Hybrid Intelligent Systems Ajith

Abraham
Visit to download the full and correct content document:
https://textbookfull.com/product/hybrid-intelligent-systems-ajith-abraham/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Intelligent Systems Design and Applications Ajith

Abraham

https://textbookfull.com/product/intelligent-systems-design-and-
applications-ajith-abraham/

Hybrid Intelligent Systems 19th International

Conference on Hybrid Intelligent Systems HIS 2019 held
in Bhopal India December 10 12 2019 Ajith Abraham

https://textbookfull.com/product/hybrid-intelligent-systems-19th-
international-conference-on-hybrid-intelligent-systems-
his-2019-held-in-bhopal-india-december-10-12-2019-ajith-abraham/

Hybrid Intelligent Systems 15th International

Conference HIS 2015 on Hybrid Intelligent Systems Seoul
South Korea November 16 18 2015 1st Edition Ajith
Abraham
https://textbookfull.com/product/hybrid-intelligent-systems-15th-
international-conference-his-2015-on-hybrid-intelligent-systems-
seoul-south-korea-november-16-18-2015-1st-edition-ajith-abraham/

Proceedings of the 16th International Conference on

Hybrid Intelligent Systems HIS 2016 1st Edition Ajith
Abraham

https://textbookfull.com/product/proceedings-of-the-16th-
international-conference-on-hybrid-intelligent-systems-
his-2016-1st-edition-ajith-abraham/
Intelligent Systems Design and Applications 19th
International Conference on Intelligent Systems Design
and Applications ISDA 2019 held December 3 5 2019 Ajith
Abraham
https://textbookfull.com/product/intelligent-systems-design-and-
applications-19th-international-conference-on-intelligent-
systems-design-and-applications-isda-2019-held-
december-3-5-2019-ajith-abraham/

Intelligent Systems Design and Applications: 18th

International Conference on Intelligent Systems Design
and Applications (ISDA 2018) held in Vellore, India,
December 6-8, 2018, Volume 1 Ajith Abraham
https://textbookfull.com/product/intelligent-systems-design-and-
applications-18th-international-conference-on-intelligent-
systems-design-and-applications-isda-2018-held-in-vellore-india-
december-6-8-2018-volume-1-ajith-abraham/

Intelligent Systems Design and Applications: 18th

International Conference on Intelligent Systems Design
and Applications (ISDA 2018) held in Vellore, India,
December 6-8, 2018, Volume 2 Ajith Abraham
https://textbookfull.com/product/intelligent-systems-design-and-
applications-18th-international-conference-on-intelligent-
systems-design-and-applications-isda-2018-held-in-vellore-india-
december-6-8-2018-volume-2-ajith-abraham/

Proceedings of the Third International Scientific

Conference “Intelligent Information Technologies for
Industry” (IITI’18): Volume 1 Ajith Abraham

https://textbookfull.com/product/proceedings-of-the-third-
international-scientific-conference-intelligent-information-
technologies-for-industry-iiti18-volume-1-ajith-abraham/

Proceedings of the Third International Scientific

Conference “Intelligent Information Technologies for
Industry” (IITI’18): Volume 2 Ajith Abraham

https://textbookfull.com/product/proceedings-of-the-third-
international-scientific-conference-intelligent-information-
technologies-for-industry-iiti18-volume-2-ajith-abraham/
Advances in Intelligent Systems and Computing 734

Ajith Abraham · Pranab Kr. Muhuri

Azah Kamilah Muda · Niketa Gandhi
Editors

Hybrid
Intelligent
Systems
17th International Conference on
Hybrid Intelligent Systems (HIS 2017)
Held in Delhi, India, December 14–16,
2017
Advances in Intelligent Systems and Computing

Volume 734

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: kacprzyk@ibspan.waw.pl
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.
The publications within “Advances in Intelligent Systems and Computing” are primarily
textbooks and 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.

Advisory Board
Chairman
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
e-mail: nikhil@isical.ac.in
Members
Rafael Bello Perez, Universidad Central “Marta Abreu” de Las Villas, Santa Clara, Cuba
e-mail: rbellop@uclv.edu.cu
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
e-mail: escorchado@usal.es
Hani Hagras, University of Essex, Colchester, UK
e-mail: hani@essex.ac.uk
László T. Kóczy, Széchenyi István University, Győr, Hungary
e-mail: koczy@sze.hu
Vladik Kreinovich, University of Texas at El Paso, El Paso, USA
e-mail: vladik@utep.edu
Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwan
e-mail: ctlin@mail.nctu.edu.tw
Jie Lu, University of Technology, Sydney, Australia
e-mail: Jie.Lu@uts.edu.au
Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico
e-mail: epmelin@hafsamx.org
Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazil
e-mail: nadia@eng.uerj.br
Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Poland
e-mail: Ngoc-Thanh.Nguyen@pwr.edu.pl
Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Kong
e-mail: jwang@mae.cuhk.edu.hk

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

Ajith Abraham Pranab Kr. Muhuri
•

Azah Kamilah Muda Niketa Gandhi

•

Editors

Hybrid Intelligent Systems

17th International Conference
on Hybrid Intelligent Systems (HIS 2017)
Held in Delhi, India, December 14–16, 2017

123
Editors
Ajith Abraham Azah Kamilah Muda
Machine Intelligence Research Labs Faculty of Information and Communication
Auburn, WA Technology
USA Universiti Teknikal Malaysia Melaka
Durian Tunggal, Melaka
Pranab Kr. Muhuri Malaysia
Department of Computer Science
South Asian University Niketa Gandhi
Chanakyapuri, Delhi Machine Intelligence Research Labs
India Auburn, WA
USA

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-3-319-76350-7 ISBN 978-3-319-76351-4 (eBook)
https://doi.org/10.1007/978-3-319-76351-4
Library of Congress Control Number: 2018935892

© Springer International Publishing AG, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved 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, express 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.

Printed on acid-free paper

This Springer imprint is published by the registered company Springer International Publishing AG
part of Springer Nature
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Welcome to the Proceedings of the 16th International Conference on Hybrid

Intelligent Systems (HIS 2017) held in South Asian University, Delhi, India, during
December 14–16, 2017. HIS 2017 is jointly organized by the Machine Intelligence
Research Labs (MIR Labs), USA, and South Asian University, Delhi, India.
Hybridization of intelligent systems is a promising research field of modern
artificial/computational intelligence concerned with the development of the next
generation of intelligent systems. A fundamental stimulus to the investigations of
hybrid intelligent systems (HISs) is the awareness in the academic communities that
combined approaches will be necessary if the remaining tough problems in compu-
tational intelligence are to be solved. Recently, hybrid intelligent systems are getting
popular due to their capabilities in handling several real-world complexities
involving imprecision, uncertainty, and vagueness. HIS 2017 builds on the success of
HIS 2016, which was held in Marrakech, Morocco, during November 21–23, 2016.
HIS 2017 was organized in conjunction with the 9th World Congress on Nature
and Bio-inspired Computing (NaBIC 2017). Nature and Bio-inspired Computing is
currently one of the most exciting research areas, and it is continuously demon-
strating exceptional strength in solving complex real-life problems. The main
driving force of the conference is to further explore the intriguing potential of
Bio-inspired Computing.
Many people have collaborated and worked hard to produce this year successful
HIS-NaBIC conferences. First, we would like to thank all the authors for submitting
their papers to the conference, for their presentations and discussions during the
conference. Our thanks to Program Committee members and reviewers, who carried
out the most difficult work by carefully evaluating the submitted papers. We have
two plenary speakers:
– Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico
– Alexander Gelbukh, Instituto Politécnico Nacional, Mexico City, Mexico
Thanks to all the speakers for their valuable time. The themes of the contribu-
tions and scientific sessions range from theories to applications, reflecting a wide
spectrum of coverage of the hybrid intelligent systems and computational

v
vi Preface

intelligence areas. HIS 2017 received submissions from over 24 countries, and each
paper was reviewed by at least five reviewers in a standard peer review process.
Based on the recommendation by five independent referees, finally 35 papers were
accepted for publication in the proceedings published by Springer, Verlag.
We would like to thank the Springer publication team for the wonderful support
for the publication of this volume.

Ajith Abraham
Pranab Kr. Muhuri
HIS 2017 General Chairs
HIS 2017 Organization

General Chairs

Ajith Abraham Machine Intelligence Research Labs, USA

Pranab Kr. Muhuri South Asian University, Delhi, India

Program Committee Co-chairs

Dilip Kumar Pratihar Indian Institute of Technology Kharagpur, India

Arpad Kelemen University of Maryland, USA
José Francisco Martínez National Institute of Astrophysics, Mexico
Trinidad
Millie Pant Indian Institute of Technology Roorkee, India
Nelishia Pillay University of Pretoria, South Africa

Advisory Board

Albert Zomaya The University of Sydney, Australia

Andre Ponce de Leon University of Sao Paulo at Sao Carlos, Brazil
F. de Carvalho
Bruno Apolloni University of Milano, Italy
Hideyuki Takagi Kyushu University, Japan
Imre J. Rudas Óbuda University, Hungary
Janusz Kacprzyk Polish Academy of Sciences, Poland
Javier Montero Complutense University of Madrid, Spain
Krzysztof Cios Virginia Commonwealth University, USA
Marina Gavrilova University of Calgary, Canada
Mario Koeppen Kyushu Institute of Technology, Japan
Mohammad Ishak Desa Universiti Teknikal Malaysia Melaka, Malaysia
Patrick Siarry Université Paris-Est Créteil, France

vii
viii HIS 2017 Organization

Ronald Yager Iona College, USA

Salah Al-Sharhan Gulf University of Science and Technology,
Kuwait
Sebastian Ventura University of Cordoba, Spain
Vincenzo Piuri Universita’ degli Studi di Milano, Italy

Publication Chairs

Azah Kamilah Muda UTeM, Malaysia

Niketa Gandhi Machine Intelligence Research Labs, USA

Local Organizing Committee

Q. M. Danish Lohani South Asian University, India

danishlohani@cs.sau.ac.in

Local Organizing Committee Members

Amit K. Shukla South Asian University, Delhi, India

Ashraf Zubair South Asian University, Delhi, India
Manvendra Janmaijaya South Asian University, Delhi, India
Amit Rauniyar South Asian University, Delhi, India
Rahul Nath South Asian University, Delhi, India
Sandeep Kumar South Asian University, Delhi, India
Taniya Seth South Asian University, Delhi, India
Deepika Malhotra South Asian University, Delhi, India

Web Service

Kun Ma University of Jinan, China

International Program Committee

Ajith Abraham Machine Intelligence Research Labs, USA

Alberto Cano University of Córdoba, Spain
Alicia Troncoso Pablo de Olavide University, Spain
Amparo Fuster Sabater Institute of Applied Physics (C.S.I.C.), Spain
Anne Laurent University of Montpellier II, France
Antonio Dourado University of Coimbra, Portugal
Antreas Nearchou University of Patras, Greece
Aswani Cherukuri Vellore Institute of Technology, India
HIS 2017 Organization
Azah Kamilah Muda
Choo Yun Huoy
Chun Wei Lin
Daniela Zaharie
Dilip Kumar Pratihar
Eiji Uchino
Elizabeth Goldbarg
Elpida Tzafestas
Enrique Dominguez
José Francisco Martínez
Trinidad
Isaac Chairez
Isabel Barbancho
Isabel S. Jesus
Ivo Pereira
José M. Merigó
José María Luna
José Ramón Villar
José Raúl Romero
Jose Santos
Jose Tenreiro Machado
José Valente De Oliveira
Julio Cesar Nievola
Kelemen Arpad
Korhan Karabulut
Kyriakos Kritikos
Lee Chang Yong
Leocadio G. Casado
Leticia Hernando
Lin Wang
Lubna Gabralla
Ludwig Simone
Luigi Troiano
Maria Nicoletti
Mario Giovanni C. A. Cimino
Mario Koeppen
Mark Embrechts
Martin Lukac
Matthias Becker
ix
Universiti Teknikal Malaysia Melaka, Malaysia
Universiti Teknikal Malaysia Melaka, Malaysia
National University of Kaohsiung, Taiwan
West University of Timisoara, Romania
Indian Institute of Technology-Kharagpur, India
Yamaguchi University, Japan
Universidade Federal do Rio Grande do Norte,
Brazil
University of Athens, Greece
Universidad de Málaga, Spain
National Institute of Astrophysics,
Optics and Electronics, Puebla, Mexico
Instituto Politécnico Nacional, Mexico
Universidad de Málaga, Spain
Instituto Superior de Engenharia do Porto,
Portugal
Instituto Politécnico do Porto, Portugal
University of Chile, Chile
University of Córdoba, Spain
University of Oviedo, Spain
University of Córdoba, Spain
Universidad de A Coruña, Spain
Instituto Superior de Engenharia do Porto,
Portugal
Universidade do Algarve, Portugal
Pontifícia Universidade Católica do Paraná,
Brazil
University of Maryland, USA
Yaşar Üniversitesi, Turkey
Foundation for Research and Technology
(FORTH) Hellas, Greece
Kongju National University, South Korea
University of Almería, Spain
The University of the Basque Country, Spain
Jinan University, China
Sudan University of Science and Technology,
Sudan
North Dakota State University, USA
University of Sannio, Italy
Universidade Federal de São Carlos, Brazil
University of Pisa, Italy
Kyushu Institute of Technology, Japan
Rensselaer Polytechnic Institute, USA
Nazarbayev University, Kazakhstan
Leibniz Universität Hannover, Germany
x HIS 2017 Organization

Millie Pant Indian Institute of Technology Roorkee, India

Mohammad Shojafar Sapienza University of Rome, Italy
Nabil Belgasmi Université de la Manouba, Tunisia
Niketa Gandhi Machine Intelligence Research Labs, USA
Noor Azilah Muda Universiti Teknikal Malaysia Melaka, Malaysia
Oscar Castillo Instituto Tecnológico de Tijuana, Mexico
Oscar Gabriel Reyes Pupo The University of Central Oklahoma, USA
Patrick Siarry Universit de Paris, France
Quan Guo Sichuan University, China
Ricardo Matsumura Araujo Federal University of Pelotas (UFPEL), Brazil
Rushed Kanawati Université Paris 13, France
Sansanee Auephanwiriyakul Chiang Mai University, Thailand
Sarjoun Doumit University of Cincinnati, USA
Sauptik Dhar University of Minnesota, USA
Serge Bezobrazov Brest State Technical University, Belarus
Sylvain Piechowiak Université de Valenciennes et du
Hainaut-Cambrésis, France
Tarun Sharma Amity University, Rajasthan
Thomas Hanne University of Applied Sciences Northwestern
Switzerland, Switzerland
Thomas Stutzle Université libre de Bruxelles, Belgium
Varun Kumar Ojha Swiss Federal Institute of Technology,
Switzerland
Yogan Jayakumar Universiti Teknikal Malaysia Melaka, Malaysia
Zeratul Izzah Mohd Yusoh Universiti Teknikal Malaysia Melaka, Malaysia

Additional Reviewers

Aurora Ramírez University of Córdoba, Spain

Ezequiel López-Rubio Universidad de Málaga, Spain
Antonio Galicia Pablo de Olavide University, Seville, Spain
José F. Torres Pablo de Olavide University, Seville, Spain
Rafael M. Luque-Baena Universidad de Málaga, Spain
Shadrack Maina Mambo Kenyatta University, Kenya
Contents

Neural Tree for Estimating the Uniaxial Compressive Strength

of Rock Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Varun Kumar Ojha and Deepak Amban Mishra
Fault Tolerant Multiple Synchronized Parallel Load Balancing
in Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
S. Sreelekshmi and K. R. Remesh Babu
Semi-Supervised Learning with the Integration of Fuzzy Clustering
and Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Indrajit Saha and Nivriti Debnath
GMM-KNN: A Method for Processing Continuous k-NN Queries
Based on The Gaussian Mixture Model . . . . . . . . . . . . . . . . . . . . . . . . . 30
Ziqiang Yu, Mincai Lai, and Lin Wang
Edge Detection for Cement Images Based on Interactive
Genetic Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Guangyue Gao, Lin Wang, Bo Yang, Liangliang Zhang, Fengyang Sun,
Ajith Abraham, and Shuangrong Liu
A Fast Satellite Image Super-Resolution Technique
Using Multicore Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Helal Uddin Mullah and Bhabesh Deka
Use of Cellular Automata to Predict Deforestation in Queretaro . . . . . . 61
Lourdes Margain, Alberto Ochoa, Lissette Martínez Almaguer,
and Rigoberto Velázquez
A Classification Algorithm for Assessing the Quality Criteria
for Business Process Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Fouzia Kahloun and Sonia Ayachi Ghannouchi

xi
xii Contents

Multilevel Image Thresholding Established on Fuzzy Entropy

Using Differential Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Abhishek Dixit, Sushil Kumar, Millie Pant, and Rohit Bansal
Analysis of GA Optimized ANN for Proactive Context Aware
Recommender System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Akshi Kumar, Nitin Sachdeva, and Archit Garg
Person Versus Non-person Classification of Twitter Handle . . . . . . . . . . 103
Himanshu Budania and Pramod Kumar Singh
Detection of Insulting Comments in Online Discussion . . . . . . . . . . . . . . 115
Anusha Gupta and Pramod Kumar Singh
A Model for Multi-processor Task Scheduling Problem
Using Quantum Genetic Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Rashika Bangroo, Neetesh Kumar, and Reya Sharma
Improving Modified Differential Evolution for Fuzzy Clustering . . . . . . 136
Jnanendra Prasad Sarkar, Indrajit Saha, Anasua Sarkar,
and Ujjwal Maulik
A Recent Brief Survey for the Multi Depot Heterogenous Vehicle
Routing Problem with Time Windows . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Bochra Rabbouch, Rafaa Mraihi, and Foued Saâdaoui
Hybridized Elephant Herding Optimization Algorithm
for Constrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Ivana Strumberger, Nebojsa Bacanin, and Milan Tuba
Drone Placement for Optimal Coverage by Brain Storm
Optimization Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Eva Tuba, Romana Capor-Hrosik, Adis Alihodzic, and Milan Tuba
Combining Fragmentation and Encryption to Ensure Big Data
at Rest Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Houyem Heni, Marwa Ben Abdallah, and Faiez Gargouri
Manufacturing Services Classification in a Decentralized Supply
Chain Using Text Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
M. D. Akhtar, V. K. Manupati, M. L. R. Varela, G. D. Putnik,
A. M. Madureira, and Ajith Abraham
Secured Biometric Template Matching by Using Linear
Discriminant Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Surbhi Vijh and Deepak Gaur
Contents xiii

Perturbation Based Efficient Crow Search Optimized FLANN

for System Identification: A Novel Approach . . . . . . . . . . . . . . . . . . . . . 204
Bighnaraj Naik, Debasmita Mishra, Janmenjoy Nayak, Danilo Pelusi,
and Ajith Abraham
Neurodegenerative Diseases Detection Through Voice Analysis . . . . . . . 213
Diogo Braga, Ana M. Madureira, Luis Coelho, and Ajith Abraham
Multi Document Summarization Using Neuro-Fuzzy System . . . . . . . . . 224
Muhammad Azhari, Yogan Jaya Kumar, Ong Sing Goh,
Ngo Hea Choon, and Aditya Pradana
A Comparison of Butterworth Noise Filteration Frequency
for Isotonic Muscle Fatigue Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Nur Shidah Ahmad Sharawardi, Yun-Huoy Choo, Shin-Horng Chong,
and Nur Ikhwan Mohamad
Recognizing Music Features Pattern Using Modified Negative
Selection Algorithm for Songs Genre Classification . . . . . . . . . . . . . . . . 242
Noor Azilah Muda, Azah Kamilah Muda, and Choo Yun Huoy
Preparation of ATS Drugs 3D Molecular Structure for 3D Moment
Invariants-Based Molecular Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . 252
Satrya Fajri Pratama, Azah Kamilah Muda, Yun-Huoy Choo,
and Ajith Abraham
Synchronization Technique via Raspbery Pi as Middleware
for Hospital Information System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Mohd Hariz Naim, Mohd Khanapi Abd Ghani,
Abd Samad Hasan Basari, Burhanuddin Mohd Aboobaider,
Lizawati Salahuddin, Wan Norhisyam Abd Rashid, and Gan Jun Kai
Validation Guideline for Small Scale Dataset Classification Result
in Medical Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Ee Kim Hwe and Zeratul Izzah Mohd Yusoh
A Grey Wolf Optimization Algorithm for Modular Granular
Neural Networks Applied to Iris Recognition . . . . . . . . . . . . . . . . . . . . . 282
Patricia Melin and Daniela Sánchez
Content Based Fraudulent Website Detection Using Supervised
Machine Learning Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
Mahdi Maktabar, Anazida Zainal, Mohd Aizaini Maarof,
and Mohamad Nizam Kassim
Wavelet Convolutional Neural Networks for Handwritten
Digits Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Chiraz Ben Chaabane, Dorra Mellouli, Tarek M. Hamdani,
Adel M. Alimi, and Ajith Abraham
xiv Contents

