Concepts Of Genetic Algorithms Research Articles

OPTIMIZING SIMILARITY THRESHOLD FOR ABSTRACT SIMILARITY METRIC IN SPEECH DIARIZATION SYSTEMS: A MATHEMATICAL FORMULATION

Speaker diarization is a critical task in speech processing that aims to identify "who spoke when?" in an audio or video recording that contains unknown amounts of speech from unknown speakers and unknown number of speakers. Diarization has numerous applications in speech recognition, speaker identification, and automatic captioning. Supervised and unsupervised algorithms are used to address speaker diarization problems, but providing exhaustive labeling for the training dataset can become costly in supervised learning, while accuracy can be compromised when using unsupervised approaches. This paper presents a novel approach to speaker diarization, which defines loosely labeled data and employs x-vector embedding and a formalized approach for threshold searching with a given abstract similarity metric to cluster temporal segments into unique user segments. The proposed algorithm uses concepts of graph theory, matrix algebra, and genetic algorithm to formulate and solve the optimization problem. Additionally, the algorithm is applied to English, Spanish, and Chinese audios, and the performance is evaluated using wellknown similarity metrics. The results demonstrate that the robustness of the proposed approach. The findings of this research have significant implications for speech processing, speaker identification including those with tonal differences. The proposed method offers a practical and efficient solution for speaker diarization in real-world scenarios where there are labeling time and cost constraints

Solving large scale industrial production scheduling problems with complex constraints: an overview of the state-of-the-art

Production scheduling is challenging and the body of literature addressing various variants of the problem is large. It can roughly be divided into two streams: The first stream addresses and generalizes established scheduling problems, being general in the sense that they are not only applicable in a particular industry. The second stream works on less generic scheduling approaches for real industry cases by enriching standard models with all the required realistic aspects, such as process overlapping or sequence dependent setup times. Furthermore, different approaches have different limitations in terms of the problem size that they can tackle. The rise of Industry 4.0 has lead to a significant increase in data collection activities and the gathered information is used to build larger and more complex models. Industrial use cases may consist of several thousand operations on a large variety of machines, while classical benchmark instances tend to range up to only a few hundred of operations. It is therefore necessary to identify and highlight approaches, that can meet the challenges of scheduling in the era of Industry 4.0 and are suitable to tackle large scale problems.In this work, we conduct a structured literature review on scheduling problems incorporating several real world aspects among a broad range of use cases. Based on the identified publications we find that advanced solution approaches for large scale scheduling problems usually belong to one out of three categories, namely metaheuristic methods, constraint programming and machine learning. Our review shows that comparably few contributions tackling (very) large scale problems exist, emphasizing the need for additional research in this field. We identify promising approaches for further research, such as powerful metaheuristics combining concepts of tabu search and genetic algorithms. We further discuss the possibility to enhance solution methods by integrating constraint programming concepts and investigating problem decomposition.

A diversity-based genetic algorithm for scenario generation

Tackling uncertainty is becoming increasingly relevant for decision-support across fields due to its critical impact on real-world problems. Uncertainty is often modelled using scenarios, which are combinations of possible outcomes of the uncertain parameters in a problem. Alongside expected value methods, decisions under uncertainty may also be tackled using methods that do not rely on probability distributions and model different decision-maker risk profiles. Scenarios are at the core of these approaches. Therefore, we propose a scenario generation methodology that seizes the structure and concepts of genetic algorithms. This methodology aims to obtain a diverse set of scenarios, evolving a scenario population with a diversity goal. Diversity is here expressed as the difference in the impact that scenarios have on the value of potential solutions to the problem. Moreover, this method does not require a priori knowledge of probability distributions or statistical moments of uncertain parameters, as it is based on their range. We adapt the available code for Biased-Random Key Genetic Algorithms to apply the methodology to a packing problem under demand uncertainty as a proof of concept, also extending its use to a multi-objective setting. We make available these code adaptations to allow the straightforward application of this scenario generation method to other problems. With this, the decision-maker obtains scenarios with a distinct impact on potential solutions, enabling the use of different criteria based on their profile and preferences.

Fuzzy Adaptive PID Control of a Shell and Tube Heat Exchanger Output Temperature

This paper presents the design and the performance comparison of a classic PID controller, a feedforward PID controller, and a fuzzy PID controller. These controllers have been implemented to control the outlet temperature of a shell and tube heat exchanger. The controller with feedforward is proposed to provide assistance in the closed-loop control, in order to reduce the disturbances that affect the system. This method is compared with the fuzzy one, which is based on the concepts of artificial intelligence, genetic algorithm, neural networks, and fuzzy logic. For this controller, Gaussian membership functions have been established, a universe of discourse ranging from -6 to 6 for the error and its derivative, and a universe of discourse ranging from -0.5 to 0.5 for the control variables. From the results of this study, it has been concluded that the tracking error in the output temperature under the variation of the set-point decreases in a faster way in the PID with feedforward respect to the classical PID. However, the fuzzy adaptive PID control is based on the optimized parameters.

GENERAL ARTIFICIAL INTELLIGENCE MODEL FOR DEVELOPING ADAPTIVE NON PLAYABLE CHARACTERS IN COMPUTER GAMES

In 2005, through inauguration of the annual AAAI General Game Playing Competition the artificial intelligence (AI) community was introduced to AI in computer games. Since then, the efforts have been made to make AI systems understand the rules of arbitrary computer games without any human help. This paper will introduce a new model based on neural networks and genetic algorithm concepts which has the potential to replace behavior tree model currently used by game developers for making AI systems. This model introduces the possibility for AI systems to be trained in the game instead of creating complex nested if-else tree to determine the next move. This adaptive learning approach for AI systems could potentially lead to development of more powerful and engaging AI systems known as Non playable characters in gaming industry. The model demonstrated in this paper is developed using Python 3 language and the game environment in which it is tested are developed in Python 3 environment as well.

An Effective Technique on Clustering in Perspective of Huge Data Set

Analysis of data plays a crucial job considering the different phenomenon. It explores the prior knowledge, consisting of development across the extensively different communities. Cluster technique is the collecting of data object placed into groups. Therefore objects are the same nature or similar place within a cluster different nature (i.e. dissimilar) put in other cluster. Differences and likeness are refereed on the attribute values say that the objects involved in measuring distance. We have reviewed a few clustering techniques for data sets in data mining of various field of computer science and engineering, statistical, machine learning and a novel attracting field of demanding efforts. Several closely related concepts of neural network, fuzzy and genetic algorithm are also discussed. In this research paper to also discussed the facebook data set to mining the attributes from the cluster set to changing the mean square error 0-10 -6 and also discuss measuring the performance.

Evolutionary-based image encryption using RNA codons truth table

Symmetric image cryptography is a mechanism in which image pixels are encrypted into some meaningless format called cipher image. Only authorized users have access to the secret code for symmetric key encryption. In this paper a new symmetric image encryption method has been proposed using the concepts of ribonucleic acid (RNA) sequence and genetic algorithm (GA), called RNA-GA. The proposed method starts by generating specified number of initial cipher images using logistic map function. Then, the initial cipher images are converted to the corresponding one-dimensional binary sequences and relevant codons array using codons truth table. The codons array are then updated using encryption key and encryption RNA tables to form the initial population of genetic algorