Bio-inspired Algorithms for Data Streaming and

Visualization, Big Data Management, and Fog

Computing Simon James Fong download

https://textbookfull.com/product/bio-inspired-algorithms-for-
data-streaming-and-visualization-big-data-management-and-fog-
computing-simon-james-fong/

Download more ebook from https://textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!

Biota Grow 2C gather 2C cook Loucas

https://textbookfull.com/product/biota-grow-2c-gather-2c-cook-
loucas/

Probabilistic data structures and algorithms for big

data applications Gakhov

https://textbookfull.com/product/probabilistic-data-structures-
and-algorithms-for-big-data-applications-gakhov/

Artificial Intelligence for Coronavirus Outbreak Simon

James Fong

https://textbookfull.com/product/artificial-intelligence-for-
coronavirus-outbreak-simon-james-fong/

Learning IPython for Interactive Computing and Data

Visualization - Second Edition Cyrille Rossant

https://textbookfull.com/product/learning-ipython-for-
interactive-computing-and-data-visualization-second-edition-
cyrille-rossant/
Innovations in Bio Inspired Computing and Applications
Ajith Abraham

https://textbookfull.com/product/innovations-in-bio-inspired-
computing-and-applications-ajith-abraham/

Disk-based algorithms for big data 1st Edition Healey

https://textbookfull.com/product/disk-based-algorithms-for-big-
data-1st-edition-healey/

Scala and Spark for Big Data Analytics Explore the

concepts of functional programming data streaming and
machine learning 1st Edition Md. Rezaul Karim

https://textbookfull.com/product/scala-and-spark-for-big-data-
analytics-explore-the-concepts-of-functional-programming-data-
streaming-and-machine-learning-1st-edition-md-rezaul-karim/

Big Data Analytics Systems Algorithms Applications

C.S.R. Prabhu

https://textbookfull.com/product/big-data-analytics-systems-
algorithms-applications-c-s-r-prabhu/

Bio-Inspired Algorithms in PID Controller Optimization

First Edition Ashour

https://textbookfull.com/product/bio-inspired-algorithms-in-pid-
controller-optimization-first-edition-ashour/
Springer Tracts in Nature-Inspired Computing

Simon James Fong

Richard C. Millham Editors

Bio-inspired
Algorithms for Data
Streaming and
Visualization, Big
Data Management,
and Fog Computing
Springer Tracts in Nature-Inspired Computing

Series Editors
Xin-She Yang, School of Science and Technology, Middlesex University, London,
UK
Nilanjan Dey, Department of Information Technology, Techno India College of
Technology, Kolkata, India
Simon Fong, Faculty of Science and Technology, University of Macau, Macau,
Macao
The book series is aimed at providing an exchange platform for researchers to
summarize the latest research and developments related to nature-inspired
computing in the most general sense. It includes analysis of nature-inspired
algorithms and techniques, inspiration from natural and biological systems,
computational mechanisms and models that imitate them in various fields, and
the applications to solve real-world problems in different disciplines. The book
series addresses the most recent innovations and developments in nature-inspired
computation, algorithms, models and methods, implementation, tools, architectures,
frameworks, structures, applications associated with bio-inspired methodologies
and other relevant areas.
The book series covers the topics and fields of Nature-Inspired Computing,
Bio-inspired Methods, Swarm Intelligence, Computational Intelligence,
Evolutionary Computation, Nature-Inspired Algorithms, Neural Computing, Data
Mining, Artificial Intelligence, Machine Learning, Theoretical Foundations and
Analysis, and Multi-Agent Systems. In addition, case studies, implementation of
methods and algorithms as well as applications in a diverse range of areas such as
Bioinformatics, Big Data, Computer Science, Signal and Image Processing,
Computer Vision, Biomedical and Health Science, Business Planning, Vehicle
Routing and others are also an important part of this book series.
The series publishes monographs, edited volumes and selected proceedings.

More information about this series at http://www.springer.com/series/16134

Simon James Fong Richard C. Millham
•

Editors

Bio-inspired Algorithms
for Data Streaming
and Visualization, Big Data
Management, and Fog
Computing

123
Editors
Simon James Fong Richard C. Millham
University of Macau Durban University of Technology
Taipa, China Durban, South Africa

ISSN 2524-552X ISSN 2524-5538 (electronic)

Springer Tracts in Nature-Inspired Computing
ISBN 978-981-15-6694-3 ISBN 978-981-15-6695-0 (eBook)
https://doi.org/10.1007/978-981-15-6695-0

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

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

The purpose of this book is to provide some insights into recently developed
bio-inspired algorithms within recent emerging trends of fog computing, sentiment
analysis, and data streaming as well as to provide a more comprehensive approach
to the big data management from pre-processing to analytics to visualisation phases.
Although the application domains of these new algorithms may be mentioned, these
algorithms are not confined to any particular application domain. Instead, these
algorithms provide an update into emerging research areas such as data streaming,
fog computing, and phases of big data management.
This book begins with the description of bio-inspired algorithms with a
description on how they are developed, along with an applied focus on how they
can be applied to missing value extrapolation (an area of big data pre-processing).
The book proceeds to chapters including identifying features through deep learning,
overview of data mining, recognising association rules, data streaming, data visu-
alisation, business intelligence and current big data tools.
One of the reasons for writing this book is that the bio-inspired approach does
not receive much attention although it continues to show considerable promise and
diversity in terms of approach of many issues in big data and streaming. This book
outlines the use of these algorithms to all phases of data management, not just a
specific phase such as data mining or business intelligence. Most chapters
demonstrate the effectiveness of a selected bio-inspired algorithm by experimental
evaluation of it against comparative algorithms. One chapter provides an overview
and evaluation of traditional algorithms, both sequential and parallel, for use in data
mining. This chapter is complemented by another chapter that uses a bio-inspired
algorithm for data mining in order to enable the reader to choose the most
appropriate choice of algorithms for data mining within a particular context. In all
chapters, references for further reading are provided, and in selected chapters, we
will also include ideas for future research.

Taipa, China Simon James Fong

Durban, South Africa Richard C. Millham

v
Contents

1 The Big Data Approach Using Bio-Inspired Algorithms: Data

Imputation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Richard Millham, Israel Edem Agbehadji, and Hongji Yang
2 Parameter Tuning onto Recurrent Neural Network and Long
Short-Term Memory (RNN-LSTM) Network for Feature
Selection in Classification of High-Dimensional
Bioinformatics Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Richard Millham, Israel Edem Agbehadji, and Hongji Yang
3 Data Stream Mining in Fog Computing Environment
with Feature Selection Using Ensemble of Swarm
Search Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Simon Fong, Tengyue Li, and Sabah Mohammed
4 Pattern Mining Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Richard Millham, Israel Edem Agbehadji, and Hongji Yang
5 Extracting Association Rules: Meta-Heuristic and Closeness
Preference Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Richard Millham, Israel Edem Agbehadji, and Hongji Yang
6 Lightweight Classifier-Based Outlier Detection Algorithms
from Multivariate Data Stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Simon Fong, Tengyue Li, Dong Han, and Sabah Mohammed
7 Comparison of Contemporary Meta-Heuristic Algorithms
for Solving Economic Load Dispatch Problem . . . . . . . . . . . . . . . . 127
Simon Fong, Tengyue Li, and Zhiyan Qu
8 The Paradigm of Fog Computing with Bio-inspired Search
Methods and the “5Vs” of Big Data . . . . . . . . . . . . . . . . . . . . . . . . 145
Richard Millham, Israel Edem Agbehadji,
and Samuel Ofori Frimpong

vii
viii Contents

9 Approach to Sentiment Analysis and Business Communication

on Social Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Israel Edem Agbehadji and Abosede Ijabadeniyi
10 Data Visualization Techniques and Algorithms . . . . . . . . . . . . . . . . 195
Israel Edem Agbehadji and Hongji Yang
11 Business Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Richard Millham, Israel Edem Agbehadji, and Emmanuel Freeman
12 Big Data Tools for Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Richard Millham
About the Editors

Simon James Fong graduated from La Trobe University, Australia, with a

First-Class Honours B.E. Computer Systems degree and a Ph.D. Computer Science
degree in 1993 and 1998, respectively. Simon is now working as an Associate
Professor at the Computer and Information Science Department of the University of
Macau. He is a Co-Founder of the Data Analytics and Collaborative Computing
Research Group in the Faculty of Science and Technology. Prior to his academic
career, Simon took up various managerial and technical posts, such as Systems
Engineer, IT Consultant, and E-commerce Director in Australia and Asia. Dr. Fong
has published over 500 international conference and peer-reviewed journal papers,
mostly in the areas of data mining, data stream mining, big data analytics,
meta-heuristics optimization algorithms, and their applications. He serves on the
editorial boards of the Journal of Network and Computer Applications of Elsevier,
IEEE IT Professional Magazine, and various special issues of SCIE-indexed
journals. Currently, Simon is chairing a SIG, namely Blockchain for e-Health at
IEEE Communication Society.

Richard C. Millham a B.A. (Hons.) from the University of Saskatchewan in

Canada, M.Sc. from the University of Abertay in Dundee, Scotland, and a Ph.D.
from De Montfort University in Leicester, England. After working in industry in
diverse fields for 15 years, he joined academe and he has taught in Scotland, Ghana,
South Sudan, and the Bahamas before joining DUT. His research interests include
software and data evolution, cloud computing, big data, bio-inspired algorithms,
and aspects of IOT.

ix
Chapter 1
The Big Data Approach Using
Bio-Inspired Algorithms: Data
Imputation

Richard Millham, Israel Edem Agbehadji, and Hongji Yang

1 Introduction

In this chapter, the concept of big data is defined based on the five characteristics
namely velocity, volume, value, veracity, and variety. Once defined, the sequential
phases of big data are denoted, namely data cleansing, data mining, and visual-
ization. Each phase consists of several sub-phases or steps. These steps are briefly
described. In order to manipulate data, a number of methods may be employed.
In this chapter, we look at an approach for data imputation or the extrapolation of
missing values in data. The concept of genetic algorithms along with its off-shoot,
meta-heuristic algorithms, is presented. A specialized type of meta-heuristic algo-
rithm, bio-inspired algorithms, is introduced with several example algorithms. An
example, a bio-inspired algorithm, the kestrel, is introduced using the steps outlined
for the development of a bio-inspired algorithm (Zang et al. 2010). This kestrel algo-
rithm will be used as an approach for data imputation within the big data phases
framework.

R. Millham (B) · I. E. Agbehadji

ICT and Society Research Group, Durban University of Technology, Durban, South Africa
e-mail: [email protected]
I. E. Agbehadji
e-mail: [email protected]
H. Yang
Department of Informatics, University of Leicester, Leicester, England, UK
e-mail: [email protected]

© The Editor(s) (if applicable) and The Author(s), under exclusive license 1
to Springer Nature Singapore Pte Ltd. 2021
S. J. Fong and R. C. Millham (eds.), Bio-inspired Algorithms for Data
Streaming and Visualization, Big Data Management, and Fog Computing,
Springer Tracts in Nature-Inspired Computing,
https://doi.org/10.1007/978-981-15-6695-0_1
2 R. Millham et al.

2 Big Data Framework

The definition of big data varies from one author to another. A common definition
might be that it denotes huge volume and complicated data sets because it comes
from heterogeneous sources (Banupriya and Vijayadeepa 2015). Because of the enor-
mous variety in definitions, big data is often known by its characteristics of velocity,
volume, value, veracity, and variety which constitutes the framework of big data.
Velocity relates to how quickly incoming data needs to be evaluated with results
produced Longbottom and Bamforth (2013). Volume relates to the amount of data to
be processed. Veracity relates to the accuracy of results emerging from the big data
processes. Value is the degree of worth that the user will obtain from the big data
analysis.

3 Evolutionary and Bio-Inspired Methods

Genetic algorithms (GA) inherited the principles of “Darwin’s Evolutionary Theory”.

Genetic algorithms provide solutions to a search problem by using biological evolu-
tion principles. Nature breeds a large number of optimized solutions which have been
discovered and deployed to solve problems (Zang et al. 2010). Genetic algorithm
adopts some common genetic expressions such as
(1) Chromosome: where the solution to an optimization problem is encoded (Cha
and Tappert 2009).
(2) Selection: a phase where individual chromosomes are evaluated and the best
are chosen to raise the next generation.
(3) “Crossover” and “mutation” are genetic methods for pairing parents to change
their genetic makeup through the process of breeding.
The first phase of a genetic algorithm produces initial population which represents
randomly generated individuals. This individual forms a range of potential solutions
in which the population size is determined by the nature of the problem. The initial
population represents the search space and the algorithm begins with an initial esti-
mate. Then the operators of crossover and mutation are applied to the population in
order to try to improve the estimate through evolution (Agbehadji 2011). The next
phase assesses the individual of a given population to determine their fitness values
through a fitness function. The higher the fitness value of individuals, the greater the
probability that the individual will be selected for the next generation. This process of
mutation, selection via the fitness function, and generation/iteration continues until
a termination criteria, or final optimal value or solution, is met.
Because of its adaptive and iterative nature, genetic algorithm may be used to
discover multiple types of optimal solutions from a set of initial random potential
solutions. Given the continuous updating of the population through the application
of genetic operators and the culling-off weak generation via a fitness function, the
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 3

gradual improvement of the population to a termination condition, or optimal solu-

tion, is made. One such solution that may be determined via genetic algorithm is
discovering an optimal path (Dorigo et al. 2006). A practical application of a genetic
algorithm is extrapolating missing values in a dataset (Abdella and Marwala 2006).
Meta-heuristic search or bio-inspired search or nature-inspired search methods
are mostly used interchangeably to refer to search algorithms developed from the
behaviour of living creatures in their natural habitat. Conceptually, living creatures
are adapted to make random decisions that can steer them either towards hunt or away
from its enemy. Meta-heuristic search methods can be combined to develop a more
robust search algorithm for any complex problems. The advantage of meta-heuristic
search method is the ability to ignore a search that is not promising. Generally, meta-
heuristic search algorithm begin with random set of individuals where each represents
a possible solution. In each generation, instead of mutation, there is a random Levy
walk (which corresponds to the random movements of animals/random searches for
an optimal solution). At the end of each generation, the fitness of each individual of
that generation is evaluated via a specified fitness function. Only those individuals
that meet a prescribed threshold of fitness, as determined by a fitness function, are
allowed to continue as parents for the next generation. The succession of generation
continues until some pre-defined stopping criteria is reached; ideally, this stopping
criteria is when a near-optimal solution has been found (Fong 2016).

3.1 Development Process for Bio-Inspired Algorithms

These are the stages in developing a bio-inspired algorithms:

(a) Firstly: identify the unique behaviour of a creature in nature,
(b) Secondly: formulate basic expressions on their behaviour.
(c) Thirdly, transform the basic expression into mathematical equation, identify
some underlying assumptions, and setup initial parameters,
(d) Fourthly, write a pseudo-code to represent the basic expression,
(e) Fifth: test the code on actual data and refine the initial parameter for better
performance of the algorithm.
Usually, animal behaviour constitutes actions relative to its environment and
context; thus, a particular animal behaviour should be modelled in conjunction with
other animal behaviours, other in terms of a team of individuals or another species,
in order to achieve better results. Therefore, the nature-inspired algorithms can be
combined with other algorithms for an efficient result and more robust algorithm
(Zang et al. 2010).

3.1.1 Examples of Bio-Inspired Algorithms

Bio-inspired algorithms can focus on the collective behaviour of multiple simple

individuals (as in particle swarm) (Selvaraj et al. 2014), the co-operative behaviour
4 R. Millham et al.

of more complex individuals (as in wolf search algorithm) (Tang et al. 2012), or the
single behaviour of an individual (Agbehadji et al. 2016b). Within these categories,
such as particle swarm, there are many types (such as artificial bee colony), and
within these types, there are many applications of the same algorithm for such things
as image processing, route optimization, etc. (Selvaraj et al. 2014).
A major category of bio-inspired algorithms are particle swarm algorithms.
Particle swarm algorithms is a bio-inspired technique that mimics the swarm
behaviour of animals such as fish schools or bird flocks (Kennedy and Eberhart
1995). The behaviour of the swarm is determined by how particles adapt and make
decisions in changing their position within a space relative to the positions of neigh-
bouring particles. The advantage of swarm behaviour is that as particles make a
decision, it leads to local interaction among particles which in turn, lead it to an
emergent behaviour (Krause et al. 2013). Particle swarm algorithm that focuses on
finding the near-optimal solution includes the firefly algorithm, bats (Yang and Deb
2009) and cuckoo birds (Yang and Deb 2009).

3.1.2 Firefly Algorithm

The basis of the firefly algorithm’s behaviour is the short and rhythmic flashes it
produces. This flashing light of fireflies is used as an instrument to attract possible
prey, attract mating partners, and to act as a warning signal. The firefly signalling
system consists of rhythmic flash, frequency of flashing light and time period of
flashing. This signalling system is controlled by simplified basic rules underlining
the behaviour of firefly that can be summarized as, one firefly can be connected with
another; hence, this connection which refers to attractiveness is proportional to the
level of brightness between each firefly and brightness is affected by landscape (Yang
2010a, b, c). The attraction formulation is based on the following assumptions:
(a) Each firefly attracts another fireflies that has a weak flash light
(b) This attraction depends on the level of brightness of the flash which is reversely
proportional to their proximity to each other
(c) The firefly with the brightest flash is not attracted to any other firefly and their
flight is random (Yang 2010a, b, c).
The signal of this flashing light instrument is governed by a simplified basic rule
which forms the basis of firefly behaviour. In comparison with a genetic algorithm, it
uses what is referred to as operators that are mutation, crossover, and selection. The
firefly uses attractiveness and brightness of its flashing light. The similarity between
the firefly algorithm and the genetic algorithm is that both algorithms generate an
initial population which is updated continuously at each iteration, via fitness function.
In terms of firefly behaviour, the brighter fireflies attract those fireflies nearest to them
and those fireflies whose brightness fall below a defined threshold are removed from
subsequent population. The brightest fireflies, whose brightness have exceeded a
specified threshold, constitute the next generation and this generation continues until
either a termination criteria (best possible solution) is met or the highest number of
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 5

iterations is achieved. The use of brightness in firefly algorithm is to help attract the
weaker firefly which mimics the extrapolation of missing values in a dataset where
the fireflies represent known values and those with the brightest light (indicating
closeness to the missing values as well as nearness to the set of data including the
missing value) are selected as suitable to replace the missing value entries.

3.1.3 Bat Search Algorithms

The bat search algorithm is another bio-inspired search technique that is grounded
on the behaviour of micro-bats within their natural environment (Yang 2010a, b, c).
Bat is known to have a very unique behaviour called echolocation. This characteristic
assists bats to orient themselves and find prey within their habitat. The search strategy
of a bat, whether to navigate or to capture prey, is governed by the pulse rate and
loudness of their cry. This pulse rate governs the enhancement of the best possible
solution, its loudness affects the acceptance of the best possible solution (Fister
et al. 2014). Similar to genetic search algorithm, the bat search algorithm begins
with random initialization, evaluation of the newly generated population, and after
multiple iterations, the best possible solution is outputted. In contrast to the wolf
search algorithm that uses attractiveness, the bat search algorithm uses its pulse rate
and loudness to steer its search for a near-optimal solution. The bat search algorithm,
with its behaviour, has been applied to several optimization problems to find the best
possible solution.

3.1.4 Wolf Search Algorithm

The wolf search algorithm (WSA) is a nature-inspired algorithm that focuses on

a wolf’s preying behaviour (Tang et al. 2012). This preying behaviour, as derived
from wolves’ behaviour, demonstrates that wolves are able to hunt independently by
recalling their own trait; have the ability to join with a fellow wolf only when the
other wolf is in a better position; and have the ability to randomly escape when a
hunter appears. This expressed wolf behaviour allows them to adapt to their habitat
when hunting for prey. Because wolves have the ability to join a fellow wolf in a
better position, it implies that wolves have some trust in each other and they avoid
preying upon each other. In addition, wolves prefer to only move into territory mark
by other wolves which indicates that the territory is safe for other wolves to live in.
Moreover, if this new location is better, the motivation is stronger especially if this
new location is within territory already occupied by a fellow wolf. This wolf search
algorithm can be defined as a search procedure which begins with setting the initial
population, evaluating the candidate population and updating the current population
via fitness test, and continuous until stopping criteria is met. Particle swarm algo-
rithms, like firefly, attract its prey by using the characteristics of attractiveness and
brightness while wolf uses the characteristic of attractiveness of prey within its visual
range. Wolves also have both individual search capability and independent flocking
6 R. Millham et al.

movement. In WSA, consequently, the swarming behaviour of wolves, unlike other

nature-inspired algorithms, is delegated to individual wolf instead of a single leader,
as is the case in the particle swarm and firefly algorithms. In practice, WSA works as
if there are “multiple leaders swarming from multiple directions” to the best possible
solution instead of a “single flock” that searches for the best possible solution in one
direction at a time (Tang et al. 2012). Similar to the firefly and bat, the WSA char-
acteristic and behaviour towards attraction can be used to extrapolate the estimated
value that is near to known values in any missing data imputation method.
Nature-inspired or bio-inspired search algorithms are characterized by random-
ization, efficient local search and finding of global best results (Yang 2010a, b, c).
With the newly developed kestrel search algorithm, the random circling of a kestrel
is examined to see how it may be used to achieve a best possible solution (estimates
closest to missing values). The advantage of the random encircling method of the
kestrel, unlike other bio-inspired algorithms, is that it maximizes the local search
space, and in so doing, it creates a wider range of possible solutions, based on a
hovering position, in order to assess and obtain the best possible solution.

