxfull.com

Computational Intelligence, Communications, and Business

Analytics: Second International Conference, CICBA 2018,
Kalyani, India, July 27–28, 2018, Revised Selected Papers,
Part I Jyotsna Kumar Mandal
https://textbookfull.com/product/computational-intelligence-
communications-and-business-analytics-second-international-conference-
cicba-2018-kalyani-india-july-27-28-2018-revised-selected-papers-part-
i-jyotsna-kumar/
textboxfull.com

Computational Intelligence, Communications, and Business

Analytics: Second International Conference, CICBA 2018,
Kalyani, India, July 27–28, 2018, Revised Selected Papers,
Part II Jyotsna Kumar Mandal
https://textbookfull.com/product/computational-intelligence-
communications-and-business-analytics-second-international-conference-
cicba-2018-kalyani-india-july-27-28-2018-revised-selected-papers-part-
ii-jyotsna-kuma/
textboxfull.com

Proceedings of International Joint Conference on

Computational Intelligence IJCCI 2019 Mohammad Shorif
Uddin
https://textbookfull.com/product/proceedings-of-international-joint-
conference-on-computational-intelligence-ijcci-2019-mohammad-shorif-
uddin/
textboxfull.com

Contemporary Advances in Innovative and Applicable

Information Technology Proceedings of ICCAIAIT 2018
Jyotsna Kumar Mandal
https://textbookfull.com/product/contemporary-advances-in-innovative-
and-applicable-information-technology-proceedings-of-
iccaiait-2018-jyotsna-kumar-mandal/
textboxfull.com
Advances in Intelligent Systems and Computing 1276

Jyotsna Kumar Mandal ·

Imon Mukherjee · Sambit Bakshi ·
Sanjay Chatterji · Pankaj K. Sa Editors

Computational
Intelligence and
Machine Learning
Proceedings of the 7th International
Conference on Advanced Computing,
Networking, and Informatics
(ICACNI 2019)
Advances in Intelligent Systems and Computing

Volume 1276

Series Editor
Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences,
Warsaw, Poland

Advisory Editors
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
Rafael Bello Perez, Faculty of Mathematics, Physics and Computing
Universidad Central de Las Villas, Santa Clara, Cuba
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
Hani Hagras, School of Computer Science and Electronic Engineering
University of Essex, Colchester, UK
László T. Kóczy, Department of Automation, Széchenyi István University,
Gyor, Hungary
Vladik Kreinovich, Department of Computer Science, University of Texas
at El Paso, El Paso, TX, USA
Chin-Teng Lin, Department of Electrical Engineering, National Chiao
Tung University, Hsinchu, Taiwan
Jie Lu, Faculty of Engineering and Information Technology
University of Technology Sydney, Sydney, NSW, Australia
Patricia Melin, Graduate Program of Computer Science, Tijuana Institute
of Technology, Tijuana, Mexico
Nadia Nedjah, Department of Electronics Engineering, University of Rio de
Janeiro, Rio de Janeiro, Brazil
Ngoc Thanh Nguyen , Faculty of Computer Science and Management
Wrocław University of Technology, Wrocław, Poland
Jun Wang, Department of Mechanical and Automation Engineering
The Chinese University of Hong Kong, Shatin, Hong Kong
The series “Advances in Intelligent Systems and Computing” contains publications
on theory, applications, and design methods of Intelligent Systems and Intelligent
Computing. Virtually all disciplines such as engineering, natural sciences, computer
and information science, ICT, economics, business, e-commerce, environment,
healthcare, life science are covered. The list of topics spans all the areas of modern
intelligent systems and computing such as: computational intelligence, soft comput-
ing including neural networks, fuzzy systems, evolutionary computing and the fusion
of these paradigms, social intelligence, ambient intelligence, computational neuro-
science, artificial life, virtual worlds and society, cognitive science and systems,
Perception and Vision, DNA and immune based systems, self-organizing and
adaptive systems, e-Learning and teaching, human-centered and human-centric
computing, recommender systems, intelligent control, robotics and mechatronics
including human-machine teaming, knowledge-based paradigms, learning para-
digms, machine ethics, intelligent data analysis, knowledge management, intelligent
agents, intelligent decision making and support, intelligent network security, trust
management, interactive entertainment, Web intelligence and multimedia.
The publications within “Advances in Intelligent Systems and Computing” are
primarily proceedings of important conferences, symposia and congresses. They
cover significant recent developments in the field, both of a foundational and
applicable character. An important characteristic feature of the series is the short
publication time and world-wide distribution. This permits a rapid and broad
dissemination of research results.
Indexed by SCOPUS, DBLP, EI Compendex, INSPEC, WTI Frankfurt eG, zbMATH,
Japanese Science and Technology Agency (JST), SCImago.

More information about this series at http://www.springer.com/series/11156

Jyotsna Kumar Mandal · Imon Mukherjee ·
Sambit Bakshi · Sanjay Chatterji · Pankaj K. Sa
Editors

Computational Intelligence
and Machine Learning
Proceedings of the 7th International
Conference on Advanced Computing,
Networking, and Informatics (ICACNI 2019)
Editors
Jyotsna Kumar Mandal Imon Mukherjee
Department of Computer Science and Department of Computer Science and
Engineering Engineering
University of Kalyani Indian Institute of Information Technology
Kalyani, West Bengal, India Kalyani
Kalyani, West Bengal, India
Sambit Bakshi
Department of Computer Science and Sanjay Chatterji
Engineering Department of Computer Science and
National Institute of Technology Rourkel Engineering
Rourkela, Odisha, India Indian Institute of Information Technology
Kalyani
Pankaj K. Sa Kalyani, West Bengal, India
Department of Computer Science and
Engineering
National Institute of Technology Rourkel
Rourkela, Odisha, India

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-981-15-8609-5 ISBN 978-981-15-8610-1 (eBook)
https://doi.org/10.1007/978-981-15-8610-1

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore
Pte Ltd. 2021
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

The 7th International Conference on Advanced Computing, Networking, and Infor-

matics (ICACNI, 2019) was held during 20–21 December 2019, at the Indian Institute
of Information Technology Kalyani, India. There were theoretical research works
from Machine Learning and other algorithm domains and their applications.
This book focuses on both theory and applications in the broad areas of compu-
tational intelligence and machine learning. ICACNI 2019 presents research papers
in the areas of advanced computing, networking, and informatics. It brings together
contributions from scientists, professors, scholars, and students and presents essen-
tial information on the topic. It also discusses the practical challenges encountered
and the solutions used to overcome them, the goal being to promote the “translation”
of basic research into applied research and of applied research into practice. The
works presented here also demonstrate the importance of basic scientific research in
a range of fields.
Some of the works are on Natural Language Processing domain like a sarcasm
detection tool using Deep Neural Network. An article is on sentiment analysis using
VADER for Twitter data. The book consists of an article on Latent Dirichlet Allo-
cation (LDA)-based domain-specific sentiment analysis applicable for the Twitter
dataset. Type classification problem is addressed in another article. A model to
retrieve news articles related to the news currently being read based on their context
and background information is also presented. The user-defined parameters for clus-
tering algorithm are done in another article by harnessing the capabilities of Genetic
Algorithm (GA).
There are a few articles based on image processing tasks also. Some other authors
devised few techniques based on some circuit parameters. A numerical model of the
mammalian ventricular cell to exhibit the effects of pacing and stimulus variations
in the Luo–Rudy Phase I model is also presented in this book. Another work is on
distributed algorithm for computational mobile entities, popularly known as swarm
robots that operate and move in continuous space to arrange themselves in a straight
line. A Deep Learning Model (DLSTM) is also presented to forecast the air pollu-
tion. A Canadian Middle Atmosphere Model (CMAM30) data is analyzed for the
temperature variation from 1000 to 0.001 hPa from 2009 to 2010 to study the effect
of window size on two-dimensional Fourier spectra.
v
vi Preface

An efficient adaptive Weighted Hybrid Queue Scheduling Scheme (WHQSS) is

also contributed by the authors. An integrated improved hybrid cuckoo-GKO-based
low power consumption routing protocol is suggested by another author. Convolu-
tional Neural Network (CNN) model-based steganalysis is suggested by the authors.
A novel watermarking-based approach is proposed by some authors to enable digital
media assets to be maintained with their metadata persistently and robustly.
The editors would like to thank the authority of IIIT Kalyani, India, for providing
infrastructural support to host the conference at IIIT Kalyani. Our sincere gratitude
to the authorities of Springer Nature for publishing the presented papers of this
conference as a book series.
This volume will be useful to the researchers, budding engineers, and undergrad-
uate and postgraduate students of related disciplines.

Kalyani, India Jyotsna Kumar Mandal

Kalyani, India Imon Mukherjee
Rourkela, India Sambit Bakshi
Kalyani, India Sanjay Chatterji
Rourkela, India Pankaj K. Sa
Editors
Contents

Applications of Neural Network

Minutiae Points Extraction Using Faster R-CNN . . . . . . . . . . . . . . . . . . . . . 3
Vivek Singh Baghel, Akhilesh Mohan Srivastava, Surya Prakash,
and Siddharth Singh
Genetic Algorithm-Based Optimization of Clustering Data Points
by Propagating Probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Shailja Dalmia, Aditya Sriram, and T. S. Ashwin
Detection of Malaria Parasites in Thin Blood Smears Using
CNN-Based Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Sabyasachi Mukherjee, Srinjoy Chatterjee, Oishila Bandyopadhyay,
and Arindam Biswas
A Deep Learning Approach for Predicting Air Pollution in Smart
Cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Banani Ghose and Zeenat Rehena
Structural Design of Convolutional Neural Network-Based
Steganalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Pratap Chandra Mandal

Natural Language Processing

Sarcasm Detection of Media Text Using Deep Neural Networks . . . . . . . . 49
Omkar Ajnadkar
Sentiment Analysis Using Twitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Ujjayanta Bhaumik and Dharmveer Kumar Yadav
A Type-Specific Attention Model For Fine Grained Entity Type
Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Atul Sahay, Kavi Arya, Smita Gholkar, and Imon Mukherjee

vii
viii Contents

A Two-Phase Approach Using LDA for Effective Domain-Specific

Tweets Conveying Sentiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Pradnya Bhagat and Jyoti D. Pawar
News Background Linking Using Document Similarity Techniques . . . . . 87
Omkar Ajnadkar, Aman Jaiswal, P. Gourav Sharma, Chandra Shekhar,
and Arun Kumar Soren

GRSS
Formation of Straight Line By Swarm Robots . . . . . . . . . . . . . . . . . . . . . . . . 99
Arijit Sil and Sruti Gan Chaudhuri
A Novel Technique to Utilize Geopolitical Risk as a Factor
for Predicting Gold Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Debanjan Banerjee and Arijit Ghosal
Effect of Window Size on Fourier Space of CMAM30 Model Data
Generated for Satellite Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Subhajit Debnath and Uma Das
Design and Analysis of an Efficient Queue Scheduling Scheme
for Heterogeneous Traffics in BWA Networks . . . . . . . . . . . . . . . . . . . . . . . . 127
Tanusree Dutta, Prasun Chowdhury, Santanu Mondal,
and Rabindranath Ghosh
Simulation of Cardiac Action Potential with Deterministic
and Stochastic Pacing Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Ursa Maity, Anindita Ganguly, and Aparajita Sengupta
A Study on Development of PKL Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
K. A. Khan, Md. Afzol Hossain, Salman Rahman Rasel,
M. Ohiduzzaman, Farhana Yesmin, Lovelu Hassan, M. Abu Salek,
and S. M. Zian Reza
Analysis of Lakes Over the Period of Time Through Image
Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Sattam Ghosal, Abhishek Karmakar, Pushkar Sahay, and Uma Das
Video Watermarking for Persistent and Robust Tracking
of Entertainment Content (PARTEC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Deepayan Bhowmik, Charith Abhayaratne, and Stuart Green

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

About the Editors

Jyotsna Kumar Mandal, M.Tech. in Computer Science from the University of

Calcutta in 1987, awarded Ph.D. (Engineering) in Computer Science and Engineering
by Jadavpur University in 2000. Presently, he is working as Professor of Computer
Science & Engineering, and former Dean, Faculty of Engineering, Technology &
Management, KU, for two consecutive terms during 2008–2012. He is Director,
IQAC, Kalyani University, and Chairman, CIRM, and Placement Cell. He served
as Professor of Computer Applications, Kalyani Government Engineering College,
for two years, as Associate Professor of Computer Science for eight years at North
Bengal University, as Assistant Professor of Computer Science North Bengal Univer-
sity for seven years, and as Lecturer at NERIST, Itanagar, for one year. He has 32
years of teaching and research experience in coding theory, data and network secu-
rity and authentication; remote sensing & GIS-based applications, data compression,
error correction, visual cryptography, and steganography. He was awarded 23 Ph.D.
degrees, one submitted and 8 are pursuing. He has supervised 03 M.Phil., more than
70 M.Tech., and more than 125 M.C.A dissertations. He is Guest Editor of MST
Journal (SCI indexed) of Springer. He published more than 400 research articles
out of which 170 articles in international journals. He published 7 books from LAP
Germany, IGI Global, etc. He has organized 31 international conferences and Corre-
sponding Editors of edited volumes and conference publications of Springer, IEEE,
Elsevier, etc., and edited 32 volumes as Volume Editor.

Imon Mukherjee received his Ph.D. from Jadavpur University, Kolkata, India, in
2015, respectively. Currently, he is working as Faculty in-Charge, Academics, and an
Assistant Professor (Grade-I) in Computer Science & Engineering, Indian Institute
of Information Technology Kalyani, India (an Institute of National Importance under
MHRD, Government of India). Earlier, he was an Assistant Professor in the Depart-
ment of Computer Science and Engineering of the Institute of Technology & Marine
Engineering, now The Neotia University, West Bengal, India, from August 2006
to December 2011. Then from January 2012, he was associated with the Depart-
ment of Computer Science and Engineering, St. Thomas’ College of Engineering
& Technology, Kolkata, India, with the same designation. His research focuses on

ix
x About the Editors

information security and data analytics. He is associated as Reviewer, PC member,

and Track Session Chair of many reputed conferences and journals like IEEE Trans-
actions on Circuits and Systems for Video Technology. He is currently supervising
5 Ph.D. students at IIIT Kalyani. He has acted as PI and Co-PI of many projects
funded by DST, Government of West Bengal, etc.

Sambit Bakshi is currently with Centre for Computer Vision and Pattern Recog-
nition of National Institute of Technology Rourkela, India. He also serves as an
Assistant Professor in the Department of Computer Science & Engineering of the
institute. He earned his Ph.D. degree in Computer Science & Engineering. His area
of interest includes surveillance and biometric authentication. He is a senior member
of IEEE since 2019. He currently serves as Associate Editor of International Journal
of Biometrics (2013–), IEEE Access (2016–), Innovations in Systems and Soft-
ware Engineering: A NASA Journal (2016–), Expert Systems (2017–), and Plos One
(2017–). He has served/is serving as Guest Editor for reputed journals like Multimedia
Tools and Applications, IEEE Access, Innovations in Systems and Software Engi-
neering: A NASA Journal, Computers and Electrical Engineering, IET Biometrics,
and ACM/Springer MONET. He is serving as the Vice-Chair for IEEE Computational
Intelligence Society Technical Committee for Intelligent Systems Applications for
the year 2019. He received the prestigious Innovative Student Projects Award 2011
from Indian National Academy of Engineering (INAE) for his master’s thesis. He
has more than 50 publications in reputed journals, reports, and conferences.

Sanjay Chatterji has received Ph.D. from IIT Kharagpur on Indian Language
Machine Translation domain specifically on statistical and transfer-based approaches
to build Bengali-Hindi machine translation systems based on the resources that are
available. He has built morphological analyzer, POS tagger, parser, etc., for Bengali
language. In Samsung R&D Institute India Bangalore (SRIB), he has worked in the
domain of semantic analysis, specifically on aspect-based sentiment analysis, inter
concept higher order predicate relation extraction, representation and application in
search domain, DNN-based classification of intents, etc. Overall, he has experience
and expertise in application of machine learning algorithms for NLP tasks. He has
worked in SRIB for about 4 years and in multiple engineering colleges as Lecturer
and Assistant Professor for 3 years. He has published 4 journal papers (3 of which
are SCI indexed) and 16 conference papers and achieved jury award for a demo in
Nipun, 2015, in SRIB. Currently, he is working as an Assistant Professor in Indian
Institute of Information Technology Kalyani, West Bengal, India.

Pankaj K. Sa received the Ph.D. degree in Computer Science in 2010. He is

currently serving as an Associate Professor with the Department of Computer Science
and Engineering, National Institute of Technology Rourkela, India. His research
interest includes computer vision, biometrics, visual surveillance, and robotic percep-
tion. He has co-authored a number of research articles in various journals, confer-
ences, and book chapters. He has co-investigated some research and development
About the Editors xi

projects that are funded by SERB, DRDO-PXE, DeitY, and ISRO. He is the recip-
ient of prestigious awards and honors for his excellence in academics and research.
Apart from research and teaching, he conceptualizes and engineers the process of
institutional automation.
Applications of Neural Network
Minutiae Points Extraction Using Faster
R-CNN

Vivek Singh Baghel, Akhilesh Mohan Srivastava, Surya Prakash,

and Siddharth Singh

Abstract A fingerprint is an impression of the friction ridges of all parts of the

finger and it is a widely used biometric trait. In a fingerprint authentication system,
minutiae points are used as features to authenticate an individual. There are many
traditional techniques which have been proposed to extract minutiae points from a
fingerprint. However, efficiently locating minutiae points in a fingerprint image is still
a challenging problem. In this paper, we propose a technique to locate and classify the
minutiae points based on Faster R-CNN, which is a combination of region proposal
network (RPN) and detection network. We use IIT Kanpur fingerprint database to
evaluate the proposed technique and to show the performance.

Keywords Biometrics · Fingerprint · Minutiae Extraction · CNN · Faster R-CNN

V. S. Baghel (B) · A. M. Srivastava · S. Prakash

Indian Institute of Technology Indore, Indore 453552, India
e-mail: phd1801201005@iiti.ac.in
A. M. Srivastava
e-mail: phd1701101001@iiti.ac.in
S. Prakash
e-mail: surya@iiti.ac.in
S. Singh
National Institute of Technology Hamirpur, Hamirpur 176005, India
e-mail: ece16447@nith.ac.in
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer 3
Nature Singapore Pte Ltd. 2021
J. K. Mandal et al. (eds.), Computational Intelligence and Machine Learning, Advances
in Intelligent Systems and Computing 1276,
https://doi.org/10.1007/978-981-15-8610-1_1
4 V. S. Baghel et al.

1 Introduction

A fingerprint is a combination of continuous patterns which are known as ridges

and valleys. In each fingerprint, some unique points exist which are called minutiae
points and these points are used to discriminate a fingerprint from others. Bifurcation
and ridge-end are mainly used minutiae points in a fingerprint authentication system.
The locations where ridges bifurcate into two different ridges are called bifurcation
points, whereas points, where ridges end abruptly, are called ridge-end points. For
the last several years, deep convolutional neural networks (CNNs) [4] has been
used to solve several object recognition problems. The basic architecture of a CNN
consists of several building blocks, including different layers such as convolution
layers, pooling layers (e.g., max pooling and min pooling), fully connected layers,
and classification layer. CNNs have also been used for extracting the minutiae points
from a fingerprint image [3]. R-CNN, which is also called as regions with CNN
features, has been widely used to detect the locations of objects present in a scene
with their class labels. The extraction of the location of minutiae points with their
types (bifurcation or ridge-end) can also be mapped to object detection problem in
an image containing multiple objects. Although R-CNN achieves great results, the
training process has lots of problems because the region proposal has to be generated
for the training dataset. In R-CNN, region proposals are calculated by the selective
search method, which is a time-consuming process. The improved version of R-
CNN is Fast R-CNN, which uses CNN completely for the whole image and RoI
pooling is performed on the feature map for classification and regression. However,
it still uses a selective search method. The improved version of Fast R-CNN is called
Faster R-CNN. It is a fully convolutional network and is fast in training process as
compared to both R-CNN and Fast R-CNN. In this network, RPN is used to propose
the regions where the object of interest may be present in a scene. This motivates
us to use Faster R-CNN for the extraction of minutiae points in a fingerprint image.
To carry out the same, training data has to be prepared, which contains fingerprint
images with ground truth bounding boxes representing the minutiae locations and
their respective class labels (i.e., bifurcation or ridge-end).
Rest of the paper is organized as follows. Section 2 discusses some of the existing
techniques used for minutiae points extraction. Description of Faster R-CNN is given
in Sect. 3. The proposed method is discussed in the Sect. 4. Experimentation and
results are discussed in the Sect. 5. The last section discusses the conclusions.

2 Literature Review

Fingerprint biometrics is a widely explored research area in biometrics and there

are many works that have been carried out in this area. Ratha et al. [13] proposed a
method to extract the minutiae points based on orientation flow and they have com-
pared their result with the ground truth (the manually extracted minutiae points). Jain
Minutiae Points Extraction Using Faster R-CNN 5

et al. [6] have proposed an online fingerprint verification system and their minutiae
extraction algorithm is shown to be much faster and reliable as compared to the
technique proposed in [13]. Hong et al. [11] have proposed a method of enhancing
the fingerprint image to extract minutiae points correctly. Jain et al. [8] proposed
a new representation of features in a fingerprint image instead of minutiae points.
This feature representation is called fingercodes. Jain et al. have explained in [7,
9] about the importance of fingerprint authentication. They have also explained the
feature extraction of a fingerprint image and the matching techniques for fingerprint
authentication. The features, which are basically minutiae points, are extracted by
traversing a 3 × 3 filter onto the thinned fingerprint image. Lavanya et al. [2] pro-
posed a Gabor filter-based enhancement algorithm for minutiae extraction, which
can simultaneously extract ridge orientation and ridge frequency. Other than above-
mentioned techniques, there are many works which have been carried out in the
area of fingerprint authentication for minutiae points extraction based on traditional
approaches.
In recent years, studies show that deep neural network architectures [4] are poten-
tial candidates for feature extraction. In the domain of minutiae extraction, there
are many works which use deep neural networks for minutiae extraction. Jiang
and Liu [10] presents fingerprint minutiae detection based on multi-scale CNN and
employ two separate CNN for minutiae detection and minutiae classification. In
their work, the minutiae detection is carried out using two steps. In the first step,
classification of image patches into minutiae and non-minutiae patch is carried out,
whereas, in the second step, classification is done for minutiae into ridge-end and
bifurcation. Tang et al. [15] have proposed an approach to extract minutiae points
from latent fingerprint images and have compared the accuracy of the authentication
system with the accuracy, while minutiae points are calculated manually. Jiang et
al. [12] proposed a novel minutiae extraction approach by using two nets, i.e., judge
net and locate net. In this work, fingerprint image has been cropped into several
overlapped patches. Classification of these patches into minutiae and non-minutiae
is carried out using judge net, whereas locate net in this work is used to locate the
minutiae points into the overlapped patches.

3 Proposed Method

We propose an automated approach for the extraction of minutiae points from a

fingerprint image. The technique is based on Faster R-CNN [14] which is a fully
convolutional neural network and is end-to-end trainable. The complete schematic
of Faster R-CNN used in the proposed technique for extraction of minutiae points
is shown in Fig. 1. As discussed in the introduction section, there are different types
of minutiae points present in a fingerprint image. Here, bifurcation and ridge-end
points are used as the minutiae points and are detected in a fingerprint image.
6 V. S. Baghel et al.

Classifier

Located Minutiae Points

Fully Connected with Class Labels
Fingerprint Region Proposal Network (RPN) Layer
Bifurcation
Image
Ridge-end
Feature Region
Proposals Bifurcation
ConvNet Map Region Proposal Ridge-end
(ResNet50) Layer

Ridge-end

RoI Pooling
Layer
Anchor
Bifurcation Ridge-end Boxes
Regressor
Coordinates of bounding boxes Detection Network
(Ground Truth)

Fig. 1 The schematic diagram of Faster R-CNN used in fingerprint minutiae extraction

3.1 Ground Truth Preparation

The preparation of ground truth is an essential step before performing training of the
network in the proposed technique. Ground truth for training the network are patches
of size m × m around the minutiae points present in a fingerprint image. These are
represented in the form of bounding boxes and the coordinates of bounding boxes
are used for training purpose. Coordinates of these bounding boxes along with class
label (minutiae type, viz. bifurcation or ridge-end) are stored in an annotation file.
At the time of training of Faster R-CNN, this annotation file is used as an input to
provide the details of the ground truth.

3.2 Training of Faster R-CNN

Faster R-CNN is mainly a combination of RPN and detection network. RPN is a

combination of ConvNet, region proposal layer, and two parallel layers classifica-
tion and regression. Pre-trained ResNet50 [5] is used as a ConvNet in our proposed
approach and hence there is no training required for ResNet50. ResNet50 is a con-
volutional neural network which is used to classify the objects. It generates feature
maps from the fingerprint images and these feature maps are the output of the last
convolution layer or the base layers of ResNet50. These feature maps and ground
truth information obtained in the previous step are further feed-forwarded to the
region proposal layer. Region proposal layer uses anchor boxes of different sizes
and sliding-window strategy to find out the maximum overlap of anchor boxes with
ground truth. In RPN, after the region proposal layer, two parallel layers which are
classification and regression layers exist. Both of these layers are also convolution
layers. The training (fine-tuning) of RPN is performed using pre-trained weights of
Minutiae Points Extraction Using Faster R-CNN 7

base layers, anchor boxes of different sizes, fingerprint images, and the ground truth.
The tuning of weights is mainly performed for region proposal, classification, and
regression layers of RPN, whereas its other remaining part uses pre-trained weights
of ResNet50. The region proposals generated by RPN are used to train the detection
network. RoI pooling layer, fully connected layers, and two parallel layers, i.e., clas-
sification and regression layers are the part of the detection network. Classification
and regression layers are the fully connected layer in the detection network. The
training of detection network is performed using region proposals generated by the
RPN. Backpropagation is used to train both RPN and detection network.

3.3 Localization of Minutiae Points

After training the Faster R-CNN, to carrying out localization of minutiae points, a
fingerprint image is given to the network as an input. In RPN, ConvNet generates
the feature map which is further feed-forwarded to the region proposal layer. Region
proposal layer computes the region proposals based on the anchor box sizes which
have been used during the training of RPN. These region proposals are further feed-
forwarded to the detection network where RoI pooling layer flattens these regions in
the form of feature vectors of same sizes. Computed feature vectors are further used to
classify and detect the location of minutiae points by fully connected, classification,
and regression layers of the detection network. At the end, location and type of
minutiae points are obtained as outputs.

4 Experimental Analysis

This section evaluates the performance of the proposed technique. It also provides the
details of the database and the values of hyperparameters. We have used IIT Kanpur
fingerprint database for carrying out the experimentation. This database contains
1376 subjects where for each subject, there are four samples present in the database.
This gives a total of 5504 images for training, validation, and testing.
The preparation of ground truth is a very important step to train the network
on a given dataset. To accomplish the same, first minutiae points are extracted by
using VeriFinger Demo Version [1]. Further, the bounding boxes of size 13 × 13 are
calculated around the extracted minutiae points for each fingerprint image. Further,
the training is performed on thinned fingerprint images. For thinning of fingerprint
images, there are mainly two steps involved. The first step performs the enhancement
on fingerprint images by using Gabor filters of different orientations and convert
images into binary. In the second step, by using a morphological function, enhanced
binary images are converted into thinned binary images. These thinned images are
used for training purposes. During the training, Faster R-CNN takes an annotation
file as an input. Training has been performed on 70% images of the total fingerprint
8 V. S. Baghel et al.

Table 1 Hyperparameter values for Faster R-CNN used in the proposed technique
Hyperparameter Value
Optimizer-RPN and Detection network “Adam”
Learning rate 0.00001
Number of epochs 20
Epoch length 128
Anchor box sizes 128, 256, and 512 px
RPN Stride 16
Overlaps for classifier RoIs Minimum-0.1, Maximum-0.5
Overlaps for RPN Minimum-0.3, Maximum-0.7
Non-maximum suppression Overlap threshold-0.8

images available in the database. Remaining 30% images are used to validate and
test the proposed technique. The values of hyperparameters for the training of Faster
R-CNN are given in Table 1. It contains different parameter values like types of
optimizer, learning rate, number of epoch, epoch length, anchor boxes scale, etc.
Anchors also play a very important role in the training of the Faster R-CNN network.
They are actually bounding boxes of different sizes that are used to propose the region
by RPN. The ratios between width and height of boxes are chosen as 0.5, 1.0, and
1.5. All possible combinations of these ratios and the size of anchors are used to
perform the training of the network.
The results of the proposed technique on some sample images of the database have
been shown in Fig. 2 where ground truth and detected minutiae points are shown.
As we can see, the proposed technique detects minutiae points quite efficiently. In
Fig. 2a, it can be clearly seen that most of the minutiae points are extracted when we
compare it with the ground truth. In some cases, the number of detected minutiae
points using Faster R-CNN is less as compared to the ground truth. An example
of such cases is depicted in Fig. 2b. It is evident from Fig. 2 that even though the
less number of minutiae points are detected in some cases, their locations and the
class labels (either bifurcation or ridge-end) are predicted correctly. This shows that
the precision value of the proposed technique is high. These results are significant
to show that the proposed technique has worked quite well and this can be further
improved in future to extract minutiae points in more challenging fingerprint images
such as noisy, wet, and dry fingerprint images.

5 Conclusions

In this work, a new technique has been proposed to extract minutiae points from
a fingerprint image using Faster R-CNN. Faster R-CNN, which is a combination
of RPN, and detection network, has a good capability in locating specific objects
Minutiae Points Extraction Using Faster R-CNN 9

Ground Truth Detected Minutiae Ground Truth Detected Minutiae Ground Truth Detected Minutiae

(a)

Ground Truth Detected Minutiae Ground Truth Detected Minutiae Ground Truth Detected Minutiae

(b)

Fig. 2 Results of minutiae points detection using the proposed technique: a Sample images with
large number of detected minutiae points, b Sample images with relatively lesser number of detected
minutiae points as compared to (a)

in a scene. We exploit it for the purpose of locating minutiae points in a fingerprint

image where the specific objects (patches) being located are the regions having ridge-
end and the bifurcation points at its center in a fingerprint image. In the proposed
technique, the first feature map is computed from a fingerprint image by using Con-
vNet which is a part of RPN and further, is feed-forwarded to region proposal layer.
Region proposal layer takes feature map and ground truth as inputs and generates
region proposals which are subsequently feed-forwarded to detection network. RoI
pooling layer which is a part of detection network pools the region proposals into
feature vectors. By using these feature vectors, the types of minutiae points and the
bounding boxes around their locations are computed by the classifier and regressor
of the detection network. The proposed technique has been evaluated on IIT Kanpur
fingerprint database. The detected minutiae points in a fingerprint image are com-
pared with the ground truth. It has been observed that the proposed technique shows
encouraging performance in the detection of minutiae points.

References

1. Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. The MIT Press (2016)
2. Cao, K., Jain, A.K.: Automated latent fingerprint recognition. IEEE Trans. Pattern Anal. Mach.
Intell. 41(4), 788–800 (2019)
3. Ratha, N.K., Chen, S., Jain, A.K.: Adaptive flow orientation-based feature extraction in finger-
print images. Pattern Recogn. 28(11), 1657–1672 (1995)
10 V. S. Baghel et al.

4. Jain, A., Lin Hong, Bolle, R.: On-line fingerprint verification. IEEE Trans. Pattern Anal.
Machine Intell. 19(4), 302–314 (1997)
5. Hong, L., Wan, Y., Jain, A.: Fingerprint image enhancement: algorithm and performance eval-
uation. IEEE Trans. Pattern Anal. Mach. Intell. 20(8), 777–789 (1998)
6. Jain, A.K., Prabhakar, S., Hong, L., Pankanti, S.: Filterbank-based fingerprint matching. IEEE
Trans. Image Process. 9(5), 846–859 (2000)
7. Jain, A.K., Flynn, P., Ross, A.A.: Handbook of Biometrics, 1st edn. Springer Publishing Com-
pany, Incorporated (2010)
8. Jain, A.K., Ross, A.A., Nandakumar, K.: Introduction to Biometrics. Springer Publishing Com-
pany, Incorporated (2011)
9. Lavanya, B.N., Raja, K.B., Venugopal, K.R., Patnaik, L.M.: Minutiae extraction in fingerprint
using gabor filter enhancement. In: Proceedings of International Conference on Advances in
Computing, Control, and Telecommunication Technologies, pp. 54–56 (2009)
10. Jiang, H., Liu, M.: Fingerprint minutiae detection based on multi-scale convolution neural
networks. In: Proceedings of Chinese Conference of Biometric Recognition, vol. 10568, pp.
306–313 (2017)
11. Tang, Y., Gao, F., Feng, J.: Latent fingerprint minutia extraction using fully convolutional
network. In: Proceedings of IEEE International Joint Conference on Biometrics, pp. 117–123
(2017)
12. Jiang, L., Zhao, T., Bai, C., Yong, A., Wu, M.: A direct fingerprint minutiae extraction approach
based on convolutional neural networks. In: Proceedings of International Joint Conference on
Neural Networks, pp. 571–578 (2016)
13. Ren, S., He, K., Girshick, R., Sun, J.: Faster R-CNN: Towards real-time object detection with
region proposal networks. In: Proceedings of Neural Information Processing Systems, pp.
91–99 (2015)
14. He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceed-
ings of IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)
15. Verifinger SDK Demo. https://www.neurotechnology.com/verifinger.html. [Online; Accessed
26 Oct 2019]
Genetic Algorithm-Based Optimization
of Clustering Data Points by Propagating
Probabilities

Shailja Dalmia, Aditya Sriram, and T. S. Ashwin

Abstract Clustering is among the pivotal elementary operations in the field of data
analysis. The efficiency of a clustering algorithm depends on a variety of factors
like initialization of cluster centers, shape of clusters, density of the dataset, and
complexity of the clustering mechanism. Previous work in clustering has managed
to achieve great results but at the expense of a trial and error approach to achieve
optimal values for user-defined parameters which have a huge bearing on the quality
of the clusters formed. In this work, we propose a solution that optimizes the user-
defined parameters for clustering algorithm called Probability Propagation (PP) by
harnessing the capabilities of Genetic Algorithm (GA). In order to overcome this
sensitivity in PP, a novel optimization technique is applied by obtaining the optimal
values of δ and s using GA by maximizing inter-cluster spread and minimizing intra-
cluster spread among the clusters being formed. The proposed method was found
to retrieve top chromosomes (bandwidth and s) with a similar number of clusters,
thus eliminating the sensitivity of user-defined parameters which is optimized by
using GA.

Keywords Data Clustering · Probability Propagation · Density · Genetic

Algorithm

S. Dalmia · A. Sriram · T. S. Ashwin (B)

National Institute of Technology Karnataka, Surathkal Mangalore 575025, Karnataka, India
e-mail: ashwindixit9@gmail.com
S. Dalmia
e-mail: shailjadalmia11@gmail.com
A. Sriram
e-mail: adityasriram1995@gmail.com

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer 11
Nature Singapore Pte Ltd. 2021
J. K. Mandal et al. (eds.), Computational Intelligence and Machine Learning, Advances
in Intelligent Systems and Computing 1276,
https://doi.org/10.1007/978-981-15-8610-1_2
12 S. Dalmia et al.

1 Introduction

Over the past few decades, applications of clustering have tremendously grown in
varied fields as a precursor for data science use cases [4]. The classical k-means
clustering algorithm suffers from preliminary centers for clustering which has an
adverse effect on the quality of clusters and the speed of convergence. To solve the
issue of arbitrary initialization, Frey and Dueck [1] had put forth an effective method
for clustering called Affinity Propagation (AP). The flipside to this solution is the
complex nature of message passing instructions employed by this method.
Motivated by AP, the PP can pin-point clusters with both spherical and non-
spherical shapes. It starts with a stochastic matrix of probabilities and propagates
iteratively until a group of attractors become stable. The AP, PP, and Markov Clus-
tering (MCL) use message passing techniques but the rules of message passing is
much simpler in case of PP with the initialization of stochastic matrix being much
different in this case as compared to the other two. The quality of most cluster-
ing algorithms is deeply influenced by the initialization parameters which result in
different clustering results for varying parameters.
To this end, the PP algorithm does suffer from initialization of two user-defined
parameters, namely bandwidth and s which controls the neighbors set and shape
of clusters, respectively, and thus forms clusters based on the initialization of user-
defined parameters.
In order to address this drawback in PP algorithm, a novel optimization technique
based on Genetic Algorithm (GA) has been proposed.

2 Related Work

Even though clustering plays a pivotal role in the modern applications, very few
algorithms are in place for optimization of such clustering algorithms to retrieve
better clustering results which are insusceptible to any user-defined parameters. To
this end, Frey et al. [1] proposed AP clustering which does not necessitate the amount
of clusters to be recognized beforehand and has edge over other algorithms based
on efficiency and precision but is not appropriate for large-scale data clustering.
Dongen et al. [2] in their work were able to come up with MCL algorithm which is
a simulation of random flow in the graph, however, the overall process is very slow.
DBSCAN is a density-based clustering algorithm which classifies points in space as
core, non-core, and outliers points where high-density region cluster together with
neighboring points leaving outliers to lie in low-density region as different clusters
are formed, but appropriate combination of E radius and minPts cannot be determined
in case of large datasets [3].
Each of the existing clustering algorithms has their own shortcomings based on
different parameters. We aim to conceive optimization for PP clustering algorithm
Genetic Algorithm-Based Optimization of Clustering Data Points … 13

which identify non-spherical and spherical clusters with varying density by employ-
ing a message passing technique with optimal values of initialization parameters
using GA such that it can perform well on large as well as small datasets.

3 Methodology

The proposed optimization of PP algorithm is performed using GA to eliminate

the need for user-defined parameters like bandwidth and s and provide consistent
clustering results. The work by Gana et al. [4] was incorporated along with the
optimization of parameters to minimize intracluster distance and maximize inter-
cluster distance between data points.
Figure 1 represents the flow diagram of the proposed algorithm. The idea is to
evolve the parameter by selecting the fittest chromosomes which would result in better
clustering results rather than varying parameters which would differ the results.

3.1 Optimized PP Algorithm with Genetic Algorithm

The PP algorithm [4] deals with two main user-defined parameters which governed
the cluster orientation in the space. The bandwidth parameter was used as a threshold
value between two data point. The other parameter called s determines the cluster
size. In our work, the proposed model is to optimize these user-defined parameters
by applying GAs [5] to find the ideal values of bandwidth and s such that an optimal
number of clusters can be found. GA begins with the initialization of population with
the selection of the fittest chromosomes accompanied by crossover and mutation.

Fig. 1 Flow diagram of GA-incorporated PP algorithm

14 S. Dalmia et al.

There are two main steps in the procedure of the application of GA:
1. Individual Representation In the proposed procedure, the chromosomes con-
sists of randomized real numbers to symbolize bandwidth and s. The length of
chromosomes is 2 (bandwidth and s). Initially, the population size is fixed and
random initialization of k (population size) chromosomes is done with 2*k genes
values of bandwidth and s.
2. Evaluation of Fitness The purpose of clustering is to rift data into clusters. A
resultant barrier should have below given assets:
• Homogeneity: Data going to the identical cluster should be as alike as probable.
• Heterogeneity: Data going to the dissimilar cluster should be as unlike as prob-
able.
3. The distinction of the data points to be in same or different clusters (as pointed
above) is done on the basis of the distance between them.
In the proposed method, the Davies–Bouldin (DB) index is used to evaluate the
ideality of the clusters which helps to calculate the fitness assessment of the discrete
chromosome. The techniques used below have been used in [13] to identify the
best number of clusters, however, in this work; a novel approach is proposed to
optimize the user-defined constraints through these techniques as will be explained
subsequently. The DB index is applied in [9, 10, 12] also to gauge the soundness of
the clusters. This is defined as the ratio of the sum of cluster scatter which is within
the separation between clusters described in [13]. Then the DB index is demarcated
as follows:
1 
K
DB = Ri,qt . (1)
K i=1

Algorithm 1 Optimized PP Algorithm with GA

1: Randomize initialization of bandwidth parameter and s
2: Initialize the number_of_generations(NGEN), population_size, mutation_rate, and
crossover_rate.
3: while NGEN >0 do
4: for each bandwidth and s do
5: Obtain the clusters as specified in Algorithm 1.
6: Calculate the intercluster and intracluster distances [13]
7: Calculate the DB index as per Eq. (1)
8: Calculate fitness value as 1/DB and maximize it.
9: end for
10: Apply crossover and mutation operator over the chromosomes.
11: NGEN–
12: end while
13: select the top 10 fittest chromosomes
Genetic Algorithm-Based Optimization of Clustering Data Points … 15

A small value of DB index would be most fit, hence to minimize DB index 1/DBi
is maximized (DB index calculated for discrete i). Thus, maximum fitness value
chromosome is selected to perform crossover and mutation [6–8]. Crossover is two-
point crossover and a fixed crossover rate of 50% is applied to be performed on
highly fit chromosome.
To incorporate variety, mutation is performed after crossover for a fixed muta-
tion rate of 10%. After performing one iteration, the above procedure is recurred for
a static number of iterations after which highly fit chromosomes denote the user-
defined bandwidth and s upon which the PP algorithm is used to get an optimal
number of clusters. Algorithm 2 presents the optimized PP algorithm through GA.
Lines 1 and 2 deal with the initialization of various parameters. Lines 3–11 iterate
over the number of generations, and evaluate the fitness value for every chromo-
some. Line 10 applies the biological operators like crossover and mutation over the
chromosomes to incorporate variety and produce another set of chromosomes for
the next generation. Line 13 selects the top 10 fittest chromosomes, giving us the
optimal value of bandwidth and s.

