See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/346957172

Artificial-intelligence-for-improving-the-optimization-of-nphard-problems-
A-review2020International-Journal-of-Advanced-Trends-in-Computer-
Science-and-Engineering

Article in International Journal of Advanced Trends in Computer Science and Engineering · October 2020
DOI: 10.30534/ijatcse/2020/73952020

CITATIONS READS

5 1,741

3 authors, including:

Saadia Bahti Jaafar Abouchabaka

Université Ibn Tofail Université Ibn Tofail
2 PUBLICATIONS 13 CITATIONS 118 PUBLICATIONS 714 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Saadia Bahti on 11 December 2020.

The user has requested enhancement of the downloaded file.

 ISSN 2278-3091
Volume 9, No.5, September - October 2020
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420
International Journal of Advanced Trends in Computer Science and Engineering
Available Online at http://www.warse.org/IJATCSE/static/pdf/file/ijatcse73952020.pdf
https://doi.org/10.30534/ijatcse/2020/73952020

Artificial Intelligence for Improving the Optimization

of NP-Hard Problems: A Review
Said Tkatek1,Saadia Bahti2, YounesLmzouari3, Jaafar Abouchabaka4
1
LaRI, Faculty of Sciences, IbnTofail University, Morocco,[email protected]
2
LaRI, Faculty of Sciences, IbnTofail University, Morocco, [email protected]
3
LaRI, Faculty of Sciences, IbnTofail University, Morocco, [email protected]
4
LaRI, Faculty of Sciences, IbnTofail University, Morocco, [email protected]

 effective and valid for all data.

ABSTRACT
In the real world, the resource ("human", "machine", etc.)
Artificial intelligence (AI) techniques play an important management problem arises in various contexts, and the
role in the development and optimization of complex literature is of interest for RAP, which is an NP-hard problem
problems in various fields. This study presents a systematic [3], [4] and proposes various practical applications [1]. In
analysis of previously publications on the use of AI general, the classical assignment problem consists in finding a
techniques to optimize the NP-hard problems in the area of one-to-one correspondence between a set of tasks and a set of
human resource assignment to multi-constrained activities in agents, whose objective is to minimize the total cost of the
order to increase the productivity and performance of a agents' work. Several examples of variance of the assignment
company, and in the area of vehicle routing in order to cost problem can be found in the literature, such as: Assignment of
minimization. These two problems belonged to the NP-hard jobs to machines, jobs to workers, workers to machines,
family. Among these AI techniques used for a better students to groups, or computations to compute nodes... etc.
optimization of this type of problems, we discuss the genetic Like other combinatorial optimization problems, the RAP has
algorithms (GA), hybrid GA, parallel GA, GA with Big Data
been studied and solved by exact methods, specific heuristics,
and improved AI techniques with special features are also
meta-heuristics and by the different ways of hybridization of
discussed and presented through the different contributions of
these approaches [1]. The objective is to find the optimal
the researchers.
assignment that minimizes the total cost or maximizes the
overall benefit [5]. In addition, the assignment problem is
.Key words: Artificial Intelligence, Big Data, Genetic
usually a sub-problem of a complex problem such as the
Algorithm, NP-hard problem, Optimization.
travelling salesman problem TSP, the Job Shop Scheduling
problem JSP, ...etc.
1. INTRODUCTION
Similarly, in the field of logistics and supply chain
In a highly competitive environment, organizations are led to management, the vehicle routing problem is considered one of
integrate human resources into their core strategy. Aware that the most important and practical problems. It is a
the latter plays a decisive role in the success of the combinatorial optimization and operations research problem.
organization, managers of organizations are looking for more First introduced by Danzig and Ramser as a generalisation of
efficient and more optimal tools to identify the adequate the travelling salesman problem TSP. Whose principle is to
profiles (resources) of each position/task to maintain and cover a transport network to deliver goods to customers or to
develop their productivity [1], [2]. This has led managers to cover the roads of this network by one or more vehicles. The
solve their problems, no matter how difficult, by introducing resolution of this problem requires its decomposition into a
optimization techniques. Solving an optimization problem sub-problem of assignment (the assignment of customers to
consists in using radically different methods in order to obtain vehicles) [6]. As a result of environmental concerns, a new
an optimal (maximum or minimum optimal) solution. category of VRPs has emerged that aims to minimise
environmental costs on the one hand, and the VRP's common
In everyday life, especially in the industrial field, problems operational costs on the other hand [7].
are complex and large in size, making them difficult to solve.
However, the competition between optimization search Given the continuous increase in the complexity of problems
methods distinguishes a type of problem classified as that are difficult to solve, researchers aim to exploit the
difficult, whose name is NP-hard problem. To date, none of principles and techniques of artificial intelligence, operational
the proposed algorithmic solutions can be considered research and the various methods of metaheuristics, in order
to solve these problems using the different forms of
hybridization of these techniques and methods. As well as
7411
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

BigData techniques that allow to exploit interesting data for if and only if it satisfies these two conditions: P ∈ NP and ∋ P′
improved decision making. ∈ NP such as P′ is reduced to P by a polynomial algorithm
[10].
Artificial intelligence techniques mainly consist of (a)
Evolutionary algorithms EA (Genetic algorithm GA, Genetic According to the literature, resource assignment consists of
Programming GP, Evolutionary Programming EP, Evolution matching human resources to tasks, machines, projects,
Strategies ES, Differential Evolution DE), (b) Fuzzy logic FL, teams, sites, schedules, etc. [11], its considered of one of
(c) Swarm Intelligence SI (Particle swarm optimisation PCO, NP-hard problem [3],[4]. The resource assignment issue is
Ant Colony Optimisation ACO, Artificial Bee Colony ABC), present in any organization that depends on resources to
and (d) Artificial neural network ANN [8]. These techniques generate tangible throughput or provide services [11]. The
have been developed and studied by a large number of determining criteria for staff assignment are related to the
researchers and developers. They have shown significant personality of the resource, its skills and the work
potential for solving difficult problems. environment .

One of the most answered and advanced search techniques of Other studies have presented the problem of resource
artificial intelligence is the genetic algorithm. It is a allocation such as the process of allocating resources between
meta-heuristic derived from the evolutionary phenomenon of different projects, units, etc. [1],[12],[13]. In order to achieve
biological organisms. Its objective is to find the most optimal a specific objective of the organization, this resource may be a
solution for a given problem [9]. For this reason, the genetic person, property, equipment or capital.The personnel
algorithm has been selected mainly for examination in the assignment problem was first addressed by Votaw and Orden
solution of resource assignment problems, as well as its in 1952 [14], illustrating a method of assignment that helps to
hybridization, although other techniques of artificial match a task to the most appropriate agent, in order to achieve
intelligence. the most efficient result. Many articles address the RAPunder
several variances that appear in several areas such as: staff
Therefore, this paper is a summary of existing literature on the transfer/mobility [15]–[18], scheduling [19],[20],
resource assignment problem. To do this, we are looking for Timetabling Problem [21]–[23], maintenance [24] and also
larger studies of RAP and VRP problem. team formation [21],[25]–[28].

Although this work cannot be exhaustive, it covers a large In recent decades, the resource assignment problem has been
number of bibliographical references. In Section 2, we present widely considered by researchers in several fields. Notably
a description of the complex resource assignment problem. the resource allocation in the educational field, Indeed, Hertz
The section 3 is devoted to the genetic algorithm: illustration and Robert[1], [29] present the assignment problem as a
of the principles of GA and its variances, algorithmic sub-problem of the course scheduling problem, teachers are
formulation, and the application examples of GA to solve the assigned to courses, then courses are assigned to time slots,
resource assignment problem, as well as its variances. In the and finally time slots are assigned to classrooms. In the
sections 4 we presentthe various variances of GA: Basic GA, literature, the timetabling problem appears in applications
hybridization and Parallelism. Pf GA. The section 5describes related to time slot assignment such as conferences, exams,
the importance of bringing together Big Data techniques and courses, meetings, etc. [22],[23]. On the other hand, the
Genetic algorithms in solving the assignment problem. In assignment problem arises in robotics and control theory due
Section 6, we will distinguish the other methods and to applications involving the assignment of tasks or objectives
approaches of Artificial Intelligence used to solve this to agents (Robots) [30].
problem. Finally, we conclude the paper and raise prospects
for future research. The RAP in software engineering is the simultaneous
allocation of multiple developers to multiple projects in order
2. ABOUT THE NP-HARD PROBLEMS: RESOURCE to maximize both developer and project attributes. Otero [31],
ASSIGNMENT AND VR PROBLEMS [32] argue that it is beneficial to generate systematic
The resolution of difficult combinatorial optimization employee allocation processes that take into account the skill
problems very often relies on so-called "approximate" set of candidates and provide the best fit in order to increase
methods. They do not aim at solving a problem in an optimal quality, reduce costs, and decrease training time, and they
way. A combinatorial optimization problem is NP-hard if and propose a multi-criteria decisional approach to allocating
only if its decision variant is an NP-complete problem, or if resources to software engineering tasks. In addition, Palacios
the complexity of an optimization problem is related to that of [25], [27] present a recommendation system designed to help
the decision problem associated with it. In particular, if the project managers configure multiple teams for the work
decision problem is NP-complete, then the optimization packages defined by Scrum Projects. The system is based on
problem is said to be NP-hard [2]. On the other hand, we can fuzzy logic, coarse set theory and semantic technologies.
say that a problem is NP-complete if and only if it belongs to Silva and Costa [25], [28] presented a framework for human
NP and NP-hard. Also, a decision problem P is NP-complete resource allocation in information systems projects. The main
7412
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

objective is to determine the adequacy between the set of 3.1 Genetic Algorithm
skills available in a candidate member of a team project and Among the instances of evolutionary algorithms, we find the
the skills required for the latter in order to minimize the time genetic algorithms which are according to the literature the
needed for the realization of a requested project. most used to solve optimization problems. The basic
principles of the genetic algorithm were first developed by
For large (multi-site) organizations, the management of Holland [35]. It is an artificial intelligence research
resource mobility is always a problem [17], [33]. S. Acharyya meta-heuristic, inspired by the laws of species evolution and
[15] presented the problem of personnel transfer as a general natural genetics. The genetic algorithm (GA) is one of the
problem of stable matching, comparing different methods most important artificially techniques because of its
(simulated annealing, genetic algorithms, taboo search). The flexibility, parallelism and speed in finding the optimal
work consists in matching two lists of preferences, one of the solution close to the global solution. Genetic algorithms
employee and the other of the employer, so that a maximum attempt to simulate the law of evolution of species in their
number of employees are allocated transfers, while reducing natural environment: an artificial transposition of the basic
the number of blocked pairs (job, site). S. Tkatek [34] presents principles of genetics and the laws of survival enunciated by
an expert system to manage limited human resource mobility Darwin. Genetic Algorithms are methods for exploring all the
in a multi-site company. It is based on models with constraints solutions of a problem using the same mechanisms as those
on reassignment, HR recruitment and the genetic algorithm to involved in natural selection (Selection, Crossover,
solve these models. On the other hand, S. Tkatek [17] has Mutation)(Figure 1). They are mainly used in the fields of
developed an expert system to simplify the management of optimization and learning.
human resources mobility within a multi-site company
without taking into account the hard constraints arising from Yes
the complex problems of mobility. This system integrates Population Evaluation Converged Solution
initialization of Cost
several algorithms such as the bilateral permutation
(transposition) algorithm, circular permutation algorithm and
virtual (factious) post algorithm. No

Other problem type studied in this paper is the Vehicle Mutation Selection
Routing Problems (VRP) that are part of a spectrum of
operational research that has existed for several decades. This
involves determining the routes of a fleet of vehicles in order
to deliver a list of customers, or carrying out intervention or Crossover
visit tours. In practice, this is often considered equivalent to
minimizing the total distance traveled, or minimizing the Figure 1: Flowchart of optimization with a genetic algorithm [36]
number of vehicles used, and then minimizing the total
distance for this number of vehicles. This problem is a classic A Genetic algorithm consists in implementing a set of main
extension of the traveling salesman problem, and belongs to operations to allow the passage from one generation to
the class of NP-hard problems. another. These operations are defined as sequences:

3. GENETIC ALGORITHM AS AN ARTIFICIEL Initialization: The first step of the genetic algorithm is to
INTILLEGENCE METHODE TO OPTOMIZE THE produce an initial population of individuals according to a
COMPLEX PROBLEM uniform (coding), non-homogeneous (variation of
The optimization of hard problems is the subject of a class of individuals) distribution that is the starting point for future
problems whose resolution cannot be obtained by an exact generations.
method in polynomial time. For this, optimization methods
have been widely used to obtain the optimal solution Selection: this is a mechanism that consists of designating the
(minimum, maximum optimal). The resolution of this class of best susceptible individuals on the basis of their performance
problems is then done by heuristic and meta-heuristic that could be crossed to reproduce the future population. The
methods, hence the use of artificial intelligence techniques. literature proposes different selection methods, citing:
The latter is mainly composed of the evolutionary algorithms selection by tournament, by rank, random selection, etc [37].
EA (Genetic algorithm), Fuzzy logic FL, Swarm Intelligence
SI (Particle swarm optimization PCO, Ant Colony Crossover: the search for the solution space is done by
Optimization ACO), and Artificial neural network ANN [8]. creating new generations from the old ones. The evolution of
The literature review proves the wide use of genetic the population in each generation is obtained thanks to the
algorithms to solve complex problems. In this section, genetic selection of individuals for the crossing phase. Crossover is
algorithms will be defined and then essential studies that use the operation of exchange of genetic material (part) between
these algorithms to solve the HRAP will be presented. two selected individuals of the population (parents), to give
birth to one or two new individuals named child(ren).

7413
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

Different crossover types of a genetic algorithm exist in the [19] proposes a genetic algorithm to solve a spatial
literature, for example: uniform crossover operator, cycle assignment problem of human resources in multi-site
crossover, ordered crossover, crossover with reduced enterprises. H. Algethami [42] conducted a comparative
surrogate, shuffle crossover [36]. study of the various genetic operators of two simple GAs
addressed to solve a workforce Scheduling and routing
Mutation: the individuals (children) resulting from the problem.
cross-breeding operations will be subjected to the mutation
operation as a final step in the creation of a new generation. 4. VARIANCES OF GENETIC ALGORITHM
This operation consists of modifying one or more genes at
4.1 Hybrid Genetic Algorithm
randomly chosen locations in the chromosome by replacing it
with a random value. Thus, a random modification of the Combinatorial optimization problems occur in different
solution's characteristics, allowing the introduction and application areas [43] such as production, health, engineering,
maintenance of a sufficient level of genetic diversity within education. The assignment problem is a classical
the population of solutions. combinatorial optimization problem [43], looking for the
optimal solution to assign a resource to a task, machine,
Evaluations: the evaluation phase is the main key to the project, team, site, etc. [10]. The various fields of application
genetic algorithms, whose objective is to evaluate the and the importance of these problems have led to the birth of
performance of the new generated individuals. This operation broad methods of solution, as far as Operational Research
is done using the objective function to calculate the fitness or (OR) and Artificial Intelligence (AI) [43], [44].
selective value, associated to each individual. During the
selection process, these values will determine the candidates In recent years, so-called "Hybrid" solving methods have
apt for reproduction. gained increasing popularity for solving hard optimization
problems [44].
Replacement: A closing phase in the generating process of a
new population of individuals (solutions). Its purpose is to The literature shows that in recent years, the number of
build up from the parent and child populations a new algorithms that do not only follow the concepts of a single
population for the next iteration. Therefore, deciding who Metaheuristic is increasing, they combine different
should stay and who should disappear. There are different algorithmic ideas, coming from other resolution approaches
selection strategies, depending on the criteria chosen: such as Operational Research (OR) and Artificial Intelligence
selection by descent (children automatically replace their (AI), these resolution approaches are called "Metaheuristic
parents regardless of their fitness), selection by competition Hybrids" or "Metaheuristic Hybrids" [44].
(the survivors are the winners of the competition between
parents and children), elitist selection (only the best Having a more efficient system is the main motivation behind
individuals of a population (parents and children) are retained the hybridization of metaheuristics, exploiting and uniting the
for the new generation). advantages of the chosen approaches. In fact, the choice of an
adequate hybrid method is the solution to obtain high level
Stopping criteria: The population generation and selection performance in solving real world problems [44].According
process of a genetic algorithm is repeated until a stopping to the literature, different meta-heuristic hybridization
criteria is met. This may be one of the conditions mentioned in methodologies can be distinguished as follows:
[38]. Once the algorithm is stopped, the best individuals in the
population are retained as the solutions to the initial problem. - The combination (parts) of different metaheuristics.
- By choosing problem-specific algorithms to be
combined with metaheuristics.
3.2 Genetic Algorithm Application on NP_hard Problems: - By combining metaheuristics with other more
case of HRAP general approaches from operational research (OR)
Thanks to their flexibility and performance in solving hard and artificial intelligence (AI) [44].
optimization NP-hard problems, genetic algorithms have
many examples of applications in different fields [9]. In general, hybridization is the mechanism of combining two
or more classical optimization methods into a single
The literature presents a wide variance of GAs developed and algorithm. Indeed, according to E.-G. Talbi [45]-[47],
applied to address resource allocation problems. In the hybridization approaches, especially hybrid metaheuristics,
industrial field, and in order to minimize total conveyor generally give good results, and are now gaining popularity
downtime, G. Celano [39], [40] use GA to solve a worker for solving combinatorial optimization problems. In addition,
assignment problem in a manual mixed-model U-shaped hybridization methods are implemented for the resolution of
assembly line. R. Dornberger [18], [41] applied genetic HRAP.
algorithms to solve the train drivers assigning problem to
regular train services. Based on multiple constraints, S.Tkatek

7414
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

Jat and Yang [48] propose two algorithms, the first is genetic The type 2 class contains parallelization strategies in which
hybrid, the second is based on taboo research to solve the source of parallelism comes from the explicit
problems of course planning after registration. In order to decomposition of the search space by partitioning the set of
solve the problem of work planning, assignment of workers, decision variables into disjoint subsets that are optimized by a
Tao et al [49] propose a Petri network modeling method and a set of processors. The same algorithm is executed on each
hybrid genetic algorithm and simulated annealing algorithm processor but on different regions of the search space.
(GASA). Minzu and Beldiman [50], [51] propose a
meta-heuristic hybrid based on the genetic algorithm and the In the type 3 class, several search threads evolve
Kangoroo algorithm to solve the single-machine scheduling simultaneously in parallel from the different initial solutions.
problem. Similarly, in 2007, the latter used this hybridization Thus, they explore different regions of the search space. The
to present a discrete optimization system, solving two real main objectives of this class of parallel strategies are to
problems in the manufacturing domain: single-machine improve the final solution by exploring different parts of the
scheduling problem and workstation task assignment problem search space, and at a second level, to accelerate the overall
[51]. execution time. Indeed, parallel searches are generally
executed for a shorter period of time than the sequential
In addition, S. Tkatek [52] presents a new hybrid algorithm algorithm. In some metaheuristics, the total number of
system that improves the resolution of a staff reassignment iterations in the sequential algorithm is divided by the number
problem by proposing an optimal solution. This system is of simultaneous searches.
based on the implementation of two hybrid algorithms:
Hybrid Flow Genetic Algorithm with Adaptive Immigration Hybrid models can also be created from these three types. For
Genetic HFGA-AIG, and Hybrid Flow Genetic - Sequentially example, type 3 parallelism can be used at the top level and
Simulated Annealing Algorithms with Adaptive Immigration type 1 or type 2 parallelism at the bottom level. This will
Genetic HFGA-SA-AIG.Chen et al. [53], [54] present a create a kind of hierarchical parallelism.
meta-heuristic hybrid called "hybrid evolutionary algorithm",
combining the genetic algorithm with an extreme PARALLEL
optimization method, to solve a scheduling problem in the METAHEURISTICS
manufacturing domain. Zobolas, Tarantilis, and Ioannou [55],
[56] present a hybrid method. While combining a construction
heuristic, a GA and a VNS (variable neighbourhood search).
4.2 Parallel Genetic Algorithm Type 1 TExplicit Type 3
Operation Space Multi-Sear
With the development of parallel architectures, parallelism is Parallelizat Decompos ch
used as a means of solving larger problems, and more ion ition Threads
complex combinatorial problems. Parallelism in optimization
algorithms is used for various reasons. The traditional goal is
to accelerate the algorithm execution time to more complex
parallels of strategies in which parallelism also aims at Independent Co-operative
Search Search
diversifying the search in different regions (such as in parallel
AGs). In the following we focus on parallel metaheuristics,
studying a classification proposed by Crainic and Toulouse in Figure 2: Taxonomy of parallel metaheuristics [57]
[57]. This classification is followed by a section on Parallel
Genetic Algorithm (PGA). • Parallel Genetic Algorithms
• Parallel Metaheuristic Classes There are two main possible methods for parallelism. The first
In [57], three generic classes of parallel metaheuristics are is data parallelism, where the same instruction will be
defined and studied in the context of three main executed on many data simultaneously. The second is control
metaheuristics: genetic algorithms, simulated annealing, and parallelism, which involves the simultaneous execution of
tabu search. Parallelization strategies applied to various instructions [58]. Data parallelism is sequential in
metaheuristics are classified according to the source of nature because only the data manipulation is parallelized
parallelism. The three classes of parallelization strategies are while the algorithm will be executed as a sequential
shown in Figure 2. instruction within a certain time. Thus, the majority of parallel
genetic algorithms opt for data parallelism.
Type 1 class: contains parallelization strategies in which the
sequential algorithm is unmodified. Thus, the path explored is Parallel genetic algorithms arise from the need for
the same in both algorithms. For example, in GAs, the computation for extremely complex problems for which the
evaluation and transformation steps can be parallelized. running time using sequential genetic algorithms is a
limitation [59]. The use of parallel genetic algorithms aims at
breaking down a problem into several sub-problems, solving

7415
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

them simultaneously on several processors, thus improving existence of lost data. The feature selection process becomes
search performance and increasing the probability of finding an important step in many Data Mining and Machine
the best solution. Learning algorithms to reduce the dimensionality of the
optimization problems in question. It is with this in mind that
In general, parallel algorithms behave in the same way as researchers Tareq Abed Mohammed, OguzBayat, Osman N.
sequential algorithms, but with different small systems. In Uçan and ShaymaaAlhayali in their paper [65], proposed a set
theory, parallel genetic algorithms can be divided into of hybrid and effective genetic algorithms to solve the
subtasks to maintain a certain balance in the distribution of characteristic selection problem. These algorithms use a new
activities. Then, each processor can process one of the gene weighted mechanism capable of adaptively classifying
sub-populations derived from the initial population of the characteristics into three types (relative strong, weak and
genetic algorithm. There are several methods for parallelizing unstable) during the algorithm's evolution. The performance
a genetic algorithm. The first and most intuitive is the global of the proposed algorithms is studied using different datasets
one, which essentially consists in parallelizing the fitness and characteristic selection algorithms.
function assessment of individuals holding a single stock. A
better option for parallelization of the genetic algorithm is to For comparison, they have used the all following algorithms
divide the population into sub-populations that evolve for in their test:
separately and exchange individuals after a number of  Low Weighted Gene Genetic Algorithm (LWGGA) that
generations [59]. uses a proposed weight-based mechanism to exclude low
characteristics of the selected Dataset, if their weight is very
The literature presents different studies that have applied low.
parallel genetic algorithm to solve the HRAP, such as, by  High Weighted Gene Genetic Algorithm (HWGGA)
using parallel genetic algorithm and simulated annealing, Fan that uses the proposed weight-based mechanism to select and
Yang implement solutions to solve the traveling salesman use strong characteristics in new solutions or sets of generated
problem [60]. Similarly, HarunRasit [61] parallelised a characteristics, if they have very high weights.
genetic algorithm on Hadoop Cluster by using MapReduce  Weighted Gene Genetic Algorithm (WGGA) that uses
Framework. both the low and high weighting mechanisms described in the
previous LWGGA and HWGGA algorithms.
5. BIG DATA AND GENETIC ALGORITHM  Artificial Neural network (ANN) that collects their
knowledge by detecting patterns and relationships in data.
The major increase in the amount of generated data makes it  Sequential Forward Selection (SFS), which is a
very important to develop new robust and scalable tools that bottom-up algorithm for feature selection problems; it starts
are able to extract the hidden knowledge and information the search from an empty set and gradually adds features one
from the big data sets[62]. When the dataset that we are by one.
dealing with has a massive volume of data and includes both  Sequential Backward Selection (SBS), which is the
structured and unstructured data, it is called a Big Data[63]. reverse version of the SFS algorithm for characteristic
The Big Data becomes a specific and separated field in selection problems, it starts searching among a complete set of
computer engineering society since it is hard to be processed all characteristics, then gradually deletes the characteristics
using the traditional database and software techniques. one by one.
 The genetic algorithm (GA).
Exploring knowledge in Big Data is appealing in the state of
the art of Data Mining research[64]. There are many Data The results obtained by the researchers, show that the
Mining algorithms currently in use. Most of them can be proposed algorithms could outperform other feature selection
classified in one of the following categories: Decision trees algorithms and effectively improve classification
and rules, nonlinear regression and classification methods, performance compared to the Datasets tested.
example-based methods, probabilistic graphical dependency
models, and relational learning models. In the other hand, some studies applied GA to Data Mining
learning tasks, like the one where researchers came up with
Big Data can be used for solving optimization, classification the idea of using genes to extract data with great efficiency
and NP-hard problems. The most efficient method to solve the [66]. They also showed how this extracted data can be used
problems we are going to be interested in is the genetic effectively for different purposes. Instead of being content
algorithm.Using GA over Big Data creates great robust, with the general notion of probabilities, they used the concept
computationally efficient and adaptive systems. Genetic of expectations and I modified the theory of expectations to
Algorithms can be used to improve Big Data Clustering and obtain the desired results. In addition to this, the study
the effectiveness of the Big Data Analysis. presented by researchers Tan Jun-shan, He Wei and Qing Yan
[67], have proposed a group of Data Mining algorithms based
5.1 Optimization Problem with GA-Big Data
on a genetic algorithm to find the best way to manage
The hard y in solving Big Data optimization problems is due information about a company's employees. Note that this
to many factors, including the high number of features and the

7416
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

information includes age, income level, health status, gender by Kennedy and Eberhart [71] in 1995, developed from
and skill level. swarm intelligence, which is one of the techniques of artificial
intelligence, while relying on the natural behavior of swarms
5.2 Vehicle Routing Problem
and animal gathering. Researchers, I. X. Tassopoulos and G.
The VRP belongs to a class of operational research and N. Beligiannis, in their article [72],[73] have designed a PSO
combinatorial optimization problems. It involves determining algorithm to solve a high school timetabling problems.
vehicle routes in order to deliver a list of customers, or to
carry out maintenance visits (repairs, inspections) or visits More recently, W. H. El-Ashmawi [33] proposes a new PSO
(medical, commercial, etc.). The objective is to minimize the to solve a team formation problem by minimizing the cost of
cost of delivering goods. This problem is a classic extension communication between expertscalled theimproved particle
of the business traveler problem, and is part of the NP-hard optimization with the new swap operator (IPSONSO).
class of problems.
Some researchers, like ShaoqingZheng who proposed in his 6.2 Ant Colony Optimization
article[68], a dynamic VRP model and solves it by GA. The The ant colony optimization algorithm dates back to the early
latter will make it possible to take into account the real road 1990s, introduced by Dorigo [74]. It is based on the social
conditions and achieve the desired results in the route plans. behaviour of ant colonies and their ability to find the shortest
He assumed that the real-time traffic data is updated every 15 path between a food source and their nest [1]. It is one of the
minutes in the transport network, the journey time on each most popular swarm intelligence techniques for solving
section of road has been initialized thanks to a Big Data optimization problems. Initially, this meta-heuristic has been
analysis, and customer demand is dynamically updated from proposed to tackle the problem of street vendors (TSP).
time to time. The analysis example shows that its route
adjustment model and strategy can minimize the total cost of The literature presents studies that have applied ACO
transport by transporting vehicles from several depots under algorithms to solve the HRAP problem, such as: W. Gutjahr
the time window constraints. The research results help [75] use the ant colony optimization algorithm to address a
transport companies to use vehicles efficiently and receive dynamic regional nursing planning problem in Australia to
more benefits. minimize nurse and hospital preferences and costs. The results
of the ACO approach show significant improvements. In
5.3 Clustering Problem
addition, M. Mazlan [28] developed a ACO algorithm to solve
With regard to clustering, which is an unavoidable the university courses scheduling problem.
pre-processing step for Big Data processing. It allows
engineers to better understand the structure of the data. 6.3 Artificial Neural Network
Clustering therefore consists in dividing the data into groups Artificial neural networks ANNs are based on the biological
of similar objects. Each group (cluster) involves the elements concepts of the human brain [76], which consists of a set of
that are most similar to the rest of the cluster members, and simple processing units. Signals are sent to each other via a
more different from the other elements of the group. To have a large number of weighted connections to communicate [8].
good clustering performance, researchers Different research has applied neural network techniques to
SayedeHouriRazavi, ShahrokhAsadi andHarleenKaur, in solve the problem of assignment across its different variances.
their paper [69], used an efficient grouping genetic algorithm Mengyuan Lee [77] has implemented a Deep neural network
to analyze an anonymous number of clusters in the Big Data, algorithm to solve a linear sum assignment problem, dividing
this algorithm can detect the appropriate number of data sets the problem into several sub-problems. Said Medjkouh [5]
with high clustering accuracy. proposes an Artificial Neural Network ANN to deal with
assignment problems by applying it on three different real life
In addition, AfsanehMortezanezhad and EbrahimDaneshifar, examples: military application, express company application,
in their article [70], proposed a new automatic clustering and finally Students to Projects assignment.
algorithm based on the genetic algorithm, in which it is not
6.4 Fuzzy Logic
mandatory to know the number of clusters. The proposed
algorithm uses an unsupervised learning paradigm to classify Fuzzy logic is an important area of research on which a lot of
data points into clusters. Their algorithm uses a very short research is focused. The theoretical principles of fuzzy logic
chromosome coding and proposes relevant crossing and were formulated in the 1960s by A. Zadeh [78] based on his
mutation operators leading to a very good clustering theory of fuzzy subsets. The literature offers a variety of
performance. research that uses fuzzy logic to solve the resource allocation
problem. HamedJafari [79] proposes fuzzy mathematical
6. OTHER ARTIFICIAL INTELLIGENCE programming models to solve nurse scheduling problem. In
RESOLUTION METHODS another hand, Emmanuel Ferreyra [80] applies a type-2 fuzzy
logic system to allow the estimation of the number of
6.1 Particle Swarm Optimization engineers needed to be allocated for a certain number of jobs.
Among the most recent meta-heuristics used to solve HRAP is
particle swarm optimization. This method was first introduced

7417
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

7. CONCLUSION 7. R. Alizadeh Foroutan, J. Rezaeian, and I.

In this paper, we present the artificial intelligence solving Mahdavi.Green vehicle routing and scheduling
methods developed to optimize NP-difficult problems in problem with heterogeneous fleet including reverse
logistics in the form of collecting returned goods,
several areas, in particular human resource allocation
Appl. Soft Comput. J., Vol. 94, p. 106462, 2020.
problems. As RAP and VRP problems are NP problems, this
8. H. Rahmanifard and T. Plaksina.Application of
study underlines the importance of finding an optimal solution artificial intelligence techniques in the petroleum
to them. industry: a review, Artif. Intell. Rev., Vol. 52, no. 4, pp.
2295–2318, 2019.
The literature review reveals the large number of publications 9. M. Tabassum.a Genetic Algorithm Analysis Towards
produced to cover theoretical and experimental research of the Optimization Solutions, Int. J. Digit. Inf. Wirel.
problem in different application areas. This review highlights Commun., Vol. 4, no. 1, pp. 124–142, 2014.
the growing interest in this type of research. 10. J. Chen, I A. Kanj, and G.Xia.Labeled search trees and
amortizedanalysis: Improved upper bounds for
In order to underline the interest accorded to this problem in np-hard problems, Algorithmica, no. 43,pp. 245–273,
recent years, the list of references cites various sources 2005.
(articles, conferences, books...) that deal with techniques, 11. K. Holness, C. Drury, and R. Batta.A systems view of
approaches, algorithms or theoretical developments directly personnel assignment problems,Hum. Factors Ergon.
related to this research paradigm. In particular, the use of Manuf. Serv. Ind., Vol. 16, pp. 285–307, May 2006.
Genetic Algorithms, Hybrid GA, Parallel GA, Big Data with 12. M. Osman, M. Abo-Sinna, and A. allah Mousa.An
GA to solve the problem of human resource allocation and its effective genetic algorithm approach multiobjective
variances. resource allocation problems (MORAPs), Appl. Math.
The AI-inspired solving methods applied to optimize HRAP Comput., Vol. 163, pp. 755–768, Apr. 2005.
are identified according to the chosen solving techniques. The 13. C.-M. Lin and M. Gen.Multiobjective resource
state of the art shows that evolutionary algorithms are the allocation problem by multistage decision-based
hybrid genetic algorithm, Appl. Math. Comput., Vol.
most widely used AI approaches, including the use of GA as
187, pp. 574–583, Apr. 2007.
an artificial intelligence method. 14. D. F. Votaw and A. Orden.The personnel assignment
problem, Symp. Linear Inequalities Program, Vol. pp.
The state of the art also suggests that hybrid and parallel 155–163, Jan. 1952.
algorithms are commonly used because of their ability to 15. S. Acharyya and A. K. Datta.Matching formulation of
solve difficult NP problems. Increasingly, we are finding that the Staff Transfer Problem: meta-heuristic
additional efforts are needed, combining different artificial approaches, OPSEARCH, pp. 1–40, 2019.
intelligence techniques, metaheuristic methods, and big data 16. T. Said, A. Otman, A. Jaafar, and R. Najat.A
techniques, to achieve promising hybrid methods of problem Meta-heuristically Approach of the Spatial
solving. Consequently, much remains to be done to improve Assignment Problem of Human Resources in
the optimization of real NP-hard (HRAP) problems based on Multi-sites Enterprise, Int. J. Comput. Appl., Vol. 78,
big data and GAs parallelism. no. 7, pp. 15–21, 2013.
17. T. Said, A. Otman, A. Jaafar, and R. Najat.An Expert
REFERENCES System to Simplify the Management of Human
Resources Mobility, J. Comput. Sci., Vol. 3, no. 1, 2018.
18. Y. Zhu and L. Wu. Structure Study of Multiple
1. S. Bouajaja and N. Dridi.A survey on human resource Traveling Salesman Problem using Genetic
allocation problem and its applications, Oper. Res., Algorithm, 2019 34rd Youth Acad. Annual Conf. of
Vol. 17, no. 2, pp. 339–369, 2017. Chinese Association of Automation, 2019, pp. 323-328.
2. S. Tkatek, O. Abdoun, J. Abouchabaka, and N. 19. H. Algethami, A. Martínez-Gavara, and D. Landa-Silva.
Rafalia.An optimizing approach for multi constraints Adaptive multiple crossover genetic algorithm to
reassignment problem of human resources, Int. J. solve workforce scheduling and routing problem, J.
Electr. Comput. Eng., Vol. 6, no. 4, pp. 1907–1919, 2016. Heuristics, Vol. 25, no. 4–5, pp. 753–792, 2019.
3. S. Tkatek, O. Abdoun, J. Abouchabaka, and N. Rafalia.A 20. J. Lin, L. Zhu, and K. Gao. A genetic programming
hybrid heuristic method to solve an assignment hyper-heuristic approach for the multi-skill resource
problem of human resource, Int. Rev. Comput. Softw., constrained project scheduling problem, Expert Syst.
Vol. 10, no. 9, pp. 977–986, 2015. Appl., Vol. 140, p. 112915, 2020.
4. A. Corominas, R. Pastor, and E. Rodríguez.Rotational 21. O. Ramos-Figueroa, M. Quiroz-Castellanos, E.
allocation of tasks to multifunctional workers in a Mezura-Montes, and O. Schütze. Metaheuristics to
service industry, Int. J. Prod. Econ., vol. 103, Sep. 2006. solve grouping problems: A review and a case study,
5. B. Xue and G. B. Hacene.Sparse Clustered Neural Swarm Evol. Comput., Vol. 53, p. 100643, 2020.
Networks for the Assignment Problem, COGNITIVE 22. M. Yusoff and N. Roslan. Evaluation of Genetic
2017no. c, pp. 69–75, 2017.
Algorithm and Hybrid Genetic Algorithm-Hill
6. P. Siarry.Metaheuristics, Switserland: Springer, 2016.

7418
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

Climbing with Elitist for Lecturer University 36. O. Abdoun and J. Abouchabaka. A Comparative Study
Timetabling Problem, pp. 363–373, 2019. of Adaptive Crossover Operators for Genetic
23. M. Mazlan, M. Makhtar, A. F. K. Ahmad Khairi, and M. Algorithms to Resolve the Traveling Salesman
A. Mohamed. University course timetabling model Problem, Int. J. of Comput. Appli., Vol. 31, no. 11, 2011.
using ant colony optimization algorithm approach, 37. J. Dreo, A. PETROWSKI, P. Siarry, and E. Taillard.Méta
Indones. J. Electr. Eng. Comput. Sci., Vol. 13, no. 1, pp. heuristiques pour l’Optimisation Difficile,2003.
72–76, 2019. 38. C. Reeves. Genetic Algorithms, in Handbook of
24. M. Touat, F. Benbouzid-Si Tayeb, S. Bouzidi-Hassini, Metaheuristics, M. Gendreau, JY. Potvin, International
and B. Benhamou.A fuzzy genetic algorithm for Series in Operations Research & Management Science,
single-machine scheduling and flexible maintenance 2010, vol 146, pp. 109-139.
planning integration under human resource 39. A. Ammar, H. Pierreval, and S. Elkosentini. Workers
constraints, in Proc. - Int. Conf. Tools with Artif. Intell. assignment problems in manufacturing systems: A
ICTAI, 2018, 2017-Novem, pp. 551–558. literature analysis, in Proc. 2013 Int. Conf. Ind. Eng.
25. A. Costa, F. Ramos, M. Perkusich, A. Freire, H. Almeida, Syst. Manag. IEEE - IESM 2013, 2013, no. October.
and A. Perkusich.A search-based software engineering 40. G. Celano, A. Costa, S. Fichera, and G. Perrone. Human
approach to support multiple team formation for factor policy testing in the sequencing of manual
scrum projects, in Proc. Int. Conf. Softw. Eng. Knowl. mixed model assembly lines, Comput. Oper. Res., Vol.
Eng. SEKE, 2018, Vol. 2018-July, no. June, pp. 474–479. 31, no. 1, pp. 39–59, 2004.
26. W.H.El-Ashmawi, A. F. Ali, and M. A. Tawhid. An 41. R. Dornberger, L. Frey, and T. Hanne. Single and
improved particle swarm optimization with a new Multiobjective Optimization of the Train Staff
swap operator for team formation problem, J. Ind. Planning Problem Using Genetic Algorithms, 2008
Eng. Int., Vol. 15, no. 1, pp. 53–71, 2019. IEEE Congress on Evolut.Comput., 2008, pp. 970-977.
27. R. Colomo-Palacios, I. González-Carrasco, J. L. 42. H. Algethami and D. Land_silva. Diversity Based
López-Cuadrado, and Á. García-Crespo. Resyster: A Adaptive Genetic Algorithm for a Workforce
hybrid recommender system for scrum team roles Scheduling and Routing Problem, 2017 IEEE
based on fuzzy and rough sets, Int. J. Appl. Math. Congress on Evolut.Comput.,2017, pp. 1771-1778.
Comput. Sci., Vol. 22, no. 4, pp. 801–816, 2012. 43. R. Guefrouchi and M.-K. Kholladi. Parallel Hybrid
28. L. C. e Silva and A. P. C. S. Costa. Decision model for Meta-Heuristic Genetic Algorithm with Hill Climbing
allocating human resources in information system to Resolve the QAP, The 2011 Int. Arab Conf. on Inf.
projects, Int. J. Proj. Manag., Vol. 31, no. 1, pp. Techn. (ACIT'2011), 2011, no. December.
100–108, 2013. 44. G. R. Raidl, J. Puchinger, and C. Blum. Metaheuristic
29. A. Hertz and V. Robert. Constructing a course schedule hybrids, Int. Ser. Oper. Res. Manag. Sci., Vol. 272, pp.
by solving a series of assignment type problems, Eur. 385–417, 2019.
J. Oper. Res., Vol. 108, no. 3, pp. 585–603, 1998. 45. E.-G. Talbi. A Taxonomy of Hybrid Metaheuristics, J.
30. M. M. Zavlanos and G. J. Pappas. Dynamic assignment Heuristics, Vol. 8, pp. 541–564, Jan. 2002.
in distributed motion planning with local 46. E.-G. Talbi, S. Cahon, and N. Melab. Designing cellular
coordination, IEEE Trans. Robot., Vol. 24, no. 1, pp. networks using a parallel hybrid metaheuristic on the
232–242, 2008. computational grid, Comput. Commun., Vol. 30, pp.
31. C. Kandemir and H. A. H. Handley. Work process 698–713, Feb. 2007.
improvement through simulation optimization of task 47. E.-G. Talbi and El-Ghazali. Metaheuristics: From
assignment and mental workload, Comput. Math. Design to Implementation, New Jerzey: Wiley, 2009,
Organ. Theory, Vol. 25, no. 4, pp. 389–427, 2019. vol. 74.
32. C. E. Otero, L. D. Otero, I. Weissberger, and A. Qureshi. 48. S. Jat and S. Yang. A hybrid genetic algorithm and
A multi-criteria decision making approach for tabu search approach for post enrolment course
resource allocation in software engineering, timetabling, J. Sched., Vol. 14, pp. 617–637, Dec. 2011.
UKSim2010 - UKSim 12th Int. Conf. Comput. Model. 49. Z. Tao, T. Xiao, and C. Hao. Petri net and GASA based
Simul., 2010, pp. 137–141. approach for dynamic JSP, Int. Conf. on Mechatronics
33. S. Tkatek, O. Abdoun, J. Abouchabaka, and N. Rafalia. A and Automation, 2007, 2007.
multiple knapsack approach for assignment problem 50. V. Minzu and L. Beldiman. A parallel hybrid
of human resources, J. Theor. Appl. Inf. Technol., Vol. metaheuristic for the single machine scheduling
87, no. 3, pp. 374–379, 2016. problem, Proc. IEEE Int. Symp. Assem. Task Plan.,2003,
34. S. Tkatek, O. Abdoun, and J. Abouchabaka. An expert vol. 2003-Janua, no. August 2003, pp. 134–139s.
system for a constrained mobility management of 51. V. Minzu and L. Beldiman. Some aspects concerning
human resources, 2017 Int. Colloq. Logist. Supply the implementation of a parallel hybrid
Chain Manag. Compet. Innov. Automob. Aeronaut. Ind. metaheuristic, Eng. Appl. Artif. Intell., Vol. 20, no. 7,
LOGISTIQUA 2017, 2017, pp. 53–58. pp. 993–999, 2007.
35. J. H. Holland. Adaptation in natural and artificial 52. S. Tkatek, O. Abdoun, J. Abouchabaka, and N. Rafalia. A
systems, An Introd. Anal. with Appl. to Biol. Control hybrid genetic algorithms and sequential simulated
Artif. Intell., Jan. 1994. annealing for a constrained personal reassignment