Model on Oil Platform Using Brain Storm Optimization Algorithm . . . 311

Lourdes Margain, Alberto Ochoa, Lissette Martínez Almaguer,
and Rigoberto Velázquez
Encryption Based on Neural Cryptography . . . . . . . . . . . . . . . . . . . . . . 321
Sayantica Pattanayak and Simone A. Ludwig
Genetic Algorithm to Generate Music Compositions:
A Case Study with Tabla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Subodh Deolekar, Ninad Godambe, and Siby Abraham
Navigation Control of Mobile Robot in Unknown Environments
Using Adaptive Cuckoo Search Algorithm . . . . . . . . . . . . . . . . . . . . . . . 341
Prases K. Mohanty, Shubhasri Kundu, and Harshal Dewang
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
 Neural Tree for Estimating the Uniaxial
Compressive Strength of Rock Materials

Varun Kumar Ojha1(B) and Deepak Amban Mishra2,3

1
ETH Zürich, Zürich, Switzerland
ojha@arch.ethz.ch
2
Indian Institute of Petroleum and Energy, Visakhapatnam, India
deepakamban@gmail.com
3
Institute of Geonics of CAS, Ostrava-Poruba, Czech Republic

Abstract. Uniaxial Compressive Strength (UCS) is the most important

parameter that quantifies the rock strength. However, determination of
the UCS in laboratory is very expensive and time-consuming. Therefore,
common index tests like point load (Is-50), ultrasonic velocity test (Vp ),
block punch index (BPI) test, rebound hardness (SRH) test, physical
properties have been used to predict the UCS. The objective of this
work is to develop a predictive model using a neural tree predictor
that estimates the UCS with high accuracy and assess the effectiveness
of different index tests in predicting the UCS of rock materials. UCS
and indices such as BPI, Is-50, SRH, Vp , effective porosity and density
were determined for the granite, schist, and sandstone. The constructed
model predicted the UCS with a high accuracy and in a quick time
(9 s). Additionally, the destructive mechanical rock indices BPI and Is-
50 proved to be the best index tests to estimate the UCS.

Keywords: Uniaxial compressive strength · Index tests

Rock materials · Heterogeneous flexible neural tree
Feature analysis

1 Introduction
Uniaxial compressive strength (UCS) is the most familiar parameter used in rock
engineering projects. In rock mass rating, proposed by Bieniawski [1], UCS is the
only parameter to assess the strength of rock material. However, high quality
machined specimens are required for determining UCS in the laboratory [2].
This makes the test expensive and time-consuming. Therefore, different empirical
predictive models are being used to estimate the UCS of rock materials indirectly
from the index tests, which require little or no specimen preparation and are
less expensive than the uniaxial compression test. Amongst different predictive
models, regression analysis is the most commonly used. Lists of such relationships
between the index tests and UCS obtained by different researchers have been
given in [3].
c Springer International Publishing AG, part of Springer Nature 2018
A. Abraham et al. (Eds.): HIS 2017, AISC 734, pp. 1–10, 2018.
https://doi.org/10.1007/978-3-319-76351-4_1
2 V. K. Ojha and D. A. Mishra

In the last decade, however, the use of computational intelligence (CI) meth-
ods to establish predictive models has gained much interest in the areas of rock
mechanics. A comprehensive list of such works is presented in Table 1. These
research works indicate the requirement of improvement in the CI methods to
estimate UCS from the index test results and to assess the effectiveness of indi-
vidual index tests for predicting the UCS of rock materials.

Table 1. Previous studies on the use of computational intelligence (CI) techniques in

estimating UCS and other rock engineering parameters from index tests.

Ref Rock types Inputs Output/s CI model/s r/r2

[3] Granite; Schist; BPI; Is-50; SRH; Vp UCS FIS (M) r = 0.99
Sandstone
[4] Sandstone; Hardness; Density; UCS NN r =0.96
Granodiorite Porosity; Grain size
[5] Greywacke Vp ; Is-50; BPI; TS UCS; EM FIS (M) r = 0.8
[6] Greywacke; UCS; Unit weight EM NN r = 0.82
Agglomerate
[7] Sandstone Packing Density; UCS NN r = 0.87
Concavity-convexity;
Quartz %
[8] Granite Porosity; Vp ; UCS WG NN; FIS r = 0.96;
r = 0.93
[9] Gypsum Water content; Is-50; UCS; EM NN; ANFIS r2 = 0.88;
Vp r2 = 0.94
[10] Travertine Vp ; Is-50; SRH; UCS; EM NN r2 = 0.64
porosity
[11] Sandstone; Shale; Porosity; Bulk UCS NN r = 0.91
Limestone density; Water
saturation
[12] Granite TS; BPI; Is(50); Vp UCS NN; ANFIS r = 0.6;
r = 0.69
[13] Granite Quartz %; Plagioclase UCS ANFIS r = 0.87
%; Orthoclase %
[14] Shale Dry density; Is-50; TFA PSO–NN r = 0.94
BTS; Vp ; SRH
[15] Granite Dry density, Vp ; Qtz UCS; EM NN r = 0.91
content; Plg content
[16] Granite; Schist; BPI; Is-50; SRH; Vp UCS FIS (TSK); NN r = 0.98
Sandstone r = 0.94
Note: FIS - Fuzzy Inference System; ANFIS - Adoptive Neuro-Fuzzy Inference
System; M - Mamdani, TSK – Takagi-Sugeno-Kang; PSO Particle Swamp Optimiza-
tion; BPI - Block Punch Index; Is-50 - Pointload Strength; SRH - Rebound Hard-
ness; Vp - Ultrasonic P-wave Velocity; TS -Tensile Strength; EM - Elastic Modulus;
WG -Weathering Grade; TFA - Internal Friction Angle.
 Neural Tree for Estimating the Uniaxial Compressive Strength 3

Although several predictive models have been proposed to estimate the UCS
from index tests, the influence of various index tests in predicting UCS is not
entirely understood. More specifically, so far, to the best of our literature sur-
vey, no study has been done on assessing the effectiveness of different index
tests. Moreover, it is always advantageous if development time and effort of the
predictive modeling can be reduced. To achieve these goals, the use of the neu-
ral network (NN) or similar CI methods on a limited set of experimental data
for predictive modeling can be a great benefit. Hence, the aim of this work is
to develop a predictive model that can estimate the UCS with a high accuracy
and assess the effectiveness of different index tests in predicting UCS of rock
materials.
The multiobjective heterogeneous flexible neural tree (HFNTM ) produces a
tree-like model [17], in which the nodes of the tree are similar to the NN nodes
was used in this study. Interestingly, HFNTM differs from NN [18] in its struc-
tural configuration, and it differs from the commonly used regression tree [19]
in its node type. Moreover, the tree-like structure in the HFNTM is created by
using multiobjective genetic programming (MOGP) [17,20]. Therefore, the pri-
mary advantages of using HFNTM over other CI techniques lie in its ability of
the automatic adaptation into the structure and the input feature selection.
Such an ability like HFNTM is necessary and has been successfully used in
several real-life applications [21]. However, this method has not been exploited
in estimating the UCS from index tests. In this work, index tests namely
block punch index (BPI), point load strength (Is-50), Schmidt rebound hard-
ness (SRH), ultrasonic P-wave velocity (Vp ), and physical properties namely
effective porosity (ηe ) and density (ρ) are determined and used for estimating
the UCS of granite, schist, and sandstone. An improved CI technique, HFNTM ,
was employed for this purpose and it was used to determine the effectiveness
of index tests BPI, Is-50, SRH, Vp , ηe , and ρ in the prediction of UCS of three
rocks. Such kind of individual test effectiveness assessment using HFNTM has
not been studied in the past, and it is the noble contribution of this research
along with the predictive modeling. The evaluation of individual tests provides
a detailed insight of UCS prediction of rock materials.

2 Materials and Methodology

Three completely different rock types, granite, schist, and sandstone were con-
sciously investigated in the laboratory to capture a broad scenario in evaluating
UCS. Core samples of granite, schist, and sandstone were collected from Malan-
jkhand Copper Project, Malanjkhand, India; UCIL mine at Jaduguda, India;
and SCCL, Kothagudem, India respectively. Each core sample (20 from each
rock type) was cut into four pieces as required for uniaxial compression, point
load, block punch, and Schmidt rebound hardness tests (as mentioned in [2]).
Results of the entire laboratory investigation were taken from [16]. During the
point load and block punch tests, few granite and sandstone specimens failed in
invalid failure modes. For these specimens, BPI and Is-50 were indirectly calcu-
lated using simple regression equations (developed from same rock types of the
4 V. K. Ojha and D. A. Mishra

same locality as this study) presented in [3]. A total 60 samples (20 from each
rock types) were used for the predictive modeling.

3 Predictive Models
3.1 Computational Intelligence Techniques
Discovering knowledge contained in data and developing predictive models are
vital tasks performed by the CI methods. Moreover, predictive modeling identi-
fies the underlying relationship between an input variable X = {x1 , x2 , . . . , xN }
and an output variable d = {d1 , d2 , . . . , dN } through the learning parameter w,
which defines the said relationship. A CI method finds the learning parameter w
by usually reducing the root mean square error (RMSE) e as per Eq. (1) between
the predicted output y = {y1 , y2 , . . . , yN } and the desired output d. Hence, the
learning of a CI method indicate the search for a proper learning parameter w.
In this study, the RMSE e was evaluated as:


1  N
e=  (yi − di )2 , (1)
N i=1

where N denotes the total training/test samples. Additionally, correlation coef-

ficient r as per Eq. (2) was also used in this study. The correlation coefficient r
computes the correlation between the predicted output y and the desired output
d, which is written as:
N
(yi − ȳ)(di − d̄)
r = N i=1 N (2)
i=1 (yi − ȳ) i=1 (di − d̄)
2 2

where ȳ and d̄ are the average of predicted and desired outputs. The correlation
coefficient ranges from −1 to 1, where a value 1 indicates the best performance.
In this study, CI methods such as fuzzy inference system (FIS), multilayer
perceptron (MLP), and adaptive neuro-fuzzy inference system (ANFIS) were
considered apart from the proposed HFNTM . The FIS, MLP, and ANFIS are
most widely used CI methods for modeling industrial and engineering prob-
lems [22]. The brief introduction to the mentioned CI methods are as follow:
• An FIS is a rule-based system, where a set of IF-THEN rules are designed
from the given input variables. Subsequently, the rules are inferred (conclusion
drawn from the rules set) to predict the output of the system for any given
input [23].
• An MLP is an NN model, which is an imitation of human-like learning. An
MLP is a layered network of neural nodes (computational node) arranged in
a layered structure. The nodes in an MLP are connected with synaptic links.
The primary form of MLP training for a given dataset is the discovery of
appropriate values for the synaptic connections [18].
• An ANFIS is a combination both FIS and NN-based systems, where an FIS
is designed as an NN-like model. ANFIS is typically a six-layered system in
which the layers indicate the IF and THEN parts of an FIS [24].
 Neural Tree for Estimating the Uniaxial Compressive Strength 5

3.2 Multiobjective Heterogeneous Flexible Neural Tree (HFNTM )

An HFNTM similar to any other CI methods often tries to minimize the error
e given in Eq. 1 for the given dataset by optimizing its parameter using some
learning algorithms (HFNTM employ multiobjective genetic programming and
differential evolution for this purpose). The structure of HFNTM is a tree-like
that has internal nodes as the computational nodes (analogous to MLP neurons),
branches (similar to MLP synaptic links), and leaf nodes (to represent input
variables).
Mathematically, an HFNTM , denoted as G, is a union of internal node V
and the leaf node T . The internal node V is a set of internal (computational)
nodes and, the leaf node T is a set of inputs [17]. Hence, an HFNTM G can be
expressed as:
U (k) U (k) U (k)
G = V ∪ T = {v2 , v3 , . . . , vtn } ∪ {x1 , x2 , . . . , xd } (3)
U (k)
where vi (i ∈ {2, 3, . . . , tn}) indicates an internal (computational) node
that takes two or more arguments; whereas, the leaf node takes no argument.
The function U (k) randomly invoke an activation function at a computational
node from a set of activation function: {Gaussian, tangent hyperbolic, bipolar
sigmoidal, unipolar sigmoidal, and Fermi}. HFNTM training has two aspects.
Firstly, discovering appropriate tree structure and secondly, the optimization of
tree parameters. These two HFNTM training parts are performed in two phases:

• Phase 1: Tree structure training using multiobjective genetic programming.

• Phase 2: Tree parameter training using differential evolution (DE) [25].

A detailed description of the aforementioned two phases are described in [17].

3.3 Input Feature Analysis

Feature analysis was conducted to assess the effectiveness of different index

tests in predicting UCS of rock materials. To perform such feature analysis,
20 HFNTM models were created. Each HFNTM models provided information
selected inputs and models prediction strength (in terms of RMSE). Hence, a
list of 20 models with the account of their chosen inputs and RMSE was pre-
pared.
To analyze the selected input features and their predictability, two per-
formance indicators feature selection rate R and predictability score P were
used [21]. Feature selection rate R is the measure of the total number of times a
specific set of input feature was occurred in the prepared list M = [m1 , . . . , m20 ]
of 20 models. Here, |M | indicating the size of the list was 20. Therefore, the
input-feature selection rate is computed as:
|M |
1 
Rj = I(mi = Aj ) (4)
|M | i=1
6 V. K. Ojha and D. A. Mishra

where Rj is the selection rate of j-th input feature set Aj ∈ P({BPI, Is-50,
SRH, Vp , ηe , ρ }), and function I(mi = Aj ) is a function that returns “1” if
j-th input-feature set Aj is selected by the i-th model mi , otherwise, it returns
“0.” Feature selection rate Rj = 1 is the highest (i.e., all the models selected
the input-feature set (Aj ) and Rj = 0 is the lowest (i.e., no model selected the
input-feature set Aj ).
The predictability score Pj of an input feature set Aj is on the other hand is
necessary to determine along with the selection rate Rj because the models in
the list may not be equal in their performances. To determine the predictability
score Pj of j-th input-feature set Aj , the performance Fj (typically the RMSE)
of the corresponding input-feature set Aj was at first computed as per Eq. (5):

|M |
ei × I(mi = Aj ) if|Aj | = 1
Fj = i=1 |M | |M | (5)
i=1 ei × I(mi = Aj )/ i=1 I(mi = Aj ) if|Aj | > 1

where ei indicates the RMSE of i-th model. The performance Fj for |Aj | = 1 is
the sum of RMSEs and Fj for |Aj | > 1 is the average RMSEs of all models that
selected a subset Aj . Accordingly, the predictability score Pj corresponding to
an input-feature set Aj was computed by normalizing the performance as [21]:

Fj
Pj = (6)
maxj=1 to z (Fj )

where function max(.) evaluate the maximum performance value among all Fj .
Analogous to the selection rate Rj , the predictability score Pj = 1 for an input-
feature set Aj describes the heights impact on the model’s predictability and
the score Pj = 0 describes the least impact on the model’s predictability.

4 Result and Discussion

4.1 Models Prediction
The developed best HFNTM model using the parameter setting mentioned in [17]
and two-fold training is shown in Fig. 1, where the leaf nodes indicate the input
features and the root node gives the predicted output UCS of the model. UCS
values estimated from the model are plotted against their corresponding exper-
imentally determined UCS values both for training and test data (Fig. 2). The
predictive performance of the developed model was examined through e, and r
computed as per Eqs. 1 and 2 and complexity χ (total number of parameters in
the developed models). The e, r, r2 , and χ obtained in this study was compared
with the e, r, r2 , and χ obtained from FIS (TSK), MLP and ANFIS models
as presented in [16] for the same dataset and same setting of training and test
set-up. The obtained results are presented in Table 2.
From the Fig. 2 and Table 2, it can be said that correlation coefficient r of the
developed HFNTM model is similar to that of the FIS (TSK), MLP and ANFIS
models. However, based on the RMSE of test data, it can be said that the
 Neural Tree for Estimating the Uniaxial Compressive Strength 7

UCS

v32

IS-50 IS-50
v32

BPI
v21 v21

BPI BPI BPI BPI

Fig. 1. Developed multiobjective heterogeneous flexible neural tree.

HFNTM model is better than other models. Besides the statistical parameters,
computational time of the HFNTM model was 9 s, and the complexity was 11
parameters for such a wide range of data obtained for three completely different
rock types. Considering all these factors, RMSE, r, r2 , and computation time,
it can be said the HFNTM is a very effective tool for determining the UCS
from index tests. Additionally, the HFNTM is a less complex model than the
other CI model. As shown in Table 2, the total parameters in the HFNTM were
smaller than the other CI models. Hence, HFNTM is a better model in both
implementation and computational points of view.

Table 2. Performance of the models for the selected training and test datasets.

FISa MLPa ANFISa HFNTM

Train Test Train Test Trainc Testc
e 9.54 14.33 16.9 7.87 13.72 10.92 5.87
r 0.98 0.96 0.94 0.99 0.97 0.98 0.99
r2 0.97 0.92 0.89 0.98 0.94 0.97 0.99
b b
χ 5800 25 44 11
Note: a results of FIS, ANN, and ANFIS were taken
from [16]; b approximate calculation; c results of the
proposed method.

4.2 Feature Analysis Results

A total 20 models were created using HFNTM for feature analysis. Since the
models were created by employing an evolutionary algorithm, the inputs that
contributed most towards UCS prediction were primarily selected. Hence, a list
8 V. K. Ojha and D. A. Mishra

Fig. 2. Estimated UCS plotted against measured UCS: training (left) and test (right).

of selected inputs by each model was prepared with the account of their RMSEs.
Afterward, a comprehensive feature analysis of all 6 input features was per-
formed. For the input feature analysis, two performance indicators, feature selec-
tion rate R as defined in Eq. (4) and feature predictability score P as defined in
Eq. (5), were adopted.
The effectiveness of individual index tests BPI, Is-50, SRH, Vp , ηe and ρ
in estimating the UCS were examined through R and P and setting |Aj | = 1.
Table 3 presents the results of feature analysis carried out for the all 6 individual
inputs. The input features BPI and Is-50 represent the destructive mechanical
indices; whereas, SRH and Vp represent the non-destructive rock indices. Effec-
tive porosity (ηe ) and density (ρ) are the determined physical properties of the
concerned rocks. It can be observed that R and P of destructive mechanical
indices BPI and Is-50 were much higher than that of R and P values of non-
destructive and physical rock indices.

Table 3. Significance of individual input features.

# Input features Selection rate (R) Predictability score (P )

1 A1 = {BPI} 0.862 0.899
2 A2 = {Is-50} 0.959 1
3 A3 = {SRH} 0.525 0.547
4 A4 = {Vp } 0.147 0.154
5 A5 = {ηe } 0.339 0.354
6 A6 = {ρ} 0.574 0.599

Among the individual input features, R and P of Is-50 are 0.959 and 1 respec-
tively. Next in the list is BPI with R and P of 0.862 and 0.899 respectively. From
the performed feature analysis, it can be said that Is-50 is the best index to pre-
dict the UCS of rock materials. Since BPI also has significantly high R and P ,
we can say that destructive mechanical rock indices are the best proxy for esti-
mating the UCS of the rock material. This finding is also in accordance with the
experimental findings of [26].
 Neural Tree for Estimating the Uniaxial Compressive Strength 9

5 Conclusions
The experimental program included the determination UCS, BPI, Is-50, SRH,
Vp , porosity and density of three different rock types (granite, schist, and sand-
stone). To estimate the UCS from the index tests, a multiobjective heterogeneous
flexible neural tree (HFNTM ) model was proposed. The inputs and outputs for
the developed HFNTM model were used from the rock materials experimental
results. The constructed model efficiently estimated the UCS based on the infor-
mation gathered from the experimental data in a very quick time (9 seconds).
Additionally, the HFNTM was a less complex model than the other CI model.
Therefore, this model can be effectively used in estimating the UCS. Developed
HFNTM also assess the effectiveness of different index tests in predicting UCS of
rock materials with the help of feature selection rate R and predictability score
P . From R and P values of the individual input features (index test results), it is
found that Is-50 is the best proxy for the UCS. Among the different types of index
tests—destructive indices, non-destructive indices, and physical properties—the
destructive mechanical rock indices BPI and Is-50 are found to be the best index
tests to estimate the UCS.

References
1. Bieniawski, Z.T.: Engineering Rock Mass Classifications, p. 251. Wiley, New York
(1989)
2. ISRM: The complete ISRM suggested methods for rock characterization, testing
and monitoring. In: Ulusay, R., Hudson, J.A. (eds.) Suggested Methods Prepared
by the Commission of Testing Methods, Kozan Ofset, Ankara, ISRM, 19742006.
Compilation Arranged by the ISRM Turkish National Group (2007)
3. Mishra, D.A., Basu, A.: Estimation of uniaxial compressive strength of rock mate-
rials by index tests using regression analysis and fuzzy inference system. Eng. Geol.
60, 54–68 (2013)
4. Meulenkamp, F., Grima, M.A.: Application of neural networks for the prediction
of the unconfined compressive strength (UCS) from Equotip hardness. Int. J. Rock
Mech. Min. Sci. 36, 29–39 (1999)
5. Gokceoglu, C., Zorlu, K.: A fuzzy model to predict the uniaxial compressive
strength and the modulus of elasticity of a problematic rock. Eng. Appl. Artif.
Intel. 17(1), 61–72 (2004)
6. Sonmez, H., Gokceoglu, C., Nefeslioglu, H.A., Kayabasi, A.: Estimation of rock
modulus: for intact rocks with an artificial neural network and for rock masses
with a new empirical equation. Int. J. Rock Mech. Min. Sci. 43, 224–235 (2006)
7. Zorlu, K., Gokceoglu, C., Ocakoglu, F., Nefeslioglu, H.A., Acikalin, S.: Prediction
of uniaxial compressive strength of sandstone using petrography-based models.
Eng. Geol. 96, 141–158 (2008)
8. Gokceoglu, C., Zorlu, K., Ceryanc, S., Nefeslioglu, H.A.: A comparative study on
indirect determination of degree of weathering of granites from some physical and
strength parameters by two soft computing techniques. Mater. Charact. 60, 1317–
1327 (2009)
10 V. K. Ojha and D. A. Mishra

9. Yilmaz, I., Yuksek, G.: Prediction of the strength and elasticity modulus of gypsum
using multiple regression, ANN and ANFIS models. Int. J. Rock Mech. Min. Sci.
46, 803–810 (2009)
10. Dehghan, S., Sattari, G.H., Chehreh, C.S., Aliabadi, M.A.: Prediction of uniax-
ial compressive strength and modulus of elasticity for Travertine samples using
regression and artificial neural networks. Min. Sci. Tech. 20, 41–46 (2010)
11. Rabbani, E., Sharif, F., Kooliv, M., Salooki, M.A.: Application of neural network
technique for prediction of uniaxial compressive strength using reservoir formation
properties. Int. J. Rock Mech. Min. Sci. 56, 100–111 (2012)
12. Yesiloglu-Gultekin, N., Gokceoglu, C., Sezer, E.A.: Prediction of uniaxial compres-
sive strength of granitic rocks by various nonlinear tools and comparison of their
performances. Int. J. Rock Mech. Min. Sci. 62, 113–122 (2013a)
13. Yesiloglu-Gultekin, N., Sezer, E.A., Gokceoglu, C., Bayhan, H.: An application
of adaptive neuro fuzzy inference system for estimating the uniaxial compressive
strength of certain granitic rocks from their mineral contents. Expert Syst. Appl.
40, 921–928 (2013b)
14. Armaghani, D.J., Hajihassani, M., Bejarbaneh, B.Y., Marto, A., Mohamad, E.T.:
Indirect measure of shale shear strength parameters by means of rock index tests
through an optimized artificial neural network. Measurement 55, 487–498 (2014)
15. Armaghani, D.J., Mohamad, E.T., Momeni, E., Narayanasamy, M.S., Amin,
M.F.M.: An adaptive neuro-fuzzy inference system for predicting unconfined com-
pressive strength and Youngs modulus: a study on Main Range granite. Bull. Eng.
Geol. Environ. 74, 1301–1319 (2015)
16. Mishra, D.A., Srigyan, M., Basu, A., Rokade, P.J.: Soft computing methods for
estimating the uniaxial compressive strength of intact rock from index tests. Int.
J. Rock Mech. Min. Sci. 80, 418–424 (2015)
17. Ojha, V.K., Abraham, A., Snášel, V.: Ensemble of heterogeneous flexible neural
trees using multiobjective genetic programming. Appl. Soft. Comput. 52, 909–924
(2017)
18. Haykin, S.: Neural Networks and Learning Machines, 3rd edn. Prentice Hall, Upper
Saddle River (2009)
19. Kohavi, R., Quinlan, J.R.: Data mining tasks and methods: classification: decision-
tree discovery. In: Handbook of Data Mining and Knowledge Discovery, pp. 267–
276. Oxford University Press (2002)
20. Chen, Y., Yang, B., Dong, J., Abraham, A.: Time-series forecasting using flexible
neural tree model. Inf. Sci. 174, 219–235 (2005)
21. Ojha, V.K., Schiano, S., Wu, C.Y., Snášel, V., Abraham, A.: Predictive modeling
of die filling of the pharmaceutical granules using the flexible neural tree. Neural
Comput. Appl., 1–15 (2016)
22. Lam, H.K., Nguyen, H.T.: Computational Intelligence and Its Applications: Evolu-
tionary Computation, Fuzzy Logic, Neural Network and Support Vector Machine
Techniques. World Scientific, Singapore (2012)
23. Rezaee, B., Zarandi, M.F.: Data-driven fuzzy modeling for Takagi-Sugeno-Kang
fuzzy system. Inf. Sci. 180(2), 241–255 (2010)
24. Jang, J.S.: ANFIS: adaptive-network-based fuzzy inference system. IEEE Trans.
Syst. Man Cybern. 23(3), 665–685 (1993)
25. Storn, R., Price, K.: Differential evolutiona simple and efficient heuristic for global
optimization over continuous spaces. J. Global Optim. 11, 341–359 (1997)
26. Mishra, D.A., Basu, A.: Use of the block punch test to predict the compressive
and tensile strengths of rocks. Int. J. Rock Mech. Min. Sci. 51, 119–127 (2012)
 Fault Tolerant Multiple Synchronized Parallel Load
Balancing in Cloud

S. Sreelekshmi ✉ and K. R. Remesh Babu ✉

( ) ( )

Government Engineering College Painavu, Idukki 685603, India

sreelekshmisuresh94@gmail.com, remeshbabu@yahoo.com

Abstract. Cloud computing offers on-demand access to a shared set of resources

over the internet at lower cost. The advantage of cloud resources is that it can be
easily provisioned, configurable, and managed with minimal management efforts.
Proper load balancing is an important task in maintaining fault tolerance and
Quality of Service (QoS) in the cloud. A load balancer accepts incoming user
requests and distributes this workload across multiple Virtual Machines (VMs)
using various methods. In a single load balancer system, if the load balancer is
down none of the user tasks can’t be processed, even when the servers are ready
to process it. This paper proposes a model that will avoid the single point of failure
by using multiple load balancers. In this method, service of one load balancer is
borrowed or shared among other load balancers when any correction is needed
in the estimation of load. This improves fault tolerance of the cloud eco system
and assists in cluster capacity management.

Keywords: Cloud computing · Multiple load balancer · Fault tolerant · QoS ·

Resource allocation

1 Introduction

With the emergence of cloud computing more and more business organizations are
moving towards this platform due to its low cost, simple configuration, and virtually
unlimited resource pool with on-demand provisioning. The performance of the cloud
eco system can be enhanced if the scheduling of resources is properly done. Resource
scheduling with load balancing is one of the best methods for improving the cloud
performance. The researchers have proposed several methods for optimal scheduling of
resources in the cloud.
Even though the features mentioned above are attractive, a fault tolerant mechanism
for undisrupted performance of cloud services is the most essential one. Load balancing
mechanism will improve the performance by efficient distribution of workloads across
multiple computing resources. The load balancer distribute the load evenly along cluster
members. Load balanced cluster is an abstraction for a set of identical processors, that
host same set of services. A simple cluster with a load balancer and respective cluster
members is shown in Fig. 1. Here n servers are managed by a single load balancer. When

© Springer International Publishing AG, part of Springer Nature 2018

A. Abraham et al. (Eds.): HIS 2017, AISC 734, pp. 11–19, 2018.
https://doi.org/10.1007/978-3-319-76351-4_2
12 S. Sreelekshmi and K. R. Remesh Babu

this single load balancer fails, functioning of the entire system collapses due to the non-
availability of the load balancer cum dispatcher. This causes financial as well as credi‐
bility loss for the cloud provider.

Fig. 1. Simple cloud load balancing

In dynamic cloud the predominant issue is load balancing. As in a distributed envi‐

ronment, load balancing mechanism distributes the dynamic workload evenly across all
available nodes in the system to avoid a situation where some nodes are heavily loaded
while others are idle or doing little work. It helps to attain increased satisfaction for the
customers and high resource utilization, consequently improving overall performance
and profit of the provider. In traditional method if the load balancer is down, then entire
process crashes even when processors are ready to executes the tasks, resulting in a
single point system failure.
In mission critical applications single point failures should be avoided. If we adopt
multiple load balancers, it will increase fault tolerance of the system. This paper proposes
a fault tolerant multiple synchronized parallel load balancers to improve fault tolerance
and efficient cluster capacity management. One similar approach is Sliding window
based self-learning and adaptive load balancer (SSAL) [6] is an observation based load
balancer that can produce optimal throughput in both stable and unstable environments.
SSAL monitored the performance of cluster members in every feedback interval. The
mechanism is trying to overcome the problems due to single point failure. It is used to
make corrections in load distribution model.
The main contribution of this paper is mitigation of single point of failure by the use
of multiple systems in parallel. Here the load balancing information is shared among all
load balancers available in the eco system and also provided an analysis to find out
cluster capacity needed for better performance of the system.
This paper is organized as follows. Similar works already proposed by the
researchers are reviewed in Sect. 2. Problem identification is given in Sect. 3. The
detailed design and explanation of the proposed method is described in Sect. 4. The
Sect. 5 covers the performance analysis and the paper concludes in Sect. 6.
 Fault Tolerant Multiple Synchronized Parallel Load Balancing in Cloud 13

2 Related Works

Load balancing is a cumbersome task in cloud resource allocation. The load is unpre‐
dictable in cloud computing and it may vary depending on the demand for a particular
service. Intelligent load balancing mechanisms are needed for ensuring better perform‐
ance and QoS. Load balancing algorithms are classified into static and dynamic, bio-
inspired, and game theory based. The common static algorithms are random, round
robin, min-min, min-max and weighted round robin algorithms. The paper [1] presents
a generic load balancing scheme, which separates the allocating and migrating processes
while preserving a guaranteed level of service.
An idea to make load balancing more dynamic for better QoS management of multi-
instance applications in the cloud is presented in [2]. This approach uses a load balancer
equipped with a queue for incoming user requests to limit its number, that can be effec‐
tively sent and stored in queues. Content-aware load balancing [3] is another approach.
This load balancer manages a farm of web servers and differentiates the service required
by workload and controls admissions under extreme overload. Multi-agent based distrib‐
uted load balancing framework for cloud platform [4] uses workload prediction tech‐
nology and threshold buffering strategies to reduce the migration of VMs. The paper [5]
gives an idea for accessing resources through the internet and makes use of Max-Min
and Min-Min algorithms. These algorithms are implemented based on the study of
RASA algorithm. RASA method provides better resource utilization and response time.
A fully distributed load balancing algorithm [7] is presented to cope with the load
imbalance and single point failure.
A comprehensive discussion and comparison of latest load balancing algorithms are
done in [8, 20] and these studies provides an overview of the latest approaches in this
field. Nature inspired algorithms are good candidates for solving dynamic load balancing
in cloud environment [9, 15]. Other major approaches are based on QoS [10, 11, 18,
19], Ant colony [12, 17], weighted load balancing [13], Genetic Algorithm (GA) [14]
and PSO [16] enabled load balancing methods.

3 Problem Identification

One of the main features of cloud is on-demand scalable computing at any time. In cloud,
there is no explicit knowledge for the customer about where their task is being executed
and in which server. Cloud providers are trying to offer fault tolerant services to their
customers. But the single point failures are one of the barriers for attaining fault tolerant
continuous service. Since the load balancer is responsible for distributing the tasks
received from the end users to any one of the processors, the processing of tasks will be
halted when the load balancer is down, even when the processor is ready to execute it.
This may be due to hardware failures like, server crashes, network problems, power
failures or disk crashes. Software failures like directory/proxy server crash and database
corruption also result in single point failure. So to address these failures, a suitable
cooperative mechanism is needed for fault tolerant cloud service.
14 S. Sreelekshmi and K. R. Remesh Babu

4 System Design

The proposed system contains a number of schedulers (load balancers), each capable of
balancing the task across multiple processors. These schedulers interacts each other for
gathering information about the tasks status and tasks in the input queue. They are also
able to distribute the tasks to other processors in data center based on known capability
of each processor. After the execution of each process, it can generate a feedback based
on the current capacity of each server. The capacity calculation is done in fixed time
interval based on the number of tasks executed by the processor and tasks pending in
the queue.
This frequent monitoring and cooperative load balancing ensure desired QoS to the
end users. In the proposed cooperative load balancing, none of the load balancers are
overloaded due to sharing of information about tasks already completed, being executed
or waiting in the queues. The architecture of the proposed method is shown in Fig. 2.
The detailed explanation is given in the next sub sections.

Fig. 2. System model

4.1 Task Handling

A set of unique tasks from the end users are to be handled by a task handler, in which
identical tasks are eliminated and remaining are stored in it. Hence it can reduce the
overhead of the entire process by removing duplicate tasks. Then the tasks are distributed
among different load balancers on the basis of round robin scheme. Once the scheduler
is down, the tasks stored in the queue are transferred back to the request handler. That
is task handler stays active until the completion of processes at each scheduler.

4.2 Load Balancing and Capacity Calculation

A set of tasks is given to each load balancer, which stores them in an output queue. There
exists a dispatcher for distributing the requests to different processors based on
Another random document with
no related content on Scribd:
 XV
AS IT WAS IN MY DAY

I am like the deaf old lady who, when asked why she took a box at
the opera when she could not hear, replied, “I can see.” So it is on
piazzas at summer hotels, I do not overhear remarks, so perforce
the pleasure of gossip is denied me, but “I can see,” and no doubt do
observe more than those who have the other faculty to play upon;
also I see and moralize. Last summer in the mountains didn’t I see
young girls, young society girls, educated girls who ought to have
known better, with bare heads and bare arms playing tennis in the
hot sun; and, worse still, racing over the golf links? I could see them
from my window, equally exposed, chasing balls and flourishing
clubs. The sun in August is pitiless even on those breezy mountains,
so I was scarcely surprised when one young girl was overcome by
heat and exposure, and was brought to her mother at the hotel in a
passing grocer’s cart or lumber wagon. I tell my grandchildren who
want to “do like other girls” that is not the way “other girls” did in my
day. Grandma may be so old that she forgets, but she moralizes all
the same. These athletic girls come back to city homes so sunburnt
and with such coarse skin they have to repair to a skin specialist,
and have the rough cuticle burnt off with horrid acids, and be
polished up before the society season opens.
There are, of course, extremes, but years ago young ladies took
more care of their complexions and of their hair, too. Years back of
years, I don’t know how they did. In my day we girls loved to visit the
granddaughter of a voluble dame and listen to the old lady’s talk, just
like I am talking now. She thought we were criminally careless with
our “skins,” as she called it. Why, when she was young, her skin was
so thin and clear that “one saw little blue veins meandering her
neck.” We always heard something as reminiscent in that house to
laugh over till we saw the old lady again, and heard something
equally remarkable of her youth. She was living in the past, as I am
now, as I return to my experiences. One young girl visited me, ever
so many years ago, who wore one of those awful, long, scoop
sunbonnets all the time she was not at table or in bed. She looked
like the proverbial lily. I used to wish she would take off that
sunbonnet and say something, for she was dumb as a lily. I have
entirely forgotten her name, though she was my guest for a whole
stupid week; but I recall she was a relative or friend of the Morses. I
don’t know Mr. Morse’s name; he was called Guncotton Morse, for
he invented an explosive of that name, which the United States
Government appropriated during the war.
Years after this young girl’s visit to me I called on the charming
Morse family in Washington. He was then urging his “claim.” Every
Southerner in Washington was after a “claim” at that time. I nearly
broke my neck falling over a green china dog or a blue china cat in
their dark parlor. Enterprising Morse barricaded himself behind his
explosive, but I think he failed in his fight. I find I have wandered
from the girls having their skins burned off to the Morses and their
blue china cat!... In my days there were no specialists except cancer
doctors. I think they always flourished—there were no skin
specialists. A doctor was a doctor, nothing more nor less, and he
was supposed to know all that was necessary of the “human form
divine.” He did, too, for people did not have the new-fangled
diseases of to-day. A woman’s hospital! Oh, heavens! Only last week
I saw a friend, old enough to know better, but we never are so old we
don’t want to rid our faces of pimples and warts and wrinkles. This
friend was a sight. I was really alarmed for her. She had been to a
specialist. Her face was fiery red, all the skin removed by acid.
Yesterday I saw her again, cured of sunburn and all the ills skin is
heir to. Her complexion was that of the lily girl who wore the scoop
sunbonnet. I do not advise you to try the experiment. It is shockingly
painful, and does not always prove a success.
When I was a little girl, more than seventy years ago, mother
made me, for summer romps in the country, gloves of nankeen, that
well covered the wrist, had a hole for the thumb and a deep flap to
fall over the hand. It was lucky they were easily made, and nankeen
was not expensive, for I hated them and had a way of losing them in
the currant bushes. Maybe you never saw nankeen? Gentlemen’s
waistcoats were often made of it, and little boys’ trousers. If I lost my
scoop sunbonnet one day—and it was surprising how easily I lost it!
—it was sewed on the next. There were no such things as hatpins—
and we had pigtails anyway, so they would have been of no use.
Such tortures were inflicted when we were running wild over the blue
grass farm, but no doubt the little Creole girls on the lakeshore were
similarly protected. The hair specialist was not in evidence either.

St. Louis Cemetery, New Orleans.

Ladies had their hair done up with bandoline and pomatums made
of beef’s marrow and castor oil and scented with patchouli; hair was
done into marvelous plaits and puffs. A very much admired style
which Henriette Blondeau, the fashionable hair dresser, achieved,
was a wide plait surrounding a nest of stiff puffs. It was called the
“basket of fruit.” The front locks were tiny, fluffy curls each side the
face and long ringlets to float over the shoulders. We all remember
Henriette Blondeau. She dressed my sister’s hair in the early forties,
and she dressed mine ten years later, and I met her in the hall of the
St. Charles Hotel, plying her trade, twenty years later still, the same
Henriette, with the same ample apron, the tools of her trade sticking
out from her pockets. Now, almost forty years later still, she walks
the streets of New Orleans no more. I hope she rests somewhere in
the old French cemetery, for she knew and gossiped with so many
who are taking their long sleep in that peaceful spot.
Mother made—no doubt your grandmother did, too—the pomade
that was used on our hair. It was used, too, very freely; our locks
plastered down good and smooth and flat. You may wonder how
long hair so treated could last; just as long as hair ruffled the wrong
way and marcelled with warm irons lasts our girls to-day. Mother’s
pomade was made of beef’s marrow and castor oil. After the marrow
was rendered to a fluid state, oil was added, then perfume, the whole
beaten in a deep bowl until perfectly cold and white. Mother would
beat and beat, add a few drops more of essence of bergamot, smell
and smell and beat and smell, until she had to call a fresh nose to
see if it was all right. I remember being told to try my olfactories on
the soft, creamy stuff. A naughty brother gave my head a blow that
sent my little pug-nose to the bottom of the bowl! My face was
covered to the ears, and while mother scraped it with a spoon and
scolded Henry, she was entreating me not to cry and have tears
spoil her pomade. Maybe I might have forgotten how the stuff was
made and how it looked, but for that ridiculous prank of the dearest
brother ever was.
I have a sweet little miniature of that brother Henry, namesake of
my father’s dear friend, Henry Clay, with the queer collared coat and
flourishing necktie of the day, and his long, straight hair well
plastered with mother’s good pomade. The dear man went to Central
America, on a pleasure tour to the ruins of Uxmal in 1844. The
vessel on which he sailed for home from Campeache, in September
of that year, disappeared in the Gulf. We never had any tidings of
how, or when, or where. I remember the firm of J. W. Zacharie was
consignee of that ill-fated Doric, and how tenderly Mr. Zacharie came
to my stricken mother, and how much he did to obtain information,
and how for weeks after all hopes were abandoned my mother’s
heart refused to believe her boy was indeed lost. Every night for
months she placed with her own trembling hands a lamp in the
window of Henry’s room, to light him when he came. She never gave
up some remnant of hope. So far as I know, only one friend of that
dear brother, one contemporary, is living now, in New Orleans. She
is the last of her generation; I am the last of mine.
In those days there were few patent medicines, washes and
lotions. There was a Jayne’s hair tonic, and somebody’s chologogue,
that was a fever cure much in evidence on plantations, for quinine
and blue mass pills—others, too—were made by hand. I have made
many a pill. We had an old negro woman who was daft on the
subject of medicine. There was not an earthly thing the matter with
Hannah—she was just a chronic grumbler, begging for “any kind of
pill.” I doctored her successfully, making for her bread pills, rolling
them in a little rhubarb dust to give them a nasty taste. They did her
a world of good. Mother made our lip salve (didn’t your
grandmother?) of white wax and sweet oil. We did not have cold
cream in those days.
When by accident, or some other way, our faces tanned, a wash
overnight of sour buttermilk was all that was required. It was not very
pleasant, and nobody wanted to occupy the room with you on sour
buttermilk night. Reason obvious. Kentucky belles, who were noted
for their rosy cheeks, often increased the bloom by a brisk rubbing of
the leaves of the wild mullein. Except rice powder (and that is not a
cosmetic) no cosmetics were in use.
We can recall at a later date than my girlhood a lady from
somewhere up the coast married a finicky cotton broker in New
Orleans. They made a wedding trip to Paris, and she returned with
her face enameled. I don’t think it could have been very skillfully
done, for she had to be so careful about using the muscles of the
face that she was absolutely devoid of expression. Once, in a
moment of forgetfulness or carelessness, she “cracked a smile,”
which cracked the enamel. She returned to Paris for repairs. I saw
her on the eve of sailing, and do not know if she ever returned.
 XVI
FANCY DRESS BALL AT THE MINT IN 1850