3.1.5 Kestrel Behaviour

In keeping with Zang et al. (2010)’s prescribed method of developing a bio-inspired

search algorithm, in this case that of a kestrel bird, the behaviour is observed and
briefly summarized to depict its behaviour in a natural environment. This search
algorithm of a kestrel bird is based on its hunting characteristics that are either
hovering or perched hunt. Kestrels are highly territorial and hunt individually (Shrubb
1982; Varland 1991). One researcher, Varland (1991), recognized that during hunts,
kestrels tend to be imitative rather than co-operative. In other words, kestrels choose
“not to communicate with each other” instead they “imitate the behaviour of other
kestrels with better hunting techniques”. With mimicking better techniques comes the
improvement of their own technique. The hunt technique, however, can be dependent
on such factors such as the type of prey, current weather conditions, and energy
requirements (for gliding or diving) (Vlachos et al. 2003).
During their hunt, kestrels use their “eyesight to view small and agile prey” within
its coverage area, as defined by its visual circling radius. Prey availability is indicated
either through a “trail of urine and faeces” from ground-based prey or through the
minute air disturbance from airborne-based prey. Once the prey availability is identi-
fied, the “kestrel positions itself to hunt”. Kestrels can “hover in changing airstream,
maintain fixed forward” looking position with its eye on a potential prey, and use
“random bobbing” of its head to find the minimum distance between its “position
and the position of prey”. Because kestrels can view ultraviolet light, they are able
to discover trails of urine and faeces left by prey such as voles (Honkavaara et al.
2002).
While hovering, kestrels can perform a broader search (e.g. global exploration)
across different territories within its “visual circling radius”, are able to “maintain
a motionless position with their forward-looking eyes fixed on prey, detect tiny
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 7

air disturbances from flying prey (especially flying insects) as indicators of prey”,
and can move “with precision through a changing airstream”. Kestrels have the
ability to flap their winds and adjust their long tails in order to stay in a place
(denoted as a still position) in a “changing airstream”. While in perch mode (often
perching from high fixed structures such as poles or trees), kestrels change their
perch position every few minutes before performing a thorough search (which is
denoted as “local exploitation” based on its individual hunt behaviour) of its local
territory which requires “less energy than a hovering hunt”. While in perch mode, the
kestrel uses its ultraviolet detection capacity to discover potential prey such as voles
nearer to its perch area. This behaviour suggests that while in perch stance, kestrel
uses this position to conserve some energy and to focus their ultraviolet detection
capabilities for spotting slow moving prey on the ground. Regardless of perch or
hovering mode, skill development also plays a role. Individual kestrels with better
“perch and hovering skills” that are utilized in a larger search area possess a better
chance to swoop down faster on their prey or flee from its enemies than “individual
kestrels that develop hunting skills in local territories” (Varland 1991). Consequently,
it is important to combine hunting skills from both hovering and perch modes in order
to accomplish a successful hunt.
In order to better characterize the kestrel, certain traits are given as their defining
behaviour:
(1) Soaring: it provides a wider search space (global exploration) within their visual
coverage area

(a) Still (motionless) location with eyesight set on prey

(i) Encircles prey underneath it using its keen eyesight

(2) Perching: this enables thorough search or local exploitation within a visual
coverage radius

(a) Behaviour involves “frequent bobbing of head” to find the best position of
attack
(b) Using a trail, identify potential prey and then the kestrel glides to capture
prey

These behavioural characteristics are based on the following assumptions:

(a) The still position of the kestrel bird provides a near perfect circle. Consequently,
frequent changes in circle direction depend on the position of prey shifting the
centre of this circling direction
(b) The frequent bobbing of the kestrel’s head provides a “degree of magnified
or binocular vision” that assists in judging the distance from the kestrel to a
potential prey and calculating a striking move with the required speed
(c) “Attractiveness is proportional to light reflection”. Consequently, “the higher or
longer a distance from the trail to the kestrel, the less bright of a trail”. This
distance parameter applies to both the hovering height and the distance away
from the perch.
8 R. Millham et al.

(d) “New trails are more attractive than old trails”. Thus, the trail decay, as the trail
evaporates, depends on “the half-life of the trail”.

Mathematical Model of Kestrel’s Behaviour

Following the steps of Zang et al. (2010), a model that represents the kestrel behaviour
is expressed mathematically. The following sets of kestrel characteristics, with their
mathematical equivalents, are provided below:
• Encircling behaviour

This encircling behaviour occurs when the “kestrel randomly shifts (or changes)” its
“centre of circling direction” in response to detecting the current position of prey.
When the prey changes from its present position, the kestrel randomly shifts, or
changes, the “centre of circling direction” in order to recognize the present position
of prey. With the change of position of prey, the kestrel correspondingly alters its
encircling behaviour to encircle its prey. The movement of prey results in the kestrel
adopting the best possible position to strike. This encircling behaviour
D (Kumar 2015) is denoted in Eq. 1 as:
→ 
 = −
D C ∗−
→ 
x p (t) − x(t) (1)

 denoted in equation 2as:

Cis

C = 2 ∗ −
→
r1 (2)

where C is the “coefficient vector”, −

→
x p (t) is the position vector of the prey, and x(t)
represents the position of a kestrel, r1 and r2 are random values between 0 and 1
indicating random movements.
• Current position

The present best position of the kestrel is denoted in Eq. 3 as follows:

x(t + 1) = −
→
x p (t) − A ∗ D
 (3)

Consequently, the coefficient A is denoted in Eq. 4 as follows:

A = 2 ∗ z ∗ −
→
r2 − z (4)

where A also represents coefficient vector, D  is the encircling value acquired, −

→
x p (t)
is the prey’s position vector, x(t + 1) signifies present best position of kestrels. z
decreases linearly from 2 to 0 and this value is also used to “control the randomness”
at each iterations. The z is denoted in Eq. 5 as follows:
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 9

itr
z = z hi − (z hi − z low ) (5)
Max_ itr

where itr is the current iteration, Max_itr represents maximum number of iterations
that stop the search, zhi denotes the higher bound of 2, zlow denotes the lower bound
of 0. Any other kestrels included in this search for prey will update their position
based on the best position of the leading kestrel. In addition, the change in position in
the airstream for kestrels is dependent on the “frequency of bobbing”, how it attracts
prey and “trail evaporation”. These dependent variables are denoted as follows:
(a) Frequency of bobbing

The bobbing frequency is used to determine sight distance measurement within the
search space. This is denoted in Eq. 6 as follows:
 
k
f t+1 = f min + f max − f min ∗ α (6)

where α ∈ [0, 1] indicates a random number to govern the “frequency of bobbing

within a visual range”. The maximum frequency f max is set at 1 while the minimum
frequency f min is set at 0.
(b) Attractiveness

Attractiveness β denotes the light reflection from trails, which is expressed in Eq. (7)
as follows:

β(r ) = βo e−γ r
2
(7)

where βo equals lo and constitutes the initial attractiveness, γ denotes variation of

light intensity between [0, 1]. r denotes the sight distance s(xi , xc ) measurement
which is calculated using “Minkowski distance” expression in Eq. (8) as:


n
|xi,k − xc,k |λ ) λ
1
s(xi , xc ) = ( (8)
k=1

Consequently, Eq. 9 expresses the visual range as follows:

V ≤ s(xi , xc ) (9)

where x i denotes the current sight measurement, x c indicates all possible adjacent
sight measurement near x i , n is the total number of adjacent sights and λ is the order
(values of 1 or 2) and V is the visual range.
(c) Trail evaporation

A trail may be defined as way to form and maintain a line (Dorigo and Gambardella
1997). In meta-heuristic algorithms, trails are used by ants to track the path from their
10 R. Millham et al.

home to a food source while avoiding getting mired to just one food source. Thus,
these trails enable multiple food sources to be used within a search space. (Agbehadji
2011) While ants search continuously, trails are developed with elements attached to
these trails. These elements assist ants in communicating with each other regarding
the position of food sources. Consequently, other ants constantly follow this path
while depositing elements for the trail to remain fresh. In the same manner that ants
use trails, “kestrels use trails to search for food sources”. These trails, unlike those
of ants, are created by prey which, thus, provide an indication to kestrels on the
obtainability of food sources. The assumption with the kestrel is that the elements
left by these prey (urine, faeces, etc.) are similar to those elements left on an ant
trail. In addition, when the food source indicated by the trail is exhausted, kestrels
no longer pursue this path as the trail elements begin to reduce with “time at an
exponential rate”. With the reduction of trails’ elements, the trail turns old. This
reduction indicates the unstable essence of trail elements which is expressed as if
there are N “unstable substances” with an “exponential decay rate” of γ, then the
equation to detail how N element reduces in time t is expressed as follows (Spencer
2002):

dN
= −γ N (10)
dt
Because these elements are unstable, there is “randomness in the decay process”.
Consequently, the rate of decay (γ ) with respect to time (t) can be re-defined as
follows:

γt = γo e−λt (11)

where γo is a “random initial value” of trail elements that is reduced at each iteration.
t is the number of iterations/generations/time steps, where t ∈ [0, Max_itr] with
Max_itr being the maximum number of iterations.
⎧
⎨ γt > 1, trail is new
if γt → (12)
⎩
0, otherwise

Once more, the decay constant λ is denoted by:

φmax − φmin
λ= (13)
t 21

where λ is “the decay constant”, φmax is the maximum number elements in trail,
φmin is the minimum number of elements in trail and t 21 is the “half-life period of
a trail which indicates that a trail” has become “old and unattractive” for pursuing
prey.
Lastly, the Kestrel will updates its location using the following equation:
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 11

2 
k
xt+1 = xtk + βo e−γ r x j − xi + f tk (14)

k
where xt+1 signifies the present optimal location of kestrels. xtk is the preceding
location.
• Fitness function

In order to evaluation how well an algorithm achieves in terms of some criteria (such
as the quality of estimation for missing value), a fitness function is applied. In the
case of missing value estimation, the measurement of this achievement is in terms
of “minimizing the deviation of data points from the estimated value”. A number of
performance measurement tools may be used such as mean absolute error (MAE),
root mean square (RMSE), and mean square error (MSE).
In this chapter, the fitness function for the kestrel search algorithm uses the mean
absolute error (MAE) as its performance measurement tool in order to determine
the quality of estimation of missing values. MAE was selected for use in the fitness
function because it allows the modelled behaviour of the kestrel to fine tune and
improve on its much more precise estimation of values concern for negative values.
The MAE is expressed in Eq. (15) as follows:

1
n
MAE = |oi − xi | (15)
n i=1

where xi indicates the estimated value at the ith position in the dataset, oi denotes
the observed data point at ith position “in the sampled dataset, and n is the number
of data points in the sampled dataset”.
• Velocity

The velocity of kestrel as it moves from its current optimal location in a “changing
airstream” is expressed as:

vt+1
k
= vtk + xtk (16)

Any variation in velocity is governed by the inertia weight ω (which is also denoted
as the convergent parameter). This “inertia weight has a linearly” diminishing value.
Thus, velocity is denoted in Eq. 17 as follows:

vt+1
k
= ωvtk + xtk (17)

where ω is the “convergence parameter”, vtk is the “initial velocity”, xtk is best loca-
tion of the kestrel and the vt+1
k
is the present best velocity of the kestrel. Kestrels
explore through the search space to discover optimal solution and in so doing, they
constantly update the velocity, random encircling, and location towards the best
estimated solution.
12 R. Millham et al.

Table 1 Kestrel algorithm

• Set parameters
• Initialize population of n Kestrels using equation (3) and evaluate fitness of
population using equation (18)
• Start iteration (loop until termination criteria is met)
Compute Half-life of trail using equation (11)
Compute frequency of bobbing using equation (6)
Evaluate position for each Kestrel as in equation using equation (14)
If f (xi ) < f( xj ) then
Move Kestrel i towards j
End if
• Update position f(xi ) for all i=1 to n as in equation (17)
• Find the current best value
• End loop

Kestrel-Based Search Algorithm

Following Zang (2010) steps to develop a new bio-inspired algorithm, after certain
aspects of behaviour of the selected animal is mathematically modelled, the pseudo-
code or algorithm that incorporates parts of this mathematical model is developed
both to simulate animal behaviour and to discover the best possible solution to a
given problem.
The algorithm for kestrel is given as follows (Table 1).

Implementation of Kestrel-Based Algorithm

After the algorithm for the newly developed bio-inspired algorithm has been deter-
mined, the next step, according to Zang et al. (2010) is to test the algorithm experi-
mentally. Although kestrel behaviour, due to its encircling behaviour and adaptability
to different hunting contexts [either high above as in hovering or near the ground as
in perching] (Agbehadji et al. 2016a), is capable of being used in a variety of steps
and phases of big data mining, the step of estimating missing values within the data
cleansing phase was chosen.
Following Zang’s et al. (2010) prescription to develop a bio-inspired algorithm,
the parameters of the bio-inspired algorithm are set. The initial parameters for the
KSA algorithm were set as βo = 1 with visual range = 1. As per Eq. 5, the parameters
for the lower and higher bound, zmin = 0.2 and zmax = 0.9, respectively, were set
accordingly. A maximum number of 500 iterations/generations were set in order to
allow the algorithm to have a better opportunity of further refining the best estimated
values in each iteration.
Further to Zang’s et al. (2010) rule, the algorithm is tested against appropriate
data. This algorithm was tested using a representative dataset matrix of 46 rows and
9 columns with multiple values missing in each row of the matrix. This matrix was
designed to allow for a thorough testing of estimation of missing values by the KSA
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 13

4 Maximum Likelihood (ML) Method of estimation

-10
ML

Maximum Likelihood values

5
-10

6
-10
2 3 4 5 6 7 8 9 10
Iterations

Fig. 1 Maximum likelihood

algorithm. This testing produced the following Fig. 1: A “sample set of data (46 by 9
matrix) with multiple missing values in the row matrix was used in order to provide
a thorough test of missing values in each row of a matrix”. The test revealed the
following figure represented as Fig. 2:
Figure 2 shows a single graph of the fitness function value of the KSA algorithm
during “500 iterations”. As can be seen in this graph, the “curve ascends and descends
steeply during the beginning iterations and then gradually converges at the best
possible solution at the end of 500 iterations/generations”. The steps within the
curve symbolize looking for a best solution within a particular search space, using
a random method, until one is found and then another space is explored. The curve
characteristics indicate that at the starting iterations, the KSA algorithm “quickly
maximizes the search space and then gradually minimizes” until it converges to the
best possible optimal value.

4 Conventional Data Imputation Methods

Conventional approaches to estimate missing data values include ignoring missing

attributes or fill in missing values with a global constant (Quinlan 1989), with the
real possibility of detracting from the quality of pattern(s) discovered based on these
values. Based on the historical trend model, missing data may be extrapolated, in
terms of their approximate value, using trends (Narang 2013). This procedure is
14 R. Millham et al.

Comparative results of fitness function

3

Fitness value using MAE 2.5

1.5

0.5

0
0 50 100 150 200 250 300 350 400 450 500
Iterations

Fig. 2 KSA fitness

common in the domain of real-time stock trading with missing data values. In real-
time trading, each stock value is marked in conjunction with a timestamp. In order to
extrapolate the correct timestamp from missing incorrect/missing timestamps, every
data entry point is checked against the internal system clock to estimate the likely
missing timestamp (Narang 2013). However, this timestamp extrapolation method
has disadvantages in its high computation cost and slower system response time for
huge volumes of data.
There are other ways to handle missing data. Conventional approaches include
ignoring missing attributes or fill in missing values with a global constant (Quinlan
1989), with the real possibility of detracting from the quality of pattern(s) discovered
based on these values. Another approach was by Grzymala-Busse et al. (2005), that is
the closest fit method, where the same attributes from similar cases are used to extrap-
olate the missing attributes. Other approaches of extrapolation include maximum
likelihood, genetic programming, Expectation-Maximization (EM), Expectation-
Maximization (EM), and “machine learning approach (such as autoencoder neural
network)” (Lakshminarayan et al. 1999).
• Closest fit Method

This method determines the closest value of the missing data attribute through the
closest fit algorithm based on the same attributes from similar cases. Using the
closest fit algorithm, the distance between cases (such as case x and y) are based on
the Manhattan distance formula that is given below:
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 15


n
distance(x, y) = distance(xi , yi )
i=1

where:
⎧
⎨0 if x = y
distance(x, y) = 1 if x and y are symbolic and x = y, or x =? or y =?
⎩ |x−y|
r
if x and y are numbers and x = y

where r denotes the differences between the maximum and minimum of the unknown
values of missing numeric values (Grzymala-Busse et al. 2005).
• Maximum likelihood:

Maximum likelihood is a statistical method to approximate a missing value based

on the probability of independent observations. The beginning point for this approx-
imation is the development of a likelihood function that determines the probability
of data as a function of data and its missing value. Allison (2012), the estimation
commences with the expression of likelihood function to present the probability of
data, as a function of data and its missing value. This function’s parameters must
maximize the likelihood of the observed value as in the following formulation:

 
L(θ |Yobserved ) = f Yobserved , Ymissing |θ dYmissing

where Yobserved denotes the observed data, Ymissing is the missing data, and º is
the parameter of interest to be predicted (Little and Rubin 1987). Subsequently,
likelihood function is expressed by:
n
L(θ ) = f (yi |θ )
i=1

where f(y|8) is the probability density function of the observations y whilst θ is the set
of parameters that has to be predicted provided n number of independent observation
(Allison 2012). The value of θ must be first determined before a maximum likelihood
prediction can be calculated which serves to maximize the likelihood function.
Suppose that there are n independent observation on k variables (y1 , y2 , …, yk )
“with no missing data, the likelihood function “is denoted as:
n
L= f (yi1 , yi2 , . . . , yik ; θ )
i=1

However, suppose that data is missing for individual observation i for y1 and y2.
Then, the likelihood of the individual missing data is dependent on the likelihood
16 R. Millham et al.

of observing other remaining variables such as y3 , …, yk . Assuming that y1 and y2

are discrete values, then the joint likelihood is the summation of all possible values
of the two variables which have the missing values in the dataset. Consequently, the
joint likelihood is denoted as:

f i∗ (yi3 , . . . , yik ; θ ) = f i (yi1 , . . . , yik ; θ )
y1 y2

As the missing variable are continuous, the joint likelihood is the integral of all
potential values of the two variable that contain the missing values in the dataset.
Thus, the joint likelihood is expressed as:

f i∗ (yi3 , . . . , yik ; θ ) = f i (yi1 , yi2 , . . . , yik )dy2 dy1

y1 y2

Because each observation adds to the determination of the likelihood function,

then the summation (integral) is calculated over the missing values in the dataset.
The overall probability is denoted as the product of all observations. An example, if
there are x observations with complete data and n-x observations with data missing
on y1 and y2 , the probability function for the full dataset is expressed as:
x n
L= f (yi1 , yi2 , . . . , yik ; θ ) f i∗ (yi3 , . . . , yik ; θ )
i=1 x+1

The advantages of using the maximum likelihood method to extrapolate missing

values are that this method produces approximations that are consistent (in that it
produces the same or almost the same unbiased results for a selected large dataset);
it is asymptotically efficient (in that there is minimal sample inconsistency which
denotes a high level of efficiency in the missing value dataset); and it is asymptotically
normal (Allison 2012).
In Fig. 1, the maximum likelihood algorithm, with known variance parameter of
sigma, is tested using several small but representative sets of missing value matrices
with some rows containing no missing values, other containing one missing value,
and still others containing several missing values.

5 Conclusion

The chapter introduced the concept of big data with its characteristics namely
velocity, volume, and variety. It introduces the phases of big data management,
which includes data cleansing and mining. Techniques that are used during some of
these phases are presented. A new category of algorithm, bio-inspired algorithms,
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 17

is introduced with several example algorithms based on the behaviour of different

species of animals explained. Following Zang’s et al. (2010) rules for the develop-
ment of a bio-inspired algorithm, a new algorithm, KSA, is shown with its phases of
descriptive animal behaviour, mathematical modelling of this behaviour, algorithmic
development, and finally testing with results.
In this chapter, we chose a particular step of data cleansing, extrapolating
missing values, of the big data management stages to demonstrate how bio-inspired
algorithms work.

Key Terminology & Definitions

Big data—is a definition that describes huge volume and complicated data sets from
various heterogeneous sources. Big data is often known by its characteristics of
velocity, volume, value, veracity, and variety.
Bio-inspired—refers to an approach that mimics the social behaviour of
birds/animals. Bio-inspired search algorithms may be characterized by random-
ization, efficient local searches, and the discovering of the global best possible
solution.
Data imputation—is replacing missing data with substituted values.

References

Abdella, M., & Marwala, T. (2006). The use of genetic algorithms and neural networks to
approximate missing data in database. Computing and Informatics, 24, 1001–1013.
Agbehadji, I. E. (2011). Solution to the travel salesman problem, using omicron genetic algorithm.
Case study: Tour of national health insurance schemes in the Brong Ahafo region of Ghana. M.
Sc. (Industrial Mathematics) Thesis. Kwame Nkrumah University of Science and Technology.
Available https://doi.org/10.13140/rg.2.1.2322.7281.
Agbehadji, I. E., Fong, S., & Millham, R. C. (2016a). Wolf Search Algorithm for Numeric
Association Rule Mining.
Agbehadji, I. E., Millham, R., & Fong, S. (2016b). Wolf search algorithm for numeric association
rule mining. In 2016 IEEE International Conference on Cloud Computing and Big Data Analysis
(ICCCBDA 2016). Chengdu, China. https://doi.org/10.1109/ICCCBDA.2016.7529549.
Allison, P. D. (2012). Handling missing data by maximum likelihood. Statistical horizons. PA, USA:
Haverford.
Banupriya, S., & Vijayadeepa, V. (2015). Data flow of motivated data using heterogeneous
method for complexity reduction. International Journal of Innovative Research in Computer
and Communication Engineering, 2(9).
Cha, S. H., & Tappert, C. C. (2009). A genetic algorithm for constructing compact binary decision
trees. Journal of Pattern Recognition Research, 4(1), 1–13.
Dorigo, M., & Gambardella, L. M. (1997). Ant colony system: A cooperative learning approach to
the traveling salesman problem. IEEE Transactions on Evolutionary Computation, 1(1), 53–66.
Dorigo, M., Birattari, M., & Stutzle, T. (2006). Ant colony optimization. IEEE Computational
Intelligence Magazine, 1(4), 28–39.
Fister, I. J., Fister, D., Fong, S., & Yang, X.-S. (2014). Towards the self-adaptation of the bat
algorithm. In Proceedings of the IASTED International Conference Artificial Intelligence and
Applications (AIA 2014), February 17–19, 2014 Innsbruck, Austria.
18 R. Millham et al.