7419
Said Tkatek et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(5), September - October 2020, 7411 – 7420

problem to preferred posts, Int. J. Adv. Trends Comput. Big Data Analysis. Adaptation, Learning, and
Sci. Eng., Vol. 9, no. 1, pp. 454–464, 2020. Optimization, Vol. 19,Acharjya D., Dehuri S.,
53. Y. W. Chen, Y. Z. Lu, and G. K. Yang. Hybrid SanyalS.,Springer, 2015.
evolutionary algorithm with marriage of genetic 70. A. Mortezanezhad and E. Daneshifar. Big-Data
algorithm and extremal optimization for production Clustering with Genetic Algorithm, 2019 5th Conf. on
scheduling, Int. J. Adv. Manuf. Technol., Vol. 36, no. Knowl. Based Engin. and Inn. (KBEI), 2019.
9–10, pp. 959–968, 2008. 71. J. Kennedy and R. Eberhart. Particle Swarm
54. C. Blum, J. Puchinger, G. R. Raidl, and A. Roli. Hybrid Optimization, Proc. Fourth IEEE Int. Conf. Neural
metaheuristics in combinatorial optimization: A Networks, 1995, Vol. 4, pp. 1942-1948.
survey, Appl. Soft Comput. J., Vol. 11, no. 6, pp. 72. O. E. Musa and A. Abdelaziz. Swarm Intelligence For
4135–4151, 2011. Educational Timetabling: A Survey Of The State Of
55. G. I. Zobolas, C. D. Tarantilis, and G. Ioannou. The Art, IJARP, Vol. 3, no. 9, 2019.
Minimizing makespan in permutation flow shop 73. I. X. Tassopoulos and G. N. Beligiannis. A hybrid
scheduling problems using a hybrid metaheuristic particle swarm optimization based algorithm for high
algorithm, Comput. Oper. Res., Vol. 36, no. 4, pp. school timetabling problems, Appl. Soft Comput. J.,
1249–1267, 2009. Vol. 12, no. 11, pp. 3472–3489, 2012.
56. S. Dağ. Hybrid Metaheuristic for the Permutation 74. M. Dorigo, V. Maniezzo, A. Colorni, and M. Dorigo.
Flowshop Scheduling Problems, Journal of Positive Feedback as a Search Strategy, Tech. Rep.
Multidisciplinary Engineering Scie.and Technology, Vol. 91-016, no. June, pp. 1–20, 1991.
2, no. 2, 2015. 75. W. Gutjahr and M. Rauner. An ACO algorithm for a
57. T. G. Crainic and M. Toulouse. Parallel Strategies for dynamic regional nurse-scheduling problem in
Meta-Heuristics, Handb. Metaheuristics, no. April, pp. Austria, Comput. Oper. Res., Vol. 34, pp. 642–666, Mar.
475–513, 2003. 2007.
58. A. A.Freitas. Data Mining and Knowledge Discovery 76. J. Davis, J. Hoffert, and E. Vanlandingham. A taxonomy
with Evolutionary Algorithms, Berlin, Springer. 2009. of artificial intelligence approaches for adaptive
59. G. Luque and E. Alba. Parallel genetic algorithms: distributed real-time embedded systems, IEEE Int.
Theory and realworld applications, Springer. 2011. Conf. Electro Inf. Technol., 2016, Vol. 2016-Augus, no.
60. F. Yang. Solving Traveling Salesman Problem Using 1, pp. 233–238.
Parallel Genetic Algorithm and Simulated Annealing, 77. M. Lee, Y. Xiong, G. Yu, and G. Y. Li. Deep neural
Aug. 2010. networks for linear sum assignment problems, IEEE
61. H. RasitEr and N. Erdogan. Parallel Genetic Algorithm Wirel. Commun. Lett. Vol. 7, no. 6, pp. 962–965, 2018.
to Solve Traveling Salesman Problem on MapReduce 78. L. A. Zadeh. Fuzzy sets, Inf. Control, Vol. 8, no. 3, pp.
Framework using Hadoop Cluster, JSCSE, Vol. 3, no. 338–353, 1965.
3, pp. 380–386, 2013. 79. H. Jafari, S. Bateni, P. Daneshvar, S. Bateni, and H.
62. S. John Walker. Big Data: A Revolution That Will Mahdioun. Fuzzy Mathematical Modeling Approach
Transform How We Live, Work, and Think, Int. J. for the Nurse Scheduling Problem: A Case Study, Int.
Advert., Vol. 33, no. 1, pp. 181–183, 2014. J. Fuzzy Syst., Vol. 18, no. 2, pp. 320–332, 2016.
63. Manyika, J., Chui, M., Brown, B., Bughin, J., Dobbs, R., E. Ferreyra, H. Hagras, A. Mohamed, and G. Owusu. A
Roxburgh, C., & Byers, A. H.Big Data: The Next type-2 fuzzy logic system for engineers estimation in
Frontier for Innovation, Comptetition, and the workforce allocation domain, IEEE Int. Conf.
Productivity, McKinsey Global Institute, 2011. Fuzzy Syst., 2017, pp. 0–5.
64. E. Dumbill. Making Sense of Big Data, Big Data, Vol.
1, no. 1, pp. 1–2, 2013.
65. T. A. Mohammed, O. Bayat, O. N. Uçan, and S. Alhayali.
Hybrid Efficient Genetic Algorithm for Big Data
Feature Selection Problems, Found. Sci., no.
0123456789, 2019.
66. M. Wa and H. San. Genetic Algorithm and its
application to Big Data Analysis, Int. J. Sci. Eng. Res.,
Vol. 5, no. 1, pp. 1991–1996, 2014.
67. J. S. Tan, W. He, and Y. Qing. Application of genetic
algorithm in data mining, Proc. 1st Int. Work. Educ.
Technol. Comput. Sci. ETCS 2009, 2009, Vol. 2, pp.
353–356.
68. S. Zheng. Solving Vehicle Routing Problem: A Big
Data Analytic Approach, IEEE Access, Vol. 7, pp.
169565–169570, Jan. 2019.
69. S. Razavi, O. M. Ebadati E., and H. Kaur. An Efficient
Grouping Genetic Algorithm for Data Clustering and
Big Data Analysis, In Computational Intelligence for
7420

View publication stats