I have never heard of a society ball in a United States mint

building, before nor since, but the Kennedys, who gave this one,
were a power in the social world at that time—and ambitious beyond
their means. Rose and Josephine, the two oldest of quite a flock of
daughters, were débutantes that winter. Both were handsome and
accomplished. Rose was also a famous pianist, even in those days
when every woman strove to excel in music, and it was customary to
entertain even a casual caller with a sonata. Gottschalk declared
Rose Kennedy rendered his famous “Bamboula” better than he did
himself, and to hear her was to rise and dance.
Who was at that fancy ball? Everybody who was anybody in the
fifties. The Eustises—George and Mathilde, George as “a learned
judge” (he was son of Chief Justice Eustis), and Mathilde in pure
white and flowing veil was a bewitching nun. George, years after,
married the only child of the banker-millionaire, W. C. Corcoran, in
Washington. Mathilde married Alan Johnson, an Englishman; both
are long since dead. There was Mrs. John Slidell, of “Mason and
Slidell” fame, a “marquise,” in thread lace and velvet, her sisters, the
Misses Deslonde, “peasant girls of France.” Mathilde Deslonde
became the wife of Gen. Beauregard, and her sister, Caroline,
married Mr. R. W. Adams. All three sisters are with the departed. Col.
and Mrs. John Winthrop, “gentleman and lady of the nineteenth
century,” the jolly colonel announced. Who fails to recall, with a
smile, the Winthrops, who lived in Royal Street, near Conti; near
neighbors of the—long departed—Bonfords? The genial colonel
became a tottering old man, asking his devoted wife “who and where
are we?” before he peacefully faded away. Young De Wolf of Rhode
Island, nephew of Col. Winthrop’s, “an Arab sheik,” wore probably
the only genuine costume in the room—a flowing robe that was
catching in every girl’s coiffure, and every man’s sword and spurs, in
the dance.
All the gilded youth who wanted boisterous fun, and no jury duty,
were firemen, in those days of voluntary service. Philippe De la
Chaise wore his uniform. He later married Victoria Gasquet, and was
relegated to a “back number” shortly after.
I make no special mention of the chaperons, but, Creole like, they
were present in force. Cuthbert Slocomb was a mousquetaire, and
Augusta, in red and black, “Diablotan,” a vision of beauty and grace.
She married the Urquhart mentioned in “Musical History of
Louisiana,” as the father of Cora Urquhart Potter. Mr. Urquhart died
years ago, but his widow survives. She lives with her daughter at
Staines on the Thames, in a stone house that was a lodge of
Windsor Castle in the time of Henry VIII. Cuthbert Slocomb married
a Miss Day; his widow and daughter, Countess di Brazza, survive
him. Ida Slocomb was the noted philanthropist of New Orleans, the
widow of Dr. T. G. Richardson.
There was the stately Mrs. Martin Gordon chaperoning her
exceedingly pretty sister, Myrtle Bringier, who became the wife of
Gen. Dick Taylor, and whose descendants are among the few of
those mentioned above still living and reigning in New Orleans
society.
The mint building was made ample for the gay festivities by
utilizing committee rooms, offices and every apartment that could be
diverted for the crowd’s comfort—so, we wandered about corridors
and spacious rooms, but never beyond the touch of a gendarme—
officers, soldiers, policemen at every step. These precautions gave a
rather regal air to the whole affair.
 Augusta Slocomb Urquhart

Painted in Paris, in 1857

The belles retired to their boudoirs for a season, but the beaux had
to go to business, and what a sight some of them were for a whole
week after the fancy dress ball! They had hired costumes from
members of the French opera troupe, and their faces were “made
up” with rouge that could not be washed off; had to wear off in a
purplish stain. My brother represented Louis XIV on that occasion,
and I remember he scrubbed his cheeks until he made them almost
raw. Of no avail. In time the pinkish, purplish tint gradually
disappeared.
Shortly after that grandest and most unique entertainment Mr. Joe
Kennedy’s term expired and he retired into private life. Beautiful
Rose fell into a decline and died early. What fortunes befell that
family I know not. They seem to have faded away. The Kennedys
were a large family in those days, closely allied to the Pierce and
Cenas families, all of which were socially prominent. And now their
names are “writ in water.” I should like to know how many of this old
Creole society are living to-day! I was eighteen, one of the youngest
of the group, in the fifties.
 XVII
DR. CLAPP’S CHURCH

It is quite sixty years since Dr. Clapp’s church went up in smoke. It

was as well known to the denizens and visitors of New Orleans, in its
day, as Talmage’s Tabernacle in Brooklyn some decades later was
known far and wide. Dr. Clapp called it “The First Congregational
Church of New Orleans.” Others designated it as “Clapp’s Church.” It
was, in reality, neither one nor the other, for it was not an organized
congregation, and its building was the property of an eccentric Jew.
In a burst of admiration and generosity Judah Touro gave the church
rent free to Dr. Clapp. The structure had quite the appearance of a
“Friends’ Meeting House.” It was of unpainted brick, entirely devoid
of any ornamentation. The little steeple was only high enough and
big enough to hold the inevitable bell. One entered a narrow
vestibule, with two doors leading into the body of the church, and two
flights of stairs to respective galleries. It was further furnished with
two conspicuous tin signs—“Stranger’s Gallery on the Right,”
“Gallery for Colored Persons on the Left.” (Dr. Clapp came from
Boston.)
On entering the sanctuary one faced the organ loft, the pulpit
being at the street end between the two doors. It was a little rounded
affair, with, to all appearances, “standing room for one only.” Back of
it, to convey possibly an idea of space, and also to relieve the
intense white of the wall, was a wonderful drapery, very high and
very narrow, of red serge, pleated, looped and convoluted in an
amazing way.
Dr. Clapp, a large, handsome, middle-aged man, in a clerical black
silk robe, entered the pulpit from between the folds of that draped
monstrosity. He was dignified and reverential, preached without
notes, sometimes, but not always, using a Bible text. The music of
that church was rated as very fine, the organ was the best in the city.
(I wonder if old Judah Touro furnished that, too?) And Thomas
Cripps, the organist, managed it, con amore. There must have been
a choir to furnish the chorus, but I only call to mind Mrs. Renshaw
and her sister, Miss White, who sang solos and duets. Their finely
trained voices produced melody itself. Mr. James I. Day, tall, and
thin, and gaunt, with a hatchet face, who looked as if a squeak was
his vocal limit, had a most powerful bass voice that filled the building
and floated out onto the street. The last time I saw him he was in an
open carriage with a red velvet cushion on his lap, on which reposed
the key (as big as the famous Bastile key) of the city of New Orleans.
He was receiving Rex in an initial Mardi Gras parade. That was
years ago.
To return to church, I don’t recall any prayer books or hymnals, nor
hearing any congregational singing. The choir, of course, was
volunteer. We had yet to know a church singer could be salaried.
There was no church organization, as we know it to-day, or even at
that day. There were no officers, no deacons, no elders, far as I can
think, for my father was a devoted communicant and constant
attendant and naturally would have fitted into some church office, if
there had been any.
When Dr. Clapp announced the taking of a collection he cast his
eye over the congregation and signaled from it those persons who
were to “pass the plates.”
“Mr. Smith will take the center aisle, Mr. Jones the right aisle, Mr.
Robinson the left aisle, Mr. Dick right gallery, Mr. Harry left gallery,”
whereupon Messrs. Smith, Jones and Robinson and Messrs. Dick
and Harry would come forward, take their plates from a table under
the high pulpit and proceed to their allotted tasks. Remembering this
confirms me in the belief there were no officers of the church whose
duty it would have been to discharge such services.
There was only one service a week, a morning service and
sermon on Sundays, no night meetings, as there was really no
means of lighting the building. No Bible class, no Sunday school, no
prayer meeting, no missionary band, no church committee, no
Donors’ Society, no sewing circle, no donation party, no fairs, no
organ recital, absolutely “no nothing,” but Dr. Clapp and his weekly
sermon. The church was always filled to its utmost capacity. I recall a
host of pew holders whose names have passed into oblivion with
their bodies.
The old church stood on St. Charles Street, adjacent to the St.
Charles Hotel, so when one building went up in flames the other did,
too. The Veranda Hotel, next in importance to its neighbor, was
across the way, and from these sources always came strangers,
more than enough to fill the gallery, when they were wafted up the
stairs by the conspicuous tin sign.
Almost simultaneously with the destruction of the building,
disappeared both Dr. Clapp and Mr. Touro from public notice. By the
way, Mr. Judah Touro never had been inside the church, nor had he
ever heard Dr. Clapp preach. Of course, they are both as dead now
as the unique old church, so it matters not how, when or where they
departed. The congregation dissolved as completely. Probably not
one member, old enough at the time to know what Dr. Clapp
preached about or to be able to criticise his utterances, is living to-
day. Dr. Clapp was a loyal citizen, a charitable, kindly man, one of
the few who voluntarily remained in the city and ministered to the
stricken and buried the dead in the fearful epidemics that ravaged
the land every two or three years. His counsel reached the flotsam of
a great city, and his teachings bore fruit. He is gone now where
church organizations are not considered, but the good works he
wrought by his simple methods are placed to his credit.
 XVIII
OLD DAGUERREOTYPES