3.2 Result and Analysis

This section deals with the detailed analysis of the PP algorithm along with GA
for a favorable number of clusters for the two datasets; the synthetic dataset which
consists of 40 data points with pre-defined class labels; and the other one is a real
dataset of 62 data points which is Alizadeh V2 dataset of genetic representation of
lung cancer cells. Results were recorded for triangle and uniform kernel functions
for both synthetic and real datasets. The results for the real dataset for triangular
kernel consist of the number of clusters we get for each of s = 1, 7, and 62. The
user-defined parameters here are the bandwidth parameter and s. The analysis shows
that the number of clusters we get for each of the s parameters are approximately the
same. Tables 1, 2, 3 and 4 summarize the results for the same.
The PP algorithm gives the number of clusters for given values of bandwidth
parameters and s values; now GA is applied for getting the best optimal values

Table 1 The time (T), number of clusters (K), and fitness values (FV) are recorded for given
parameters when applied to a synthetic dataset of 40 data points for Uniform kernel
δ K T FV δ K T FV δ K T FV
s=1 s = 18 s = 27
6.3 25 0.285 0.066 6.3 25 0.265 0.066 6.3 25 0.281 0.0666
11.61 9 0.602 0.226 11.61 8 1.081 0.260 11.61 8 1.284 0.026
16.15 6 0.584 0.418 16.15 5 1.276 0.641 16.15 5 1.405 0.6425
30.38 3 0.439 0.426 30.38 2 1.274 0.019 30.38 2 1.235 0.019
16 S. Dalmia et al.

Table 2 The time (T), number of clusters (K), and fitness values are recorded for given parameters
when applied to a synthetic dataset of 40 data points for Triangle kernel
δ K T FV δ K T FV δ K T FV
s=1 s = 18 s = 27
6.3 25 0.880 0.0666 6.3 25 0.757 0.0666 6.3 25 0.774 0.0666
11.61 11 0.681 0.1592 11.61 8 1.067 0.26 11.61 8 1.092 0.26
16.15 8 0.556 0.2479 16.15 5 1.304 0.6415 16.15 5 1.257 0.6415
30.38 3 0.423 0.8410 30.38 3 0.776 0.8410 30.38 3 0.982 0.8410

Table 3 The time (T), number of clusters (K), and fitness values (FV) are recorded for given
parameters when applied to Alizadeh-V2 dataset for Uniform kernel
δ K T FV δ K T FV δ K T FV
s=1 s=7 s = 62
0.4 49 1.47 0.0225 0.4 50 1.47 0.0224 0.4 50 1.48 0.0214
0.59 29 1.43 0.0478 0.59 29 1.49 0.0478 0.59 29 1.57 0.0478
0.67 11 1.96 0.7857 0.67 10 3.19 0.5882 0.67 10 3.31 0.5882
0.79 3 1.99 0.75 0.79 3 2.02 0.6 0.79 3 2.21 0.6

Table 4 The time (T), number of clusters (K), and fitness values (FV) are recorded for given
parameters when applied to Alizadeh-V2 dataset for Triangle kernel
δ K T FV δ K T FV δ K T FV
s=1 s=7 s = 62
0.4 49 1.44 0.0224 0.4 49 2.54 0.0224 0.4 49 2.56 0.0224
0.5 29 1.44 0.0485 0.59 29 2 0.0485 0.59 29 2 0.0485
0.67 12 2 0.6 0.67 10 4.26 0.6 0.67 10 4.31 0.6
0.79 3 1.43 0.75 0.79 3 2 0.75 0.79 3 2.23 0.75

Table 5 The top chromosomes for each population set are recorded which consist of bandwidth
parameter and s values. The number of clusters (K) derived from them are also recorded. The results
are recorded for 5, 10, and 15 number of generations with a mutation rate of 10% and crossover
rate of 50%
Number of Generations = 5 Number of Generations = 10 Number of Generations = 15
δ s K δ s K δ s K
0.856 2 36 1.03 3 35 0.156 2 33
0.157 3 38 0.12 3 36 0.488 3 35
0.157 3 40 1.26 3 35 0.488 3 36
1.273 3 39 0.129 3 34 0.488 3 35
1.273 3 40 0.129 3 36 0.156 2 33
Genetic Algorithm-Based Optimization of Clustering Data Points … 17

for bandwidth and s. The results for the same have been summarized in Table 5.
The results are recorded for the top five chromosomes for 5, 10, and 15 number
of generations. The population size has been fixed to ten chromosomes, with each
chromosome consisting of two genes—one for bandwidth parameter and other for s
value. The number of clusters (K) related to these values is also recorded which give
us the most favorable number of clusters for the derived s and bandwidth parameter
through GA.

4 Conclusion and Future Work

The PP algorithm performs better than other classical clustering algorithms like
AP. When compared to the time taken, PP algorithm performs at par with other
algorithms. PP algorithm does well for synthetic as well as for a sparse real dataset
for any shape of the cluster. As an improvement over the PP algorithm, a novel
approach is formulated to calculate the optimal number of clusters by harnessing
the capabilities of the GA. At the core of this approach, we formulate the fitness
function such that the inter-cluster distance between data points is reduced and intra-
cluster distance in between clusters is increased. The fitness value of each cluster
model is recorded and a maximizing fitness function is applied to it to record the
best bandwidth parameter and ‘s’ value for which an optimal number of clusters is
obtained. For further optimizations and comparisons, different genetic operators can
be explored other than mutation and crossover. Parallelizing techniques for reducing
the time complexity of the proposed GA-based PP algorithm can be incorporated,
which is currently O(n3 ).

References

1. Gana, G., Zhangb, Y., Dey, D.K.: Clustering by propagating probabilities between data points.
Appl. Soft Comput. 41, (2016)
2. Frey, B.J., Dueck, D.: Clustering by passing messages between data points. Science 315 (2007).
www.sciencemag.org
3. van Dongen, S.: Graph clustering via a discrete uncoupling process? SIAM J. Matrix Anal.
Appl. 30(1), 121–141 (2008)
4. Ester, M., Kriegel, H.-P., Sander, J., Xu, X.: A density-based algorithm for discovering clusters
in large spatial databases with noise. KDD (1996)
5. Goldberg, D.E.: Genetic Algorithms in Search, Optimization and Machine Learning. Addison-
Wesley (1989)
6. Liu, Y., Ye, M., Peng, J., Wu, H.: Finding the optimal number of clusters using genetic algo-
rithms. In: IEEE Conference on Cybernetics and Intelligent Systems. Chengdu, pp. 1325–1330
(2008)
7. Bandyopadhyay, S., Maulik, U.: Nonparametric genetic clustering: comparison of validity
indices. IEEE Trans. Syst. Man Cybern. Part C: Appl. Rev. 31(1), 120–125 (2001)
8. Bandyopadhyay, S., Maulik, U.: Genetic clustering for automatic evolution of clusters and
application to image classification. Pattern Recogn. 35(6), 1197–1208 (2002)
18 S. Dalmia et al.