Fong, S. J. (2016). Meta-Zoo heuristic algorithms (p. 2016). Islamabad, Pakistan: INTECH.
Grzymala-Busse, J. W., Goodwing, L. K., & Zheng, X. (2005). Handling missing attribute values
in Preterm birth data sets.
Honkavaara, J., Koivula, M., Korpimäki, E., Siitari, H., & Viitala, J. (2002). Ultraviolet vision and
foraging in terrestrial vertebrates. Oikos, 98(3), 505–511.
Kennedy, J., & Eberhart, R. C. (1995). Particle swarm optimization. In Proceedings of IEEE
International Conference on Neural Networks (pp. 1942–1948), Piscataway, NJ.
Krause, J., Cordeiro, J., Parpinelli, R. S., & Lopes, H. S. (2013). A survey of swarm algorithms
applied to discrete optimization problems. Swarm intelligence and bio-inspired computation:
Theory and applications (pp. 169–191). Elsevier Science & Technology Books.
Kumar, R. (2015). Grey wolf optimizer (GWO). Available https://drrajeshkumar.files.wordpress.
com/2015/05/wolf-algorithm.pdf. Accessed 3 May 2017.
Lakshminarayan, K., Harp, S. A., & Samad, T. (1999). Imputation of missing data in industrial
databases. Applied Intelligence, 11, 259–275.
Little, R. J. A., & Rubin, D. B. (1987). Statistical analysis with missing data. New York: Wiley.
Longbottom, C., & Bamforth, R. (2013). Optimising the data warehouse. Dealing with large volumes
of mixed data to give better business insights. Quocirca.
Narang, R. K. (2013). Inside the black box: A simple guide to quantitative and high frequency
trading, 2nd ed. Wiley: USA. Available: https://leseprobe.buch.de/imagesadb/78/04/78041046-
b4fd-4cae-b31d-3cb2a2e67301.pdf Accessed 20 May 2018.
Quinlan, J. R. (1989). Unknown attribute values in induction. In Proceedings of the Sixth
International Workshop on Machine Learning (pp. 164–168). Ithaca, N.Y.: Morgan Kaufmann.
Selvaraj, C., Kumar, R. S., & Karnan, M. (2014). A survey on application of bio-inspired algorithms.
International Journal of Computer Science and Information Technologies, 5(1), 366–370.
Shrubb, M. (1982). The hunting behaviour of some farmland Kestrels. Bird Study, 29(2), 121–128.
Spencer, R. L. (2002). Introduction to matlab. Available https://www.physics.byu.edu/courses/com
putational/phys330/matlab.pdf Accessed 10 Sept 2017.
Tang, R., Fong, S., Yang, X.-S., & Deb, S. (2012). Wolf search algorithm with ephemeral memory. In
2012 Seventh International Conference on Digital Information Management (ICDIM) (pp. 165–
172), 22–24 August 2012, Macau. https://doi.org/10.1109/icdim.2012.6360147.
Varland, D.E. (1991). Behavior and ecology of post-fledging American Kestrels.
Vlachos, C., Bakaloudis, D., Chatzinikos, E., Papadopoulos, T., & Tsalagas, D. (2003). Aerial
hunting behaviour of the lesser kestrel falco naumanni during the breeding season in thes-
saly (Greece). Acta Ornithologica, 38(2), 129–134. Available: http://www.bioone.org/doi/pdf/
10.3161/068.038.0210 Accessed 10 Sept 2016.
Yang, X-S. (2010a). Firefly algorithms for multimodal optimization.
Yang, X. S. (2010b). A new metaheuristic bat-inspired algorithm. In Nature inspired cooperative
strategies for optimization (NICSO 2010) (pp. 65–74).
Yang, X. S. (2010c). Firefly algorithm, stochastic test functions and design optimisation. Interna-
tional Journal of Bio-Inspired Computation, 2(2), 78–84.
Yang, X. S., & Deb, S. (2009, December). Cuckoo search via Lévy flights. In Nature & Biologically
Inspired Computing, 2009. NaBIC 2009. World Congress on (pp. 210–214). IEEE.
Zang, H., Zhang, S., & Hapeshi, K. (2010). A review of nature-inspired algorithms. Journal of
Bionic Engineering, 7, S232–S237.

Richard Millham is currently an Associate Professor at Durban University of Technology in

Durban, South Africa. After thirteen years of industrial experience, he switched to academe and
has worked at universities in Ghana, South Sudan, Scotland, and Bahamas. His research inter-
ests include software evolution, aspects of cloud computing with m-interaction, big data, data
streaming, fog and edge analytics, and aspects of the Internet of Things. He is a Chartered
Engineer (UK), a Chartered Engineer Assessor and Senior Member of IEEE.
1 The Big Data Approach Using Bio-Inspired Algorithms: Data … 19

Israel Edem Agbehadji graduated from the Catholic University college of Ghana with B.Sc.
Computer Science in 2007, M.Sc. Industrial Mathematics from the Kwame Nkrumah University
of Science and Technology in 2011 and Ph.D. Information Technology from Durban University
of Technology (DUT), South Africa, in 2019. He is a member of ICT Society of DUT Research
group in the Faculty of Accounting and Informatics; and IEEE member. He lectured undergrad-
uate courses in both DUT, South Africa, and a private university, Ghana. Also, he supervised
several undergraduate research projects. Prior to his academic career, he took up various manage-
rial positions as the management information systems manager for National Health Insurance
Scheme; the postgraduate degree programme manager in a private university in Ghana. Currently,
he works as a Postdoctoral Research Fellow, DUT-South Africa, on joint collaboration research
project between South Africa and South Korea. His research interests include big data analytics,
Internet of Things (IoT), fog computing and optimization algorithms.

Hongji Yang graduated with a Ph.D. in Software Engineering from Durham University, England
with his M.Sc. and B.Sc. in Computer Science completed at Jilin University in China. With over
400 publications, he is full professor at the University of Leicester in England. Prof Yang has
been an IEEE Computer Society Golden Core Member since 2010, an EPSRC peer review college
member since 2003, and Editor in Chief of the International Journal of Creative Computing.
Chapter 2
Parameter Tuning onto Recurrent
Neural Network and Long Short-Term
Memory (RNN-LSTM) Network
for Feature Selection in Classification
of High-Dimensional Bioinformatics
Datasets

Richard Millham, Israel Edem Agbehadji, and Hongji Yang

1 Introduction

The introduction describes the characteristics of big data, review on method and
search strategies for feature selection. With the current dispensation of big data,
reducing the volumes of dataset may be “achieved by selecting relevant features for
classification. Moreover, big data is also characterized by velocity, value, veracity
and variety. The characteristic of velocity relates to “how fast incoming data need to
be processed and how quickly the receiver of information needs the results from the
processing system” (Longbottom and Bamforth 2013); the characteristic of volume
refers to the amount of data for processing; the characteristic of value refers to what
a user will gain from data analysis. Other characteristics of big data include “variety
and veracity.” The characteristic of variety looks at “different structures of data such
as text and images, while the characteristic of veracity focuses on authenticity of the
data source.” While these characteristics (i.e., volume, value, variety and veracity) are
significant in any big data analytics, it is important to reduce the volume of dataset and
produce value (relevant and useful features) with reduced computational cost given

R. Millham (B) · I. E. Agbehadji

ICT and Society Research Group, Durban University of Technology, Durban, South Africa
e-mail: [email protected]
I. E. Agbehadji
e-mail: [email protected]
H. Yang
Department of Informatics, University of Leicester, Leicester, England, UK
e-mail: [email protected]

