Page 34 - Fister jr., Iztok, and Andrej Brodnik (eds.). StuCoSReC. Proceedings of the 2016 3rd Student Computer Science Research Conference. Koper: University of Primorska Press, 2016
P. 34
to get answers to the following questions: 3.1 Current publishing era
• What is actually considered as a new nature-inspired Without doubt we can say that this era, after the year 2000,
algorithm? led to the hard race between scientists and even institu-
tions. People that work in universities and institutes are
• What motivates researchers to propose new algorithms? forced to publish works that will achieve a lot of citations,
since good works with many citations help universities in
• What they have if they propose a new algorithm? global rankings. Better universities have attracted more,
better students. In line with this, they could obtain the
• What is a generic recipe for proposing a new algo- government and industrial projects more easily. Projects
rithm? mean money, while more students mean more tuition fees.
Fortunately, this is not true for each country. For exam-
• What are the implications of new algorithms? ple, there are no tuition fees for students of Government
Institutions in Slovenia. Only, doctoral studies require stu-
• How to stop the invasion of new algorithms? dents to pay a tuition fee that is actually not as high as it is
abroad. In order to meet these goals universities could make
• Is proposing new algorithms basically swarm intelli- pressure on researchers to publish more and better works.
gence behavior itself? This manner is also connected with the term Publish or
perish. However, to satisfy these goals is a little bit easier
Insights on these questions will be highlighted in the remain- for mature and well-known researchers, while newbies have
der of the paper. mostly enormous problems. For example, some students in
China are even prepared to give a kidney for having a paper
2. NATURE-INSPIRED ALGORITHMS accepted in a journal with impact factor. Is not publish-
ing becoming similar to any fight sport (win at any price)?
At the start, it is very hard to define exactly what the Let us leave politics and go back to our population-based
population-based nature-inspired algorithms actually are. How- nature-inspired algorithms. So, after the year 2000 when
ever, there are many definitions and most of these definitions Ant colonies and Particle Swarms became popular, inter-
say that population-based nature-inspired algorithms are a esting and widely used algorithms, some researchers began
kind of algorithms that are inspired by natural, biological thinking whether there was a possibility to develop or just
and even social systems, and are intended to solve problems propose a new algorithm that should be based on any other
in a similar way to what nature does. Even today, there inspiration from nature. If we look into our bioshpere, we
are a few taxonomies that try to deal algorithms in different can easily find a lot of animal species, different trees, natural
groups. One of the taxonomies is a taxonomy published in processes, social behaviors for developing the optimization
2013 by Fister et al. [5] where algorithms were split into 4 algorithms. When researchers found an inspiration, they
groups [5]: then needed to coin some operators that mimic the behavior
of their inspiration and, later, put this magic into the univer-
• Swarm intelligence based algorithms, sal recipe that was presented in the previous chapter. Most
of the algorithms only used a different formula for moving
• Bio-inspired that are not swarm intelligence based, individuals and that was all the magic behind an algorithm.
Paradigm was the same, but only some minor changes were
• Physics and chemistry based and incorporated in the developing of the brand new population-
based nature-inspired algorithm. After developing, the time
• Other algorithms. has started for publishing the new algorithm. Usually, all re-
searchers need to validate their new algorithms on some well-
Algorithm 1 Generic pseudo-code of most of the known benchmark functions or on some engineering prob-
population-based nature-inspired algorithms lems. Of course, all algorithms have beaten other well-
1: initialize individuals within bounds using a particular known algorithms without any problem, although nobody
cared if researchers used special benchmarks and tested on
randomization generator special dimensions and compared with basic well-known al-
2: evaluate all individuals gorithms and not with their improved versions. At the be-
3: while termination criteria not met do ginning, they were successful and achieved good publications
4: move all individuals according to proposed formulas in journals with nice impact factors and even at good con-
5: evaluate all individuals ferences. Until 2010, almost nobody had yet cared about
6: find the best individuals new algorithms but, after 2010, many scientists began to
7: end while doubt about the originality of works. However, it was too
8: return the best individual and vizualize late. Nowadays, we are in 2016. According to the list
on Github (www.github.com/fcampelo/EC-Bestiary),
Generic pseudo-code for most of the algorithms in this tax- we counted (as of 5 August 2016) that there are 101 nature-
onomy, especially for the first and second group, is presented inspired algorithms (Fig. 1). Anyway, there are many others
in Algorithm 1. that are not on this list.
3. THE NEW POPULATION-BASED
NATURE-INSPIRED ALGORITHMS
The previous section gave readers a short overview of the
population-based nature-inspired algorithms, while this sec-
tion will be concentrated on the implications of the new
population-based nature-inspired algorithms.