9. Lai, C.C.: A novel clustering approach using hierarchical genetic algorithms. Intell. Autom.
Soft Comput. 11(3), 143–153 (2005)
10. Kumsawat, P., Attakitmongcol, K., Srikaew, A.: A new approach for optimization in image
watermarking by using genetic algorithms. IEEE Trans. Signal Process. 53(12), 4707–4719
(2005)
11. Ramasubramanian, P., Kannan, A.: A genetic-algorithm based neural network short-term fore-
casting framework for database intrusion prediction system. Soft Comput. 10(8), 699–714
(2006)
12. Chang, Y.C., Chen, S.M.: A new query reweighting method for document retrieval based on
genetic algorithms. IEEE Trans. Evol. Comput. 10(5), 617–622 (2006)
13. Lin, H.J., Yang, F.W., Kao, Y.T.: An efficient GA-based clustering technique. Tamkang J. Sci.
Eng. 8(2), 113–122 (2005)
Another Random Scribd Document
with Unrelated Content
everything which might be said, thought, or done by any of his right
honorable feeders, he was not attentive to statements, as if they
required any consideration.
After much had been said by Lord Spoonbill as to the most
probable means of obtaining possession of the lovely one, and when
his lordship’s ingenuity in devising seem to be exhausted, and when
Colonel Crop seemed to think that it was necessary that he should
say something, then did he in his usual abrupt style of eloquence
say: “Perhaps, my lord, I had better bring the lady up to London with
me?”
Here we must explain, though it is a great pity that circumstances
compel us to state the fact, that this said gallant colonel had
previously been employed in that service by other right honorable
gentlemen, and therefore he regarded the matter lightly, and thought
that he had nothing else to do than state the terms and
establishment. But Lord Spoonbill did not apprehend that the lady in
question would be so lightly won.
“Ah, Crop!” said his lordship, “you don’t know the lady. She is
almost a prude; but I love her the more for her very obstinacy.”
The colonel had some notion of what was meant by a prude; but,
according to the best of his recollection, the young lady whom he
had seen at Neverden Hall was not mightily prudish in her manners.
And then the military man smiled, and replied to his lordship’s
observation: “Trust me, if you make an offer of a handsome
establishment, I will bring her up to town in a very short time.”
Lord Spoonbill was encouraged greatly by the smile which he saw
upon the colonel’s countenance, and began to think that his friend
Crop had really some reason for what he had said. And then in the
fullness of his heart his lordship went on to say:—
“Crop, my dear friend, you know that, situated as I am now, I can
do nothing for you; but it is my full intention, if I can get leave of my
father to go over to the other side, and then if Penelope is mine,
mine on any terms, you shall be promoted immediately, take my
word for it.”
 It might have been supposed that the mention of the name of
Penelope would have been the means of bringing the colonel to
suspect that there was some error in his apprehension; but the truth
is, that he paid no very particular attention to the name, and not
having the slightest suspicion of the possibility of an error, he never
thought of investigating the matter.
The right honorable lover, and his kind friend Crop, had not made
a late dinner, and it occurred to Lord Spoonbill that, as it would be a
pity to have travelled so many miles without even seeing the young
lady, he would pay a visit to the rectory in the evening. His lordship
very naturally thought that if it were true, as the colonel brought word
from his visit at Neverden, that Mr Darnley had actually and
notoriously declined the acquaintance with Penelope, the young lady
would by this time have ascertained that fact, and might not be quite
so difficult as heretofore. Still, however, his lordship could not quite
make up his mind whether to urge his suit merely as at first, or
whether he would deliberately offer her his hand and fortune.
As his mind was not able to arrive at a complete decision on that
point, he resolved to be governed by circumstances. That was a
wise resolution, and most of the species are governed by
circumstances, plebeians as well as patricians. We hardly know how
it can be avoided.
While his lordship was gone to the rectory, Colonel Crop was
occupied at the castle in preparing for his lordship’s journey, which
was destined, in obedience to the commands of Lord Smatterton, to
take place on the following morning.
Now when Lord Spoonbill drew nigh to the rectory, he could not
help recollecting that there had lived such a man as Dr Greendale,
and he could not but call to mind the pure and virtuous character of
that good man. There came again to the recollection of Lord
Spoonbill, the kind and gently-spoken admonitions of that departed
friend; and the feelings of his lordship were roused at the
remembrance of the solemn scene which he had witnessed, when
he saw that good man breathe his last. In these emotions there was
enough to disturb his lordship in his pursuit, but not enough to divert
him from it.
It was generally allowed by every one who knew him that Mr
Darnley, the rector of Neverden, was a very good and a
conscientious man, and that as a parish priest he was most
exemplary and active in the discharge of his duties. But there still
was in his character a dryness, a hardness, a coldness, a distance,
or a something, that rendered virtue not so amiable and contagious
in him as in the late Doctor Greendale. Mr Darnley preached most
unexceptionable sermons, sound in doctrine, honest in application,
always suited to the times, and most admirably adapted to the
capacity and comprehension of his hearers. He never talked
learnedly in his pulpit to gratify the taste of the great man of his
parish, to the neglect of the instruction of the poorer part of his
audience. In that, to be sure, there was not much merit, for neither
the literature nor the capacity of Sir George Aimwell was much
above the level of the farmers and their labourers. In Doctor
Greendale however there was something more than a merely
unexceptionable style of preaching. It is true indeed, that in
consequence of the peculiar employment in which so much of his life
was spent, and the theological controversies to which so much of his
attention was directed, he did occasionally introduce into the pulpit
controversial and abstract discussions, which were not very
intelligible even to himself. But then there was something so morally
beautiful and kindly impressive in his whole manner, tone, and
deportment, that his hearers used to observe that they always felt
themselves the better for hearing him preach, though they could not
always understand what he said, or divine the object of his
arguments. And let no one despise this feeling; for if good impulses
are given to the heart, why should we think slightingly and
disrespectfully of the means?
We hardly apologize for the above digression, knowing that those
of our readers who are most worth pleasing are gratified and
delighted at looking back upon the virtues and amiableness of the
late worthy rector of Smatterton. This digression also is serviceable
for the purpose of illustrating and explaining the state of mind in
which Lord Spoonbill was when he approached the rectory.
The impression produced upon the hereditary legislator was such,
that he almost resolved to offer marriage at once, formally and
decidedly, thinking that he should thereby cancel all his former
transgressions. But when arrived at the rectory, and finding that Mr
Primrose was not within, and that Penelope was alone, his lordship,
who was rather proud of his own sweet self, began to entertain less
respectful thoughts towards the young lady, and to imagine that she
must love him for himself alone.
Penelope, at sight of Lord Spoonbill at Smatterton—when she
thought, if she thought of him at all, that he was in London—
expressed great surprise, which his lordship presently removed by
saying, that his visit to the castle was with a view to some alterations
and improvements.
As the acquaintance between Robert Darnley and Miss Primrose
was now renewed, and as great part of the source of her uneasiness
was removed, the vivacity and cheerfulness which in former days
had been so becoming an ornament to her, had now returned;
therefore did she seem in the eyes of Lord Spoonbill more lovely
than ever. Considering herself now as clearly engaged to Mr
Darnley, she no longer regarded Lord Spoonbill as a suitor, and her
manner towards him was therefore the less reserved and
constrained. All this change in the young lady’s manner was by Lord
Spoonbill considered as being a corroboration of the report which
Colonel Crop had brought from Neverden.
And now his lordship would have saved himself a great deal of
trouble, mortification, and expense, had he renewed, as fifty times in
a minute he designed, his suit for the honor of the lady’s hand. For
such was Penelope’s simplicity and plainness of spirit that, had his
lordship renewed that subject, she would have told him explicitly how
insuperable an obstacle was in the way of his hopes. But he thought
that there was a possibility and a probability of the speedy return of
Mr Primrose, who would of course interrupt the discussion; and he
also thought if it were possible to possess himself of the young lady
without the encumbering ceremony of marriage, it would be so much
the better.
His lordship therefore, instead of affecting the sheepishness and
sighing of a sentimental lover, endeavoured to make himself
agreeable by the fascination of his fluent gabble, and he poured forth
a variety of idle prate, which we cannot repeat without the charge of
plagiarism. Had a third person been witness of this interview, he
might have thought them to be the best friends in the world.
Before Lord Spoonbill had taken his leave, Mr Primrose returned;
he had merely been gossipping with his neighbour Kipperson, and
the learned Zephaniah Pringle. Mr Primrose, knowing nothing of his
lordship’s intentions, and being as unsuspicious as Penelope that he
was the person by whose means the correspondence had been
interrupted, greeted the hereditary legislator with the utmost
cordiality and respect. The father expressed the same surprise as
the daughter had, at the presence of Lord Spoonbill at Smatterton,
and received the same lying explanation, and was equally well
satisfied with it.
Even had there sprung up in the mind of Mr Primrose any
suspicion whatever that his lordship’s visit to Smatterton had any
reference to Penelope, that thought would have been immediately
removed by Lord Spoonbill saying; “My visit to Smatterton is very
short, I arrived yesterday, and I return tomorrow morning.”
“Then we feel ourselves the more honored by this call;” replied Mr
Primrose.
His lordship smilingly returned for answer; “You may indeed, Mr
Primrose. I have been occupied at the castle all the morning, and
you are the only person in the country that I have called upon. I have
not even seen my good neighbour, Sir George Aimwell. But I shall
leave my friend Colonel Crop at the castle, and he must apologize
for me.”
Then after a short pause, with a little hesitation and a slight
change of tone, his lordship continued—“Perhaps you will excuse
me, Mr Primrose, if I take the liberty to enquire whether you have
had any more favourable account of the affairs of your banker, than
you had when I had the honor to see you in London?”
Lord Spoonbill asked this question with a tone of voice that
indicated great compassion, and seemed to express a hope and
earnest wish that the answer might be of a favorable complexion; but
it would have been most pleasing to his lordship to hear a long
drawn sigh and a lamentable tale of unfortunate losses and painful
disappointments. Instead however of a melancholy shake of the
head and a dismal story, Mr Primrose said, in reply, “I thank your
lordship for your kind enquiries, but I am most happy to inform you
that I had taken alarm too soon. I know nothing about the affairs of
my banker, and I care but little; for, upon farther examination, I found
that my agent had not consigned any very large amount to the
banker’s keeping.”
“I am exceedingly happy to hear it,” replied his lordship.
That was a lie, and his lordship knew it to be a lie. And his lordship
was also perplexed for want of some device by means of which he
might carry on his designs against Penelope. There appeared not to
be any very definite indications, either favorable or otherwise, if we
except, as above mentioned, that there was greater ease and
cheerfulness in the young lady’s manner than had been recently
manifested. All this his lordship’s vanity placed to his own account;
and he took his leave of Mr Primrose and Penelope with great
courtesy and remarkable self-satisfaction.
 CHAPTER III.
The Right Honorable Lord Spoonbill we will leave to travel to
London and a hundred miles farther, even down to the newly-
purchased borough, where we will leave him to make his insulting
and hypocritical bows to the poor, mean-spirited, bought and sold,
place-hunting, booby-praising, supple, transferable, soulless dolts,
who, being taught to fetch and carry, to stand on their hind legs, and
feed out of a great man’s trough, and to yelp loyalty or sedition, as
the case may be and their purchasers may dictate, suffer themselves