© The Editor(s) (if applicable) and The Author(s), under exclusive license 21
to Springer Nature Singapore Pte Ltd. 2021
S. J. Fong and R. C. Millham (eds.), Bio-inspired Algorithms for Data
Streaming and Visualization, Big Data Management, and Fog Computing,
Springer Tracts in Nature-Inspired Computing,
https://doi.org/10.1007/978-981-15-6695-0_2
 Another Random Document on
Scribd Without Any Related Topics
"I could not hold back my impatience a minute longer, cousin Hans,"
she said at length, with a smile. "Although my good aunt Gratz
would have had me wait upstairs in my rooms until you would find it
convenient to see me. You will forgive me for this unceremonious
assault?"
I would have loved to tell her what I really thought about it; but I
put a curb on any such madness by reflecting that her anxiety had
nothing in it personal to myself.
"It would take so long in the telling," I answered. "I can scarcely tell
it to you here."
A look of regret and surprise dashed her face for the moment, and
she withdrew her hands from mine and bit her lips.
"I have done wrong in rushing to you thus. You will think it
unseemly. Will you let me know how soon you can come to me? Do
believe, cousin, I would not wittingly do anything to displease you."
I stood silent like a dumb fool; and then after a pause she added:
"I ought to have reflected you would have many things to do, and
that I—that I should be in the way. I will go."
"No, don't go," I blurted out, and then could say no more.
She looked at me in justifiable astonishment, and wrinkled her brows
in perplexity.
"The Prince was saying as we drove here that he must see you at
once, countess," interposed von Krugen, and I could have blessed
him for the words. Then he went forward and threw open the door
of the room next, and looked round as if inviting us to enter. It was
the library.
I shook myself together with an effort and gathered my scattered
wits.
"Can you spare me an interview at once?" I asked Minna.
"Cousin!" and her astonishment deepened and found expression in
her tone. "Am I not here for that very purpose—and dying to learn
the news? Come;" and she went into the room and led the way to
the far end, as it chanced to the very window from the embrasure of
which I had first seen her. "I hope your first news is that all this plot
is at an end, and that the project of the marriage is dead with it?"
I had mastered my stupid embarrassment by this time and had
found my tongue again.
"You must listen carefully to all I have to tell you, and then to what I
propose to do," I replied, and plunged at once into as plain a recital
as I could give of all that part of the proceedings which I deemed it
necessary to tell her. I dwelt upon the reasons why in my opinion it
was impossible to draw back yet, and upon all I expected to gain by
the counterplot I had devised.
"I will not see the Count von Nauheim again," she said, and her
dislike of him was the first and strongest feeling she expressed. Nor
did I grieve at this.
"He will not come here," I said. "I was going to force a quarrel on
him to make that impossible when he saved me the trouble by
putting one on me. I then warned him off the place."
"Good, very good!" she cried, her eyes flashing. "If I were to see
him again, I could not restrain my hatred. I should tell him exactly
how I feel toward him. It is loathsome even to be linked in name
with such a man. But as that is settled, I will do whatever you wish.
I knew you would be too much for them all, cousin Hans, if they did
not kill you, as I sadly feared they would. I shall never be able to
repay you," she added, looking to me and smiling. "If I were only a
man, I could——"
"What?" I asked when she stopped.
"I could at least fight with you instead of being a clog and a drag."
"You are our inspiration," I said earnestly, and at that her cheeks
flushed and she cast down her eyes.
"I wish all the trouble were over," she said presently.
"We must not be in too great a hurry. We have done very well so far.
A little pluck and dash, and slice or two of luck, and we shall get
through all right. But now tell me, can you think of any place in
Munich, or near there, where you can go secretly and hide when the
moment comes?"
"Why must I be put out of the way in this fashion? It seems like
running away at the very moment of peril, and I am not afraid. Do
you think I am a coward?"
"This is no question of bravery or cowardice. It is merely a matter of
tactics. The very keystone of this inner plot of theirs is that you shall
be missing when the cry is raised for you to ascend the throne. To
secure that these people will stick at nothing—they would even take
your life. Now, for the success of my counter-scheme, I must be able
to have you at hand just when I want you. That is all-important. You
will have to go to Munich in apparent compliance with their wishes
for you to be ready for the final coup, and we shall show no sign of
suspicion, but you will have trusty guards to protect you against
attack. My scheme is to let them carry off some one in your place,
and for that purpose I shall endeavor to get wind of their plan of
abduction. What I wish to do is to shut out suspicion that we have
fooled them until it is too late for them to change their plans. Is
there any one among your maids whom you could trust to personate
you, who is sufficiently like you in height and color and so on to be
mistaken for you by a stranger, knowing you only by description or
having only seen you once? She would of course be dressed to
represent you, and she must be sufficiently devoted to you to take a
risk and hold her tongue."
"Yes, my dressing-maid, Marie, might pass for me under such
circumstances, and I would answer for her stanchness."
"Tell her nothing until the time is close at hand. Then let her know
what has to be done. She will wear your dress and will be carried
off; you will slip away; and I shall go in a fine rage to von Nauheim
to frighten him from getting to see his captive, and thus discover the
trick. Your present task, then, will be to get ready for that part of the
scheme, and also to think of some safe place to which you can go."
"I will willingly do more, if it will help you," she said.
The completeness of her trust in me was apparent in every word she
spoke.
"There will be plenty of exciting work to follow," I replied, with a
smile, for I was pleased by her eagerness to help. "Your Majesty
may depend upon it that a throne is not to be gained without a
struggle."
"I should make a poor Queen," she answered.
"You will make a beautiful one; and if the Bavarians once get sight
of you, they will not readily let you go."
She looked at me earnestly and, with half a sigh, said:
"You should not pay me empty compliments, cousin Hans. You
should not say things you do not mean."
"Perhaps it would be truer that I must not say all I do mean," I
returned, and for the moment my eyes spoke even more than my
words; and I made haste to add, in as light a tone as I could: "Your
Majesty will have at least one devoted subject, whatever may
happen."
"I believe that with all my heart," she answered, in a tone and with a
look of confidence and trust that thrilled me. Then she smiled very
slightly, and added: "Even one subject may make a kingdom; though
I'm sadly afraid I should not be the ruler of even such a realm."
I longed to turn her jest to earnest, and assure her that if she did
not no one else ever should; but I pulled myself up on the verge,
and remembered that, after all, I was an impostor, though loyal
enough to her. And so I made no reply, and dared not even look at
her.
After a pause she rose, and, with what sounded like a half-
suppressed sigh, she went away.
I let her go, and it was not until she had left the room that the
thought struck me that my silence might have seemed currish and
curmudgeonly. Then I would have gone after her and told her, and I
made a step toward the door; but the thought of what I should say
and how to explain my meaning stopped me, and as I hesitated
Captain von Krugen came in to resume the conference we had
commenced during the drive from the station.
 CHAPTER XII
MY SCHEME DEVELOPS
I took von Krugen into my confidence as to my discoveries and
plans. I showed him the documents I had brought back from
Munich; told him of my meeting with Praga; the secret history of the
duel which had ended young Gustav's life; and, at the close, invited
him to say plainly what he thought of the counter-scheme, and of
our chances of carrying it through.
"It is about the only chance," he said, "and once on the throne there
is no reason why the countess should not stay there."
"On the contrary, there are two overpowering reasons—her own
disinclination, and the attitude of the Imperial authorities at Berlin."
"There may be a third," he growled into his beard, looking sharply at
me.
"What is that?" I asked, though I could almost guess his meaning.
But he turned the question adroitly.
"That her Majesty would have little wish for a royal marriage with an
imperially selected consort chosen by Berlin. Her Majesty has a
heart, unfortunately, and God bless her for it."
"That will be all as she pleases," said I quietly. "At any rate, our
purpose is to give her the opportunity of declining the throne, and to
save her from these villains who would hound her down."
His face grew as dark as night.
"God! if that villain ever dares to cross her path again, I'll run my
sword through his carcass, if I die the next minute; and if he doesn't
come near her, I'll seek him out the moment this business is
through, and make him fight me. He has put not one but a thousand
insults on me—and he a traitor all the time. And to think the Prince
believed in him implicitly to the last. And so did I."
"Maybe the Prince had not the private knowledge of the man that I
had, nor had you," I said unguardedly.
My companion started and looked at me in such surprise that I saw
my blunder in a moment.
"You had known him previously?" he asked slowly.
"I had known of him," I answered in a tone of indifference. "It's a
long story, and I may tell it you some day."
"It is not for me to question your Highness, of course, but I should
never betray a confidence," he replied, piqued, as I thought, that I
said no more; and for the moment I was hugely tempted to tell him
the whole story.
It might be enormous value to have a stanch ally in my full
confidence for the task I had to carry through; but, on the other
hand, I could not tell how such a man would care to take his orders
from an ex-play-actor, and I decided that I dared not run a risk at
such a crisis. So I held my tongue, and sat as if my thoughts were
busy with our plans.
"There is much to do, captain," I said at length, "and we must waste
as little time as possible in consultation. In the first place, we have
to keep open a means of communicating with Praga. Are you too
well known in Munich to go backward and forward?"
"I fear so; but there is Steinitz. He is scarcely known at all there; but
he has not yet returned from where you sent him."
I had forgotten altogether about him and his mission; and, now that
the matter was recalled to me, the length of his absence gave me an
uneasy twinge. There must be some very serious cause for so long a
delay.
"He should have been back some days ago," I replied slowly.
"Probably he will be here to-day or to-morrow, at latest, and that will
be in time for our purpose. I myself shall return to Munich in a day
or two; but I have purposely made no appointment as yet, and shall
make none till the eve of my going, because, if my absence from
here were to be known in advance, it might probably be the signal
for some attempt against the Countess Minna."
"How shall you foil the attempt when it does come?" asked von
Krugen.
"By vigilance mainly; but I mean also to appear to play into this
Baron Heckscher's hands, while in reality forcing them. I shall see
him and tell him that all here will be in Munich two days before the
Court ball. That will give them time to make their plans to strike
during those two days. Further, my present idea is that for the whole
of those two days the character of the countess shall be doubled;
this waiting-maid of hers will be dressed precisely as she herself is
dressed the whole time, and, except when any one comes to the
house who is in the house, and who knows the countess on sight,
the girl will be the countess to every one. This means that the
servants we take with us must be strangers, with the exception of
one or two on whom we can rely implicitly. And I depend on you to
make the selection."
"There are several here for whom I would answer as for myself; but
isn't there a risk in so long a doubling of the parts?"
"Maybe; but we must be content to take it. My object is so to
arrange matters that we ourselves shall virtually select the moment
when they will try to get hold of my cousin. Thus I shall make it
quite plain to them that during every moment of every hour she is in
Munich she will be strictly watched and guarded by us; but I shall
manage to let a weak link appear in the chain, and I have chosen
this one. During the two days I shall give it out that my cousin is not
well, and can only receive one or two persons. But there is to be a
reception at the palace by the King on the afternoon of the day of
the Court ball, and I shall let it appear that our vigilance must be
relaxed on the return drive from the palace to the house. It will
seem an excellent opportunity for them. But while the countess shall
go herself to the reception, I shall arrange for the maid to take her
place on the return drive with the Baroness Gratz, and my cousin will
make a sufficient change of dress in the retiring-rooms to enable her
to leave the palace unknown."
"But the Baroness Gratz?"
"You have no doubt of her loyalty?" I asked sharply. "Speak out
plainly if you have."
"None in the least. I have no cause. I meant, what of the danger to
her?"
"There will be little or none. They may indeed be glad to let her get
away, while they will do her no harm even if they keep her prisoner.
But the points in favor of such a scheme outweigh all against it. It
will suit both them and us to have the abduction made as close to
the time of the ball as possible—them, because we should then have
no time to make a disturbance; us, because the shorter time we
have to keep watch over von Nauheim to prevent his finding out the
deception the better. A few hours later we shall be absolute masters
of the situation."
"It's a scheme that stirs one's blood," cried von Krugen warmly. "But
those few hours will be anxious ones."
"Meanwhile the Duke Marx will have been caught in the toils set for
him, and will be in our power; the King will be taken at the ball, and
thus our whole course will be clear. The mimic ceremony of
abdication will take place, the cry will be raised for the Queen Minna,
and just when they are chuckling that she cannot be found I shall
lead her forward and put her in the place of honor, and make some
sort of speech in her name—probably to the effect that she will take
time to consider her course. They will be thus caught like rats in a
hole they themselves have undermined; and there will be a pretty
tableau."
"And then?"
"Well, our first step will be to look out for ourselves. The attack on
me and you will commence at the moment they believe they have
outwitted us; and the danger will spread to us all the instant they
find we have outwitted them. But our holding of their duke as a
hostage will disarm them."
"You are sure of Praga, and that he can get hold of the duke?"
"I am sure of no one but you," I returned; "and of nothing except of
things as they occur from hour to hour. We can only lay our plans
and do our best to carry them out; but in such a case any instant
may see the unexpected happening, and the shipwreck of the best
laid scheme. But I like Praga's lever—a woman is a most useful
mechanism when you understand how to use her; and when I left
Praga every vein of his was burning with a raging lust for revenge.
And he is a Corsican. But if that part of the scheme fails, we must
patch up another way, that's all. I mean to be stopped by nothing."
"By Heaven, but you're a man I love to follow!" cried my companion,
his eyes kindling with enthusiasm.
Then I saw his expression change, and he peered curiously at me.
"And to think you've never been anything but a student. One might
think you had lived in the atmosphere of intrigue all your life. The
Prince little knew you. He believed you were a milksop. How he
would have loved you for a man after his own heart. Some one must
have been lying to him sorely about you."
"Dead slanders are of no import to us, captain, nor living flattery
either," I said shortly. "We have to plan out our respective work and
to set about doing it."
And with that I told him precisely that part of the plan which would
fall to his share, and gave him suggestions as to the best way of
carrying it out. When I had fully instructed him, I sent him away,
and mapped out in my thoughts the further developments I had yet
to plan.
The absence of Steinitz gave me much uneasiness. It seemed so
grossly out of perspective that a big scheme such as was on hand
should be endangered by a trumpery little matter like the selling of a
couple of farms. Yet that was the fear I had. If Steinitz had been
able to find von Fromberg and to give him my message, he ought to
have been back long since; but if he had not found the man, I could
not stop the sale of the property. Yet if it went on it was almost
certain that the old lawyer would in some way get into
communication with the men who were selling the place for von
Fromberg, and my identity would at once be questioned.
I would have paid the money, of course, willingly enough; but
obviously I could not buy an estate from myself. Again, I could not
get over the difficulty in any such way as I had employed with Praga
—that it was a freak.
The more I considered the thing the easier it appeared to me that I
might be tripped up and exposed through it; and when the whole of
that day passed without the return of Steinitz, my anxiety grew fast.
He arrived on the following afternoon, but he brought no relief with
him. He had not found von Fromberg. He had gone to Charmes, and
had arrived there after the wedding had taken place, and then he
had set out to follow the bride and bridegroom on their tour. He had
traced them from hotel to hotel, to Nancy, Bar-le-Duc, Rheims,
Amiens, and thence to Paris; but in the French capital all sign of
them was lost, and after making many useless inquiries there he had
deemed it best to return to me and bring back the letter. I told him
he had done right, but the incident added to my disquiet. It was
such a contemptibly little thing, and yet, like a poisonous pin-prick, it
threatened to gangrene the whole venture.
To add to my annoyance and perplexity, moreover, the old lawyer
came to me again on the following day to tell me that further
negotiations had taken place for the sale of the farms, and he
pestered me to know whether I really meant to sell them out of the
family, and whether the Count von Nauheim, as the Countess
Minna's future husband, ought not to be told of the matter. His
manner showed that he had a suspicion that something was being
kept from him, and he resented it strongly.
It was obvious, of course, that if he went to von Nauheim the latter
would jump at the chance of giving me trouble, and that if any
suspicions were even hinted to him the results might be exceedingly
awkward. Yet I could do nothing; and I was so irritated by the
lawyer's persistence that I sent him away with a sharp reply that if
he wished to retain my business he had better mind his own.
I could see he was vastly astonished at this and I more than half
repented my words, but he had gone before I had quite recovered
my temper. It was unbearable, however, that just when I had all the
weight of a really important crisis on my shoulders I should be
worried by a trumpery thing of this sort. I let him go, therefore, and
tried to dismiss the matter from my thoughts, while I went on with
the completion of my plans.
Everything else went as well as we could have wished. Minna herself
entered heart and soul into the work, and in the many interviews we
had during the next few days I could not have wished for a more
loyal and trusty ally. Our little confidential conferences drew us very
close together, moreover, and I saw with great delight that her
spirits brightened.
The preparations for the critical work in Munich occupied her so fully
that her thoughts were taken away from the grief caused by the
death of her father, while the belief that success in our venture
would open up a new life for her by freeing her from the marriage
with von Nauheim and from the dreaded responsibilities of the
throne raised hopes which brought with them happiness such as she
had not known for months.
"I owe it all to you, cousin," she said once, for she grew to speak
with absolute candor and unrestraint to me. "If only you had come
to Gramberg earlier, I am sure you would have persuaded my father
to abandon the scheme altogether; although I think sometimes that
——"
"Well?" I asked when she paused.
"That it is a good thing you did not come earlier."
Her eyes were laughing, and the light in them was a pleasing thing
to see.
"Perhaps it is. But why do you think so?"
"You have a way of making unpleasant things pleasant; and you
might have persuaded me to do what he wished."
"There are not many women who would need much persuasion to
be a Queen."
"Without conditions, perhaps."
"There is one condition I would never have advocated," said I,
raising my eyes to hers. "But you will be a Queen after all, and we
your humble servants, wishful only to obey your royal commands."
"I have settled one of the first uses I shall make of my power," she
said, looking up and speaking as if seriously.
"And that will be?"
"You will be the object of it. I shall issue an order in council—Privy
Council."
"Privy Council! You are getting learned in the jargon of State. I am
afraid your Privy Council will be a very small one."
"Yes," she cried, nodding her head and smiling. "We two. And the
order will be that my chief councillor shall tell me all the story of his
life. If you won't tell it to your cousin, you must tell it to your Queen.
And I know there are secrets in it. You think I don't take notice of
you, I suppose; and never know when your thoughts are slipping
away to the past and never see that you fence with my questions,
and glide away so cleverly from the little traps I lay. You mustn't
think because you would make me a Queen that I have ceased to be
a woman—and, being a woman, to be curious."
"We have no time in these days——"
"There you go," she laughed. "I know what you'll say. You never
think of the past because you are so busy thinking of all this
business; that when a man is planning a big scheme like this, and
has all the details to arrange, he has no time, etc., etc. But you have
a secret, cousin Hans—a secret that is never out of your thoughts;
that has nothing to do with all this fresh trouble and intrigue; that
took you away from the castle for two days just after you arrived;
and that has written its lines on your face. That may be because you
can find no one to tell it to. Of course you think of me only as a girl
—you self-contained strong men always do that—and that I should
make no sort of a friend to be trusted with secrets. And yet——" she
paused, and laying her hand gently on mine said softly and wistfully,
"you have done so much for me I should like to be a little help to
you. Can I, cousin? I'm not Queen yet, you know, and cannot
command. I'm only a grateful girl, and can do no more than ask."
I was not a little disconcerted to find that she had been watching me
so closely, and I could not remain untouched by the last little appeal.
But I could not reply to it.
"You are a stanch little comrade," I answered. "But we must put off
the story until the Queen commands," I answered, smiling.
"That is at least an open postponement, if not a frank refusal. But
the Queen will command, cousin. I want to know why you would not
come here at the first; what made you change your mind; how it
was that all our ideas about you were wrong; why you are so
different from what we all expected—oh, there are a thousand
questions that sting the tip of my tongue with the desire to ask
them."
"You think a student cannot also be a man of affairs?" I said, divided
between pleasure at her interest in me and perplexity at her
questions.
"But you are not even a student. You never open a book; you never
quote things—ah, now you start because I have watched you. I can
read your eyes, although you think you can drape them with the
curtains of impassiveness. But your wit is not always on guard to
draw the curtains close enough. Yes, that's better; now they are
saying nothing."
All this time she had been looking straight into my eyes, and
laughing in gleeful triumph. And I found it embarrassing enough.
Then she changed suddenly, and said:
"Does my teasing worry you and weary you, cousin? I can school my
curiosity if it does. But you will tell me all some day?"
"Is that schooling it?" I asked, and she laughed again. "Yes, I will tell
you some day what there may be to tell. But it could do no good to
do so yet."
"Is it a sad secret?" she began again after half a minute's silence,
and would no doubt have gone on with her pretty cross-examination
had we not, fortunately for me, been interrupted by a servant, who
brought word that Steinitz, whom I had sent to Munich, had
returned, and was asking to see me instantly.
"I hope there is no trouble?" said Minna, looking alarmed.
"I anticipate none; no more, that is, than that we must break off our
conference."
"You have given me your promise," she said.
"I ought to have made a condition—that you do not read me quite
so carefully," I answered lightly as I rose.
"Then I have read aright? To me your eyes are as books."
"Yet you must be careful how you read them," said I.
"Why?"
"You may chance on the chapter with your name at the head."
"I wish I could," and she laughed and her eyes brightened. "I would
give the world to know whether it is headed Queen of Bavaria or
cousin Minna. Which is it? Tell me, at least, so much."
"It may be neither," I answered ambiguously; but she seemed to
understand something of my meaning, and to be pleased, for her
cheeks were aglow with color as I hurried away.
Steinitz was awaiting me impatiently.
"There is ugly news, your Highness," he said shortly. "I saw Praga
early this morning, and he bade me urge you to hurry at once to
Munich. He has got wind of a move on the other side, which he
prefers to tell to you alone. He will meet you to-morrow at noon
where you met before, and he declares that the strictest vigilance
must be used in regard to the countess, especially while you are
away from the castle, and that your visit to the city should be made
with the greatest secrecy."
"He told you nothing more of what he had discovered?"
"No more than I say. But I gathered his meaning to be that an
attempt of some kind is imminent to get the countess out of our
hands here."
This was likely enough, but I did not take so serious a view of the
matter as Praga, because I felt that when I had explained our
movements to Baron Heckscher he would be almost sure to select
the moment when the thing could apparently be done with the least
risk of discovery, and that would be at the last moment, when Minna
returned from the palace after the reception.
At the same time I would go to Munich. I had already planned to go
there on the following day in any event, and had announced my
intention; but I settled to start at once. I sent for von Krugen and
told him, charging him to keep the strictest watch over Minna; and
after a very brief interview with her, in which she showed the
liveliest concern for my safety, mingled, as it pleased me to think,
with regret at our separation, I started with Steinitz on what I knew
might be a critical expedition.
 CHAPTER XIII
A CHECK
Matters were now hurrying fast to a crisis; and I hoped the result of
my journey would be to complete all my preparations, and leave me
nothing to do but return to escort Minna to Munich. So far all had
gone well enough. I had no reason to think that either Heckscher or
von Nauheim had the remotest idea that I knew of their treachery;
and it was, of course, of the very essence of my plan that they
should remain in ignorance. On this account I was unwilling to meet
Praga again personally, and I resolved therefore to send Steinitz to
him as soon as we reached Munich to tell him my intentions, and to
get from him in return what he believed to be the Ostenburg move.
I myself went straight to Baron Heckscher. He received me with
apparent cordiality; but it was not difficult to see that as the day of
the crisis drew near his anxiety was growing.
"All is going well, I hope," I said, after I had greeted him. "We have
all our preparations made."
"All is going very well," he replied. "But you are a day earlier in
Munich than we anticipated."
"I have not come to remain," I answered, "although I have some
important business. My cousin is not well; and her nerves are giving
way as the day approaches. I have difficulty in keeping her courage
up. Like a woman, she has some foolish fear that at the last moment
something will happen to her—some disaster to overthrow her. But I
have nearly conquered that fear, I trust."
"How?"
"She associates the fear with her visit here, and I have assured her
that night and day, every hour and every minute, she herself will be
surrounded by absolutely stanch friends who would give their lives
for her. The death of her brother just at the moment when success
seemed to be within grasp is frightening her. Nor is that unnatural,
especially when we reflect that her nerves have again been strained
by her father's death."
My words had the effect I desired. It did not suit his plans that
Minna should be guarded in this way.
"The Countess is not ill, I trust," he said after a pause.
"Oh, no, not positively ill. But she is very young, and so full of
alarms that even I myself am inclined at times to question the
wisdom of all this." Perceiving the value of the line I had taken, I
went on to make the most of it. "Indeed, I want some very
confidential talk with you. You understand that I am resolved to go
on, and I have not breathed a word to suggest to her that there is
even an alternative course; but there are two points on which I wish
to consult you. In the first place, is it quite impracticable to abandon
the thing? I am convinced my cousin would only too gladly renounce
all claim to the throne."
He looked at me sharply and with manifest consternation.
"It is absolutely impossible, Prince, absolutely," he said emphatically.
"But you are not in earnest. Why, it would be madness, sheer
madness to think of such a thing. Since you were here we have
sounded men in all directions, and there is not one who is not
enthusiastic at the idea of getting rid once and for all of this
madman."
"But my cousin can only make a weak Queen at the best."
"My dear Prince, her weakness will be the strength of the country.
Our great object is not so much to change the person of the ruler as
to break the traditions of the ruler's power—to put on the throne
some one whose title will rest, not on any right divine, but on the
people's power and will and choice. A woman will thus be far more
dependent on the people than a man. Prince, the countess cannot
draw back."
"But supposing she were willing to acquiesce in the election of the
Ostenburg heir, and thus unite all sections of the people?"
"It is impossible, equally impossible!" he exclaimed readily. "It would
be a betrayal of us all. It is not to be thought of."
I sat as if thinking this over, but in truth this prompt rejection of the
means to do fairly what I knew he was plotting to do by foul had
filled me with anger.
"And what would be the immediate consequences of a withdrawal?"
I asked.
"Do you mean the personal consequences to the countess and
yourself?" he asked, with a suggestion of contempt for such a
consideration.
"I mean to all concerned."
"What could but be the consequences where three-fourths of a
nation had been worked up to desire a revolution and found
themselves cheated at the last moment by the—the timorousness of
those in whose name and for whose sake the whole movement has
been carried out? The badge of cowardice is a hard one to bear,
Prince, and the anger of a disappointed people would not lighten the
disgrace."
"We are no cowards, Baron Heckscher," I replied warmly, as if stung
by his taunt.
"Then you must not so act that people may mistake you."
"We will not," I returned, with an air of angry decision.
"I was sure of it, and am only sorry you thought it necessary even to
moot the suggestion. But now what is your second point? Not
another objection, I hope."
"It is merely to discuss with you the last arrangements. Under the
circumstances you will, I am sure, see the necessity for making them
as simple as possible—indeed, my cousin's health will not permit
anything else."
"Up to the moment of our great coup they cannot possibly be too
simple. Anything else would be a great mistake. Up till somewhere
about midnight of this day week, Wednesday next, the countess is of
course no one but the very charming young lady that I am assured
she is—I mean she is a private person. In that capacity she will
attend the reception, and in order that there may be no suspicion
attaching to her making a public appearance so soon after her
father's death it has been arranged that a special desire for her
attendance shall be expressed by the King. She will merely attend,
kiss hands, and pass through the presence chamber, and leave the
palace at once, should it be desired. She can return home and go to
the ball, where she should be at about ten o'clock. She must be at
hand of course when the great drama is played in which we are to
take part. When the Act of Abdication has been read, you will lead
her forward. That is all. We shall do the rest."
"And what will follow then?"
"I think she will stay at the palace. It is just in the few hours
succeeding that scene that we shall have to be alert. The King will
be missing, and a Council of State will be called on the following
morning, when she will be proclaimed to the country. After that,
events will settle themselves rapidly. We are prepared with a petition
to the Imperial authorities, which will be signed by nearly every man
of influence in the country, to recognize the succession and validate
the abdication."
"But that Act of the King will surely be found to be a forgery?" I said.
My companion smiled and shook his head.
"On the contrary, it will be genuine. We should not use such clumsy
means as forgery. We have it already written. For once his Majesty's
lunacy has done his subjects a good service," he said bitterly. "He
was minded recently to play a farce of abdication in favor of one of
his hounds, declaring with his customary facetiousness that the
Bavarians were dogs, and a fit King for them would be a hound.
Accordingly he held what he was pleased to call a Privy Council—
consisting of himself and his dogs. But those about him knew their
business, and when he thought he had abdicated in favor of his dog
they fooled him to the top of his bent, but drew the document in
such a way that the insertion of the countess's name would be an
easy matter. The addition of a date will make everything complete;
and thus when the madman thought he was only insulting his
people, he was in fact signing away his throne. He had this dog, a
clever poodle, seated in the chair in the Council Chamber, garbed in
State robes, and crowned with the crown of Bavaria. I tell you,
Prince, that one act would stir the blood of even a nation of cravens
—and we Bavarians are no cowards. My blood boils at the thought,"
he cried, clenching his fist, while his eyes flashed, and his face,
usually immobile and cold, lighted up with the fires of passion.
I joined him in a hot outburst of indignation.
"But the time is past for mere anger," he said presently. "We are
resolved to act; and that farce of his shall cost him dear. As to Berlin,
so soon as we have driven home the conviction that we are in dead
earnest, and that practically the whole country is with us, there will
be no opposition. The usual official intimation will be published that
the King's health has failed, and the rest follows naturally."
"But you are forgetting the Ostenburg interest."
"I forget nothing, Prince," he replied, somewhat curtly. "I know the
public feeling. The very inaction they are showing will make the
Duke Marx impossible in the eyes of the people. While the country
has been writhing and suffering under the insults and iniquities of
this madman, what have the Ostenburgs done? Has one of them
raised a finger to help the people or protest against this royal
mumming? Has any one of them said a word? And how do you
suppose the nation is to interpret that silence and inaction, except as
approval of what has been done? They had the better right of
succession and a strong following on their side; they have forfeited
the one by their apathy and have lost the other as a consequence;"
and he went on to give many reasons for this conclusion.
"I admit," he said at the close, "there will be some anxious hours
just after the Countess Minna is proclaimed; but, with all the will in
the world, they can do nothing. I tell you there is nothing can stay
our success nor shake your cousin's seat on the throne when she
has once taken it."
I allowed myself to appear to share his convictions, even while I
marvelled at the depth of his duplicity, and I then told him the plan
of our movements. He listened closely, and made several
suggestions which I said we would adopt; and he quite acquiesced
in my view that during the time Minna was to be in Munich she
should remain in the greatest seclusion, giving audience only to
himself and two or three others.
When I left him my task in Munich was practically finished, so far as
he was concerned; but he advised me to attend a reception at the
palace on the following day but one, the Friday, and I agreed. I felt
sure I had left the impression I had gone to create—that their best
time for abducting Minna would be at the moment of her return from
the palace; and I completed my arrangements on that basis.
Steinitz was waiting for me at the hotel with an important
communication from Praga, giving me the particulars of an intended
attempt to carry off Minna from Gramberg during the night; and
though it seemed to me a mad scheme enough, and pretty certain
to be abandoned after my interview with Baron Heckscher, I
despatched Steinitz post-haste back to the castle to put von Krugen
on his guard. Whether it were abandoned or not, the fact that we
had knowledge of it would render it certain to fail, and I felt no great
anxiety on that score.
But I soon had cause for anxiety in another direction. The two men
whom I had asked to visit Gramberg had not been there, and we
were, in fact, perilously short-handed for all the work that had to be
done. I was the more anxious, too, to get extra help because of a
weak spot in my plans, which I could not remedy without further
assistance.
If the Ostenburg agents held the person of the King, and I
checkmated them at the last moment by producing Minna and
keeping their duke in confinement, there was a chance that they
might counter my stroke by bringing the mad King back on the
scene, and thus checkmate me in turn. The only means of
preventing this would be to secure that those who held the King in
custody should be loyal to Minna; and it was for this part of the
scheme that I had hoped to make use of the two men, Kummell and
Beilager. I set out to find them, therefore.
I chanced upon them together at the house of Kummell, and it did
not take me a minute to perceive that there was a decided restraint
in their manner toward me. I had meant to be perfectly frank with
them, telling them, indeed, all I knew; but their attitude made this
impossible, and for a moment I was at a loss what line to take.
While gaining time to think, I talked at large upon the importance of
the affair generally, and at length asked them point-blank why they
had not been to Gramberg.
"We have been very busy," replied Kummell, who spoke for both;
and the answer was rather curtly given.
"Scarcely a sufficient reason, gentlemen, in an affair of this sort," I
replied in quite as curt a tone, "nor, I presume, the only one."
They hesitated, and glanced at one another.
"I think you must excuse us if we do not answer the question. In
point of fact, I am not yet in a position to do so."
"I cannot understand you, and, under the circumstances, I must
really press you very closely to be frank with me," I urged; and,
although they still hesitated and equivocated, I was resolved not to
leave without an answer, and I told them as much.
"You put us in a very awkward position, indeed, but the fact is we
had intended to make the visit, and had fixed the day, when we
were advised not to do so by Herr Bock."
"And who is Herr Bock, pray?"
So utterly unsuspicious was I of any possible mischief that I put a
good deal of indignation into the question. Yet it was a blunder of
the grossest kind, and the reply astounded me utterly.
"Herr Bock is your own lawyer, who has been negotiating the sale to
me of your late mother's property."
That confounded property again!
My four years' training on the stage stood me in good stead now,
and I masked my surprise with a laugh as I exclaimed:
"Oh, that Bock! I did not know it was you who were contemplating a
purchase. But why should that keep you away from Gramberg? Were
you afraid that a look at the property would put you out of conceit
with the bargain, or that I should charge you more, thinking you
were growing eager?"
But there was more in this than a laugh could carry off.
"No, but he has been in communication with your old family lawyer,
and together they say or think they are on the track of some kind of
strange complication which I believe in some way touches yourself;
how I do not know, but Bock advised me to wait."
"This has a somewhat serious sound, sir," I said, sternly enough to
cover my apprehension.
"I cannot help that. You asked me, pressed me, indeed, for an
answer to your question. In times like these you will understand I
feel great need to be cautious—overcautious perhaps you may deem
it. But still here it is."
"And what is the nature of this supposed ridiculous complication?"
"You must excuse me if I say no more. You know Herr Bock's
address here in Munich."
The scent was getting warm.
"I shall of course see him," I answered readily. "And I will find a
short method of dealing with a couple of meddlesome attorneys as
soon as this business of next week is through. And what then do you
propose to do?"
"I think we had better not discuss any matters except in the
presence of Baron Heckscher."
I rose to leave. I had met with my first serious check.
"I thought I could have relied implicitly upon your loyalty to the
House of Gramberg," I said loftily.
"To the House of Gramberg, yes," was the answer, stolidly spoken,
yet with a significance I could not mistake.
I went back to my hotel angry and apprehensive. I could have
twisted von Fromberg's neck for his maladroitness in hurrying to sell
his property, and then getting beyond my reach and keeping there.
Moreover, I could not see what to do. These two bungling old fools
of lawyers had no doubt been comparing notes, and probably
comparing the different handwritings of von Fromberg and myself;
and had hatched a pretty cock-and-bull story about me. Probably
they were already making all sorts of inquiries. Yet I dared not go
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

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

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com