StuCoSReC Proceedings of the 2016 3rd Student Computer Science Research Conference 34
Ljubljana, Slovenia, 12 October
• What is actually considered as a new nature-inspired Without doubt we can say that this era, after the year 2000,
algorithm? led to the hard race between scientists and even institu-
tions. People that work in universities and institutes are
• What motivates researchers to propose new algorithms? forced to publish works that will achieve a lot of citations,
since good works with many citations help universities in
• What they have if they propose a new algorithm? global rankings. Better universities have attracted more,
better students. In line with this, they could obtain the
• What is a generic recipe for proposing a new algo- government and industrial projects more easily. Projects
rithm? mean money, while more students mean more tuition fees.
Fortunately, this is not true for each country. For exam-
• What are the implications of new algorithms? ple, there are no tuition fees for students of Government
Institutions in Slovenia. Only, doctoral studies require stu-
• How to stop the invasion of new algorithms? dents to pay a tuition fee that is actually not as high as it is
abroad. In order to meet these goals universities could make
• Is proposing new algorithms basically swarm intelli- pressure on researchers to publish more and better works.
gence behavior itself? This manner is also connected with the term Publish or
perish. However, to satisfy these goals is a little bit easier
Insights on these questions will be highlighted in the remain- for mature and well-known researchers, while newbies have
der of the paper. mostly enormous problems. For example, some students in
China are even prepared to give a kidney for having a paper
2. NATURE-INSPIRED ALGORITHMS accepted in a journal with impact factor. Is not publish-
ing becoming similar to any fight sport (win at any price)?
At the start, it is very hard to define exactly what the Let us leave politics and go back to our population-based
population-based nature-inspired algorithms actually are. How- nature-inspired algorithms. So, after the year 2000 when
ever, there are many definitions and most of these definitions Ant colonies and Particle Swarms became popular, inter-
say that population-based nature-inspired algorithms are a esting and widely used algorithms, some researchers began
kind of algorithms that are inspired by natural, biological thinking whether there was a possibility to develop or just
and even social systems, and are intended to solve problems propose a new algorithm that should be based on any other
in a similar way to what nature does. Even today, there inspiration from nature. If we look into our bioshpere, we
are a few taxonomies that try to deal algorithms in different can easily find a lot of animal species, different trees, natural
groups. One of the taxonomies is a taxonomy published in processes, social behaviors for developing the optimization
2013 by Fister et al. [5] where algorithms were split into 4 algorithms. When researchers found an inspiration, they
groups [5]: then needed to coin some operators that mimic the behavior
of their inspiration and, later, put this magic into the univer-
• Swarm intelligence based algorithms, sal recipe that was presented in the previous chapter. Most
of the algorithms only used a different formula for moving
• Bio-inspired that are not swarm intelligence based, individuals and that was all the magic behind an algorithm.
Paradigm was the same, but only some minor changes were
• Physics and chemistry based and incorporated in the developing of the brand new population-
based nature-inspired algorithm. After developing, the time
• Other algorithms. has started for publishing the new algorithm. Usually, all re-
searchers need to validate their new algorithms on some well-
Algorithm 1 Generic pseudo-code of most of the known benchmark functions or on some engineering prob-
population-based nature-inspired algorithms lems. Of course, all algorithms have beaten other well-
1: initialize individuals within bounds using a particular known algorithms without any problem, although nobody
cared if researchers used special benchmarks and tested on
randomization generator special dimensions and compared with basic well-known al-
2: evaluate all individuals gorithms and not with their improved versions. At the be-
3: while termination criteria not met do ginning, they were successful and achieved good publications
4: move all individuals according to proposed formulas in journals with nice impact factors and even at good con-
5: evaluate all individuals ferences. Until 2010, almost nobody had yet cared about
6: find the best individuals new algorithms but, after 2010, many scientists began to
7: end while doubt about the originality of works. However, it was too
8: return the best individual and vizualize late. Nowadays, we are in 2016. According to the list
on Github (www.github.com/fcampelo/EC-Bestiary),
Generic pseudo-code for most of the algorithms in this tax- we counted (as of 5 August 2016) that there are 101 nature-
onomy, especially for the first and second group, is presented inspired algorithms (Fig. 1). Anyway, there are many others
in Algorithm 1. that are not on this list.
3. THE NEW POPULATION-BASED
NATURE-INSPIRED ALGORITHMS
The previous section gave readers a short overview of the
population-based nature-inspired algorithms, while this sec-
tion will be concentrated on the implications of the new
population-based nature-inspired algorithms.
StuCoSReC Proceedings of the 2016 3rd Student Computer Science Research Conference 34
Ljubljana, Slovenia, 12 October