to be called “free and independent” without having spirit enough to
kick the right honorable cubs that call them so.
Colonel Crop remained at Smatterton, under a strict charge from
Lord Spoonbill to keep an eye on the young lady, and to use his best
endeavours to forward the designs of his lordship. This gallant officer
undertook his commission with great hopes of success, and with
perfect confidence in his own wisdom; and, in pursuance of his
precious scheme, made an early call on Sir George Aimwell on the
morning on which Lord Spoonbill took his departure from town.
Sir George and the colonel were very good friends. They had
many points of sympathy. They were both of them great admirers of
good cookery. They had both a great abhorrence of unnecessary
trouble. They both thought that the only use for a gentleman to make
of books was to stick them on a shelf, or lock them in a closet. They
both thought that the common people were very vulgar, insolent, low-
lived creatures, and almost wondered that nature had not formed
them to go on four legs. With many other sympathies of this nature
the gallant colonel and the worthy magistrate were mightily well
pleased with each other.
When therefore the colonel made his second visit to Neverden, the
baronet was delighted with the attention, and anticipated with much
satisfaction sufficient talk to keep him awake till dinner time.
 “So, colonel,” said the unpaid one, “you have the compassion to
come and see me now the shooting season is quite over. This I call
really an act of friendship. But did you actually know that yesterday
was the last day of the season?”
“’Pon honor, Sir George,” replied the half-paid one, “I never
thought about it; but if I had it would have made no difference.”
“You are very polite, colonel, and honest withal, and that cannot be
said of every one.”
The baronet having thus spoken laughed, and the colonel, to shew
his politeness, laughed too, but he did not know why, unless it were
for sympathy.
“But, colonel,” continued the gratuitous dispenser of hedge and
ditch and steel-trap and spring-gun justice, “what is the meaning of
this almost flying visit which my Lord Spoonbill has just now paid to
Smatterton? My people told me this morning that the young
gentleman was off by daybreak, posting back again to town.”
Colonel Crop looked wise, or screwed his features into that kind of
expression which he was pleased to think wise; and with that kind of
tone which almost asks for incredulity, and intimates that at least
only part of the truth is told, he replied: “His lordship came down to
give orders about some improvements to be made at the castle.”
This expression answered the purpose, and the curious baronet,
being from the nature of his office accustomed to the interrogative
system, proceeded with a sceptical smile to ask: “Ah, colonel, are
you sure that is all? Is not there some sly amour going on? His
lordship is a wild young dog, I have heard.”
Thereat the baronet chuckled, and the colonel smiled so gently
and so silently that Lord Chesterfield himself would not have
pronounced him vulgar or ill bred. And the two worthy ones looked at
each other very knowingly; and thus the colonel let out the secret
without speaking a word. Some persons may imagine that the mode
in which the baronet spoke of the amatory transgressions of Lord
Spoonbill was rather too flippant for a magistrate; and those persons
may perhaps be right; but let us not omit to tell those persons, that
this same magistrate, when engaged in the discharge of his
magisterial duties, and when sitting on the bench, looked as wise
and as grave and as demure as any magistrate need to look, and
gave to transgressors a goodly word of exhortation, and dwelt with
copious and convincing eloquence on the wickedness of violating
our most excellent laws.
Everybody in that neighbourhood remembers the impressive
admonition which he gave to an old man who was convicted at the
quarter sessions of having a bit of string in his pocket, and therefore
strongly suspected of a design of a malicious nature against the
game.
“John Carter,” said the worthy baronet, “let me address to you a
few words on the sin of poaching. Poaching, John Carter—is—is a
sin of which too many are guilty, owing to the lenity of our most
excellent laws. I think that if everybody thought, as I think, of the
moral heinousness of this offence, nobody would be guilty of it.
Poaching is not yet made felony; but there is no saying how soon it
may be made so, if the crime be persisted in. It is a moral offence of
the greatest enormity, and is one of those crying, national sins, which
may one day or other bring down the vengeance of heaven on our
guilty country. Now, John Carter, if you go to gaol for six months, I
hope the tread-mill and the chaplain will work a thorough reformation
in your morals.”
Much more to the same purpose did the worthy magistrate say to
John Carter; but it did, we fear, very little good, for John Carter and
the baronet’s last pheasant were missing on one and the same
morning:
This digression we have made for the purpose of shewing how
truly eloquent and morally impressive the baronet could be when the
enormity of the offence demanded an adequate severity of
reprehension. His worship would have thought the epithet “wild
young dog” a designation by far too lenient for a poacher. But the
crime of seduction, especially in an hereditary legislator, looked in
the sight of the unpaid one as comparatively a slight offence,
scarcely an offence at all.
 More than suspecting, from Colonel Crop’s look and manner, that
a pursuit of this nature had brought the heir of Smatterton down to
the venerable seat of his ancestors, the worthy baronet proceeded to
enquire of the young lord’s convenient friend more particularly
concerning the person whose charms had fascinated the young lord.
Now the gallant colonel was not quite certain that it would be proper
and becoming for him to communicate to the worthy magistrate the
whole truth, at least what he thought to be the truth; he therefore
hesitated and looked knowing: but he had no great tact or dexterity
in keeping secrets; so, after a few ineffectual attempts to hold his
peace and keep his counsel, he gave the worthy baronet to
understand that the lady with whom he had the pleasure of dining at
Neverden Hall, was the person on whom the hereditary legislator
had cast the eyes of affection.
“What!” exclaimed the astonished baronet, “Do I hear aright? Do
you mean to tell me that Lord Spoonbill has employed you to make
dishonorable proposals to a lady under my roof, a relation too of my
own?”
Colonel Crop was thunderstruck at the mode in which the
magistrate received the information; and was still more astonished to
hear that the lady was related to the baronet. And the colonel himself
was almost angry with Lord Spoonbill, for thinking of proposing such
an arrangement with a lady so respectably connected. But when the
colonel proceeded to state that his commission was to obtain the
lady on any terms, and when the baronet understood that his
lordship was so enamoured as to meditate matrimony, that mightily
altered the case.
Then the magistrate gave full and free permission to the colonel to
have access to the lady, and state with as much eloquence as he
pleased, the proposals which he might have to make. It is indeed
true, that Sir George could not help thinking, that if Lord Spoonbill
had been disposed to make an offer of his hand to Miss Glossop, it
would have been rather more respectful to “apply personally, or by
letter post paid,” than to use the intervention of a gentleman whose
services had been employed in negociations of no very reputable
nature. The baronet however conceded, that there might be on the
part of Lord Spoonbill a more than usual degree of want of ceremony
allowed, in consequence of the difference of rank between the
parties.
The very possibility of making a relation of his a countess,
reconciled the unpaid one to the negociation of the half-paid one.
The end sanctified the means.
Now occurred again to the baronet’s recollection the visit which
Miss Glossop had paid to Smatterton castle, and he concluded of
course that it was at this visit that the charms of Arabella’s voice and
person had taken captive the heart of Lord Spoonbill. At the
contemplation of the possibility that his high-spirited relative should
make a match with so great a man as the eldest son of an earl, the
worthy magistrate felt mightily gratified.
It is somewhat admirable that Sir George Aimwell should regard
with any great satisfaction a probable marriage commenced by
negociations of so equivocal a nature. But there is something in
dignity and in high rank so fascinating and overwhelming, that what
would be insolence in others is condescension in men of title and
rank. And it is also a truth that Sir George, though himself a baronet
or lordling, and bearing the same relation to nobility as velveteen
bears to velvet, yet looked upon nobility with greater reverence and
admiration than persons more remote from that high rank regard the
possessors of titles.
When Colonel Crop had thus commenced his great negociation,
by informing the magistrate of the business for which Lord Spoonbill
had employed him, the gallant officer betook himself to the quiet
seclusion of the great house, and amused himself with looking at fine
pictures, reading the backs of books, lugging up his shirt collar,
combing out his whiskers, listening to the creaking of his boots,
picking his teeth, rattling his watch-chain, looking at himself in the
glass, wondering what was going on in London, together with a
multitude of other pretty little amusements of the same description.
The half-paid hero was not sorry that his negociation received
such a direction as it did, from the discovery that the young lady in
question was related to Sir George Aimwell; and it occurred to the
negociator, that he might very soon write to his honorable employer
and inform him of the progress which had been made in this
important and momentous business. And with the exercise of a little
ingenuity, which nobody knew so well how to use as the gallant
colonel, the occupation of writing a letter might employ a whole
morning. But as the colonel’s present employment is merely
meditation seasoned and garnished with sundries of these little
amusements which we have recorded above, we need not detain our
readers for the present at Smatterton castle, but we will introduce
them to another mansion at Smatterton, where they will meet with
the personages mentioned in this narrative, engaged in the highly
honorable and delightful occupation of paying a tea visit to Honoria
Letitia Spoonbill.
Miss Spoonbill kept very little company; she had but one party in
the year, and that was a tea-party, which assembled precisely at six
and departed as precisely at ten; and, we may add, sat still and
stared at each other as precisely for the whole period of their visit.
In this visit to the spinster of noble family most of the distinguished
persons of Smatterton were engaged. And there was something
peculiarly interesting in this visit, seeing that Mr Primrose was at the
party for the first time, and Mr Zephaniah Pringle was paying his last
visit in Smatterton. The lady of the house was particularly active to
render herself agreeable to her party; and, in order to produce as
much as possible of that kind of stuff which the people who write
newspapers call “the feast of reason and the flow of soul,” Miss
Spoonbill directed very much of her conversation to Mr Primrose,
that from him she might gather something concerning the manners,
customs, religion, literature, cookery, and politics of the Hindoos; and
to Mr Zephaniah Pringle the lady was also interrogatively attentive,
that from his exhaustless store of reading and reflection something
might be brought out for the edification and instruction of her guests.
So that poor Mr Kipperson had not any opportunity of putting forth
his agricultural erudition, and was also compelled to be painfully
mute on the subject of science. But he had the happiness of
meditation and thought, and not all the Spoonbills in the universe
could deprive him of that. And he could not help pitying from the
bottom of his heart the darkness, stupidity, grossness, and delusion
of all those persons who had no ambition after pantology. But he
consoled himself with the thought, that truth is great and must
prevail; and he knew that pantology was true, and he was sure it
would prevail, and all the world would be enlightened ultimately, and
every man would be his own Encyclopedia, and instead of seeing
here and there inscribed on the monumental marble of an individual
or two, a record of “Here lies the celebrated and learned,” we should
soon have the happiness of seeing in every church-yard in the
kingdom one general inscription of “Here lie ten thousand
pantologists.”
We thought it our duty to mention all this, because, wherever Mr
Kipperson gave the honor of his presence, he always looked
important and felt himself to be so, even if circumstances sentenced
him to silence.
Miss Primrose was also of the party, and happy was it for her that
she was innocent of all suspicion of those idle tales which the people
of Smatterton had amused themselves with concerning herself. Miss
Spoonbill indeed had slightly heard mention of the rumour, but could
not believe it; for she was sure that she must have heard it from
better authority, if it had been true; supposing no doubt that her right
honorable relative Lord Spoonbill would not have kept her in
ignorance on such a matter as this.
As Mr Primrose by his lucid, ready, and intelligent answers to Miss
Spoonbill’s multitudinous questions concerning Oriental affairs gave
great satisfaction to the lady, she felt grateful to the new inhabitant of
Smatterton for the information which had been given; and, in order to
shew her gratitude, was pleased to administer to Mr Primrose a large
quantity of gratuitous advice, touching the destiny and disposal of his
only child.
It was distinctly and strongly in Miss Spoonbill’s recollection that
there had been a time, and that not far back, when it was thought
desirable by the friends of Miss Primrose to seek out for her some
occupation or employment. Now Miss Spoonbill had no knowledge
whatever of the circumstances under which Mr Primrose had
returned to England, nor had she any thought whatever on the
subject, but she took it for granted that the father could not be less
anxious to dispose of the young lady in some situation than the uncle
and aunt had been, therefore with most considerate condescension
did Miss Spoonbill discourse to Mr Primrose on the importance of
giving the young lady an occupation. But this conversation was not
audible to the whole party, and unfortunately the part which Mr
Primrose took in it was not audible even to Miss Spoonbill herself; for
the lady’s hearing was not very good, and Mr Primrose had such an
awkward and confused manner of whispering that he was hardly
intelligible. And such was the fertility of Miss Spoonbill’s imagination,
that it supplied all the deficiencies hearing or apprehension.
“And so, Mr Primrose,” gently and lowly said Miss Spoonbill, “you
have not found a situation for your daughter yet?”
“I thank you, madam,” replied Mr Primrose, “but there is no
occasion for it.”
“Oh yes, certainly,” said Miss Spoonbill, “I think you were perfectly
right. I by no means approve of the profession of a public singer. But
I am sure my cousin Lord Smatterton, or his amiable Countess,
might find some more suitable employment. I will write to them
myself. They are very benevolent people, but they need to be
reminded now and then.”
“I beg, madam,” returned Mr Primrose, “that you will not by any
means think of giving yourself that trouble; I have it in my power to
provide for my daughter.”
“Say nothing about the trouble, Mr Primrose,” continued Miss
Spoonbill; “I insist upon writing myself, and I think I have some
influence with the Earl and Countess. I assure you, Mr Primrose, I
entertain a very great regard for your daughter.”
Mr Primrose took all the pains he could to make the lady
understand him, but it was not pleasant for him to speak out loud in
the audience of all the party, and therefore he let the subject drop as
quietly as he could, resolving to take the first opportunity of
explaining himself more distinctly and audibly than he could when
within hearing of his daughter and many others.
 CHAPTER IV.
On the following morning, Miss Honoria put her threat in
execution, or, according to her own view of the matter, fulfilled her
promise of writing to her noble relatives, exhorting them to use their
best diligence to provide some occupation for Miss Primrose, better
suited to her retired habits and education than the profession of a
public singer. The letter was written and sent off early, and on the
same morning another letter was written by Colonel Crop, addressed
to Lord Spoonbill, informing his lordship of the progress which had
been made in the negociation. The colonel’s letter was but short. It
ran thus:—
“My lord,”
“I have the honor to inform your lordship, that there seems to be
no chance of obtaining the lady, but by marriage. I have conversed
about the matter with Sir George Aimwell, and he would not hear of
any other arrangement.”
“I have the honor to be, &c. &c.”
With the result of these letters, our readers will be acquainted in
due course. In the meantime the worthy baronet found it
inconvenient to keep from her ladyship the knowledge which he had
acquired from Colonel Crop, concerning the imagined attachment of
Lord Spoonbill to Arabella. The baronet’s anxiety to communicate the
secret to Lady Aimwell did not arise from any desire to gratify her
ladyship, but from a wish to mortify her. For she had never regarded
Arabella with any great degree of kindness, or behaved towards her
with much consideration and politeness. It seemed to be the
domestic system of Neverden Hall for Sir George and her ladyship
always to be diametrically opposed to each other in their likes and
dislikes. In fact, the only thing in which they seemed to agree, was in
a determination to disagree in everything.
 The worthy magistrate was not in the best of all possible humours
when the shooting season was over, and her ladyship, who made it a
frequent subject of sneering complaint that she had so little of her
husband’s company, was generally most out of humour when she
had most of his company. Lady Aimwell usually employed her time in
reading; and her favorite books were Stackhouse’s History of the
Bible, and Richardson’s Novels, and when employed in this pursuit,
she was very partial to an empty apartment or a silent companion.
But her most amiable lord and master did not always indulge his
better half with this desired luxury. So it came to pass that the
neighbours used in the plenitude of their wisdom, and in the
impertinence of their criticism, to remark that Sir George and Lady
Aimwell did not live happily together. But they who made that remark
were very superficial and unreflecting observers. It is not to be called
a strange notion, because it is a very common and almost universal
notion, that nobody can be happy who does not live according to our
own notions of happiness. To be sure they did not know how happy
they were, and while they were grumbling and growling at each other
they possessed a satisfaction in, and derived a pleasure from, that
growling and grumbling which they professed to regard as a
calamity. And we cannot think it possible that any beings in
possession of their senses would voluntarily and repeatedly do that
which afforded them no pleasure. Sir George and his lady grumbled
at one another, and they liked grumbling, and they were happy in
grumbling.
One morning, very soon after Colonel Crop had revealed to Sir
George the object which detained him at Smatterton, the baronet
introduced himself, unasked, into the drawing-room, where her
ladyship was sitting with her favorite folio displayed before her. And
her ladyship lifted up her eyes and saw the worthy baronet, and
spoke not to him, but continued her reading. The baronet also looked
at her ladyship and was also silent. The baronet walked up and
down the room with heavy monotonous tread, and ever and anon
her ladyship looked at him as if wishing that he would have the
goodness to take himself off. But the baronet would not take himself
off; and then her ladyship frowned and sighed; and in that sigh there
was nothing of tenderness; but its tone, if a sigh can have a tone,
was indicative of sulkiness and mortification. The baronet heard the
sigh and saw the frowns, but would not take any notice of them.
When the baronet was tired of pacing up and down the room, he
threw himself upon a sofa; and then her ladyship looked at the
baronet’s boots, and the boots bore evident symptoms of having
been in the stable, and as they came into seriously close contact
with the fine chintz cover of the sofa, her ladyship grew more sulky,
and from the thick smoke of her sulkiness there broke forth at length
the fire of angry words; and her ladyship said:
“I think dirty boots are hardly fit ornaments for sofas.”
“But suppose I think they are, my lady?” replied the baronet.
To that supposition her ladyship made no reply, and indeed there
was little that could be said on the subject. Passing on from the dirty
boots to other grievances, her ladyship said:
“And pray, Sir George Aimwell, may I be permitted to ask how long
we are to have the pleasure of Miss Glossop’s company?”
“Till she is married, my lady;” responded the worthy baronet.
“Till she is married!” echoed the astonished lady; “and when is that
to be? I never heard of any prospect of that event.”
“Then, if it will afford your ladyship any pleasure, I will tell you,”
replied the baronet.
“Indeed,” replied Lady Aimwell, “I have no wish to know anything
about the young lady’s affairs. I will venture to say, that Mr Robert
Darnley knows better than to make any proposal to her.”
“Very likely, my lady,” replied the magistrate, “but somebody else
may not know better. What think you of Lord Spoonbill as a husband
for this favorite young lady of yours?”
“Lord Spoonbill! Nonsense,” said her ladyship in a tone of most
exquisite contempt. “I never took his lordship for a conjuror, but he is
hardly fool enough for that.”
“I am not aware of your ladyship’s notion of conjurors,” replied the
baronet, “but I can tell you for a fact, that I had it from Colonel Crop
that Lord Spoonbill has it in his intention to offer marriage to your
favorite young friend, and the colonel is here to negociate.”
Thereupon her ladyship was angry, and assuming an almost
theatrical attitude, and pushing aside Stackhouse’s History of the
Bible, exclaimed, “Ridiculous! Contemptible! Colonel Crop is only
laughing at you, and endeavouring to make a fool of the silly girl. But
she has vanity enough to believe anything.”
It was only on very extraordinary occasions that Lady Aimwell
expressed herself with so much energy. And though the language
here recorded does not bear the semblance of mighty vehemence or
overpowering and impassioned eloquence, yet it was very vehement
and very eloquent compared to her usual mode of expression; and
her ladyship was almost exhausted by the energy with which she
spoke, and the baronet, whom in this sentence we can hardly call
worthy, felt no compassion for her ladyship’s sufferings under the
exhaustion of her eloquence; but on the contrary, coolly smiled at her
agitation, and sneeringly said:
“Let me beg of your ladyship not to fatigue yourself by such great
exertions. You had better compose yourself. Shall I bring you a
smelling bottle?”
Thus did the baronet make a mockery at her ladyship’s
nervousness; and having made that rude speech he forthwith quitted
the apartment, leaving the discussion unfinished and the particulars
untold.
His place in the drawing-room was soon supplied by Miss Arabella
Glossop; and this second intruder was scarcely more welcome than
the first. Her ladyship, however, was so far disturbed from her
reading, that now the great volume was closed. And it was very
polite and considerate in Miss Glossop not to suffer her ladyship to
replace the heavy book on the table where it usually lay, but her
ladyship silently and frowningly resisted Miss Glossop’s politeness,
so that the young lady was under the necessity of almost forcibly
taking the ponderous volume from the feeble hands that held it. But
there was no rudeness in the violence, it was rather the earnestness
of friendly officiousness.
 There are however some tempers and some states of mind which
convert almost everything into an affront; and even civilities
themselves, offered with all the considerateness and circumspection
imaginable or devisable, are wilfully and cruelly tortured into a wrong
meaning. This dexterity of misinterpretation Lady Aimwell possessed
in a degree not becoming her rank as the lady of a baronet and a
magistrate.
It must indeed be acknowledged, that there was some little more
excuse for Lady Aimwell than for Mrs Greendale, whose temper, our
readers will recollect, was of a similar description. But there was a
great and wide difference between the manners of Miss Primrose
and the general stile and manner of Miss Glossop. The former was
gentle, quiet and unobtrusive; the latter was noisy, vulgar and
affected. And there was also a mighty difference between Sir George
Aimwell and the late Dr Greendale. The baronet was coarse,
unfeeling and inconsiderate; while the doctor was meek, polite, and
generally attentive and considerate. It is not to be wondered at that
Lady Aimwell was occasionally morose and pettish. And above all
there was this difference in the lot of the two ladies, namely, that Mrs
Greendale was, with very few exceptions indeed, master and
mistress too, while poor Lady Aimwell was neither one nor the other;
her ladyship might well enough read Stackhouse and be sulky.
Miss Glossop, observing the unusual moroseness of her ladyship,
and not wishing to give offence directly or indirectly, thought it was
but a piece of civility to ask whether her ladyship was unwell. To this
courteous interrogation, her ladyship thought not fit to give a direct
and immediate answer; but being thus compelled to speak, she did
break out with mighty vehemence against poor Miss Glossop.
“I do not know, Miss Glossop,” exclaimed Lady Aimwell, “what may
be your ideas of propriety; but I know that, when I was young, if any
one had acted as you did at the castle the other day, it would have
been considered a very great breach of decorum.”
To this very abrupt and indefinite accusation Miss Glossop
pleaded not guilty, saying, “I do not know what your ladyship alludes
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com