I think I can safely say I possess the first daguerreotype ever

taken in New Orleans. An artist came there about 1840 and opened
a studio (artist and studio sound rather grand when one views the
work to-day). That studio was at the corner of Canal Street and
Exchange Alley. The artist needed some pictures of well-known men
for his showcase, so he applied to my father, who was of the “helping
hand” variety. And dear Pa was rewarded with the gift of a picture of
himself all done up in a velvet-lined case, which he brought home to
the amazement and wonder of every member of the family, white
and black. I look at it now with a grim smile. Dear Pa’s cravat ends
were pulled out and his coattail laid nicely over one leg, and his hand
spread so that one could see he had five big fingers. His head had
been steadied straight up in a most unnatural position, with a kind of
callipers or steel braces, and he must have been told to “look up and
smile” for a full minute.
We prize that daguerreotype for its antiquity, but I hope seventy
years hence when another and another generation opens my “war
album” they will not laugh at the quaint cartes de visite it contains,
though I confess some of them begin to look rather queer already.
They were all gifts of near and dear friends, most of them with
autograph attachments, some of which were so flourishing that I had
to subscribe the names and dates on the backs.
There are Mr. and Mrs. Jeff Davis, dated 1860, before he was
President, you perceive. Though I have letters from both, I never
saw either after that date. There’s Gen. J. Bankhead Magruder, in
full uniform, far and away the most picturesque of my collection. The
first time we ever met Gen. Magruder was very soon after the
capture of the Harriet Lane in Galveston waters. The Texans were
wild and jubilant at the dashing feat, and when we reached Houston,
all travel-stained and worn out, the city was in a ferment of
excitement.
The General dearly loved to tell a good story, and the impediment
in his speech, a drawling lisp, made him vastly amusing. In his office
one day one of his aides was tinkling a banjo. A travel-stained
individual called:
“Is the General in?”
“No,” tinkle, tinkle.
“When will he be in?”
“Don’t know,” tinkle, tinkle.
“Will you tell him I called?”
“What name?” tinkle, tinkle.
“Smith.”
“I think I have heard that name before,” tinkle, tinkle. “What
Smith?”
“Gen. E. Kirby Smith, young man!”
No tinkle followed that reply. The young aide almost swooned
away. Gen. Magruder surrounded himself with Virginia gentlemen
aides, who gave him infinite trouble, he said.
In the early fifties “we met by chance, the usual way,” Major F.
Ducayet. A party driving down the old Bayou road one Sunday heard
that at Ducayet’s there would be found a rare collection of wonderful
fowls and poultry, and the owner was very gracious about showing
his assortment to visitors. After a bit of hesitation we ventured to
introduce ourselves. Mr. Ducayet received us most hospitably,
showed us through his lovely grounds and gave us the history of his
rarest feathered pets, presented the two ladies with choice bouquets
and insisted upon our partaking of refreshments. During the
conversation that ensued Mr. Ducayet said he would not be able to
increase his fancy flock, all of which had been brought him from
foreign parts by captains and sailors, as a change in the
administration would remove him from the position in the Custom
House he had held for years. One of the party at once asked him to
call on him at the St. Charles Hotel the following day, that he, being a
Democrat and a politician of influence, might exert himself in his
behalf. Mr. Ducayet retained his position. From that chance
acquaintance sprung a strong friendship. We saw much of Major
Ducayet in war times, hence the little carte de visite which
ornaments my war album.
By the side of Major Ducayet’s is the face of ex-Governor Moore of
Louisiana. He was an inmate of our modest little home in Texas
during the expiring days of the Confederacy.
I have also similar small photos of Major Tom Lee, General
Preston, General Breckinridge, Commodore John N. Maffit, General
and Mrs. Robert Toombs, General Early, Dr. Howard Smith and a
host of lesser lights, all of which were taken in Havana after the war.
Dr. Howard Smith of New Orleans was surgeon on somebody’s
(perhaps Gen. Kirby Smith’s) staff, and was our frequent guest in
Texas, a very valuable guest, too, for his skill carried some members
of my family out of the “valley of the shadow” into the sunshine. One
trip we made together from the Rio Grande into the interior of Texas,
quite a caravan of us in the party.
The first day out from Laredo there was a terrible sandstorm, cold
almost to freezing point, and never was a more disgusted party of
travelers. In a fit of despair Dr. Smith exclaimed: “I would give a
thousand dollars for a drink of brandy.” Now brandy was a luxury a
thousand dollars could not always supply, but I promptly replied: “I
will give you a whole bottle of brandy, the cork of which has not been
drawn, if you will divide it with the rest of the crowd.” Of course, the
proposition was accepted. From my carpetbag I produced a tiny toy
bottle, holding perhaps a half wine glass of the coveted liquor. It was
not easy to divide the contents liberally, but the genial doctor
appreciated the joke and did his utmost to carry out its provisions.
Years after, walking uptown in New Orleans, my escort said: “Look
at the man on that gallery. See if you know him.” I met the man’s
eyes full in my face, and passed on. It was Dr. Howard Smith, neither
of us recognizing the other. He was in ill-health, old and haggard,
and I guess I showed some of time’s footprints, too.
 XIX
STEAMBOAT AND STAGE SEVENTY YEARS AGO

In the twilight of my days, seated in my favorite chair, I rock away

many a trip from my New Orleans home to the blue grass region of
my ancestors. Dream trips they are, but dreams of real trips in the
old days when steamboats and stages were the approved, in fact,
the only, transportation for travelers.
About the Fourth of July every year our family migrated to the old
Kentucky homestead. The Fourth was not chosen with any patriotic
motive, but law courts were closed and legal business suspended,
and my father’s vacation at hand at that date. Though the
steamboats were called palatial, viewed from my rocking-chair trip
to-day I wonder how people managed to stand the inconveniences
and discomforts they provided.
There was the famed Grey Eagle, “a No. 1 floating palace” it was
called. There was the Belle of the West and the Fashion and the
Henry Clay. One time and another we churned up the muddy
Mississippi water in every one of them. Naturally the boats catered in
every way to the predilections of the plantation owners, who were
their main source of profit. The picture of Arlington which illustrates
this book was originally made to decorate a stateroom door on a fine
new river boat built in the ’50’s and adorned in that way with views of
homes along the river.
 Steamboat on the Mississippi.

(From “Forty Etchings, from Sketches Made with the Camera Lucida in North
America in 1827 and 1828,” by Captain Basil Hall, R. N.)

Grey Eagle was the finest and best, and therefore most popular
boat. I recall with amusement an eight or ten days’ trip on that
palace. The cabins were divided by curtains, drawn at night for
privacy. The ladies’ cabin, at the stern, was equipped with ten or
twelve small staterooms. The gentlemen’s cabin stretched on down
to the officers’ quarters, bar, barber shop, pantries, etc., ending in
what was called Social Hall, where the men sat about, smoking and
chewing (the latter as common a habit as cigarette smoking is now)
and talking—in other words, making themselves sociable.
On that same Grey Eagle I was for the first time promoted to the
upper berth, in a stateroom shared by an older sister. The berth was
so narrow that in attempting to turn over I fell out and landed in the
wash basin, on the opposite side of the room! My sister had to sit on
the lower berth to braid my pigtails, then sent me forth so she could
have room to braid her own. Trunks and other baggage more
unwieldy than carpetbags were piled up in the vicinity of Social Hall.
A carpetbag, small enough to be easily handled, was all there was
room for in the stateroom. There were no valises, suitcases or
steamer trunks in those days of little travel, and unless you are
three-quarters of a century old you can’t imagine a more unwieldy
article than a carpetbag of seventy years ago. Only toilet articles and
things that could not muss and tumble could be safely stored in one.
In the stateroom, where we had to sleep and dress, and, if we
could snatch a chance, take an afternoon nap, there was a corner
shelf for a basin and pitcher and one chair; two doors, one leading
out and the other leading in, transoms over each for light and
ventilation—and there you are for over a week. The cabin was
lighted with swinging whale-oil lamps, and one could light his
stateroom if one had thought to provide a candle.
Every family traveled with a man servant, whose business it was
to be constantly at beck and call. Of course, there was always a
colored chambermaid, and, equally of course, she frisked around
and seemed to have very little responsibility—no bells, no means of
summoning her from her little nodding naps if she happened to be
beyond the sound of one’s voice. The man servant’s duties,
therefore, were almost incessant. If an article was needed from the
trunks he was sent to the baggage pile for it, and often he brought
trunk trays to the staterooms. When the boat stopped “to wood”
every man servant rushed to the woodman’s cabin to get eggs,
chickens, milk, what not.
And those men had the privilege of the kitchen to prepare private
dishes for their white folks. I wonder how long a boat or hotel would
stand that kind of management to-day; but in the days where my
rocking-chair is transporting me, steamboat fare was not up to the
standard of any self-respecting pater familias. There was no ice
chest, no cold storage; in a word, no way of preserving fresh foods
for any length of time, so passengers resorted to such means as
presented themselves for their own bodily comfort. Those who had
not the necessary appendage—a man servant—foraged for
themselves, but the experienced and trusted servant, to use a
vulgarism, “was never left.”
The table for meals extended the length of the gentlemen’s cabin,
stretched out and out to its utmost length, if need be, so that every
passenger had a seat. There was no second table, no second-class
passengers—anybody was the equal of anybody else. If you could
not possibly be that, you could find accommodation on the lower
deck and eat from a tin plate.
It was quite customary, as I have mentioned, for passengers to
have private dishes, prepared by their own servants. I recall with a
smile, on one occasion, a very respectable-looking stranger boarded
our boat at Helena or some such place. At dinner he reached for a
bottle of wine. Cuthbert Bullitt touched the bottle with a fork, saying,
“Private wine.” The man, with a bow, withdrew his hand. Presently he
reached for a dish of eggs. My father said, “Excuse me, private.”
There was something else he reached for, I forget what, and another
fellow-passenger touched the dish and said “Private.” Presently
dessert was served, and a fine, large pie happened to be placed in
front of the Helena man. He promptly stuck his fork into it. “By
gracious! this is a private pie.” There was a roar of laughter.
After dinner the others, finding him delightfully congenial and
entertaining, fraternized with him to the extent of a few games at
cards. He was wonderfully lucky. He left the boat at an obscure river
town during the night, and the next day our captain said he was a
notorious gambler. From his capers at table the captain saw he was
planning a way of winning attention to himself, therefore under cover
of darkness he had been put ashore. My father, who did not play,
was vastly amused when he found the smart gambler had carried off
all the spare cash of those who had enjoyed the innocent sport.
Flatboats floating all manner of freight down stream were a
common sight on the river. Arrived at their destination, the boats,
which were only huge rafts with no propelling power, were broken up
and sold for lumber, and the boatmen traveled back up stream in
packets to repeat the process. Cousin Eliza Patrick used to relate
the trip her family made in about 1820 on a flatboat from Kentucky to
Louisiana. The widowed mother wished to rejoin a son practicing
medicine in the latter state, so she sold her land, and loaded her
family and every movable object she possessed—slaves, cattle,
farm implements, household effects—upon a huge “flat” and they
floated by day and tied up to the bank by night, carrying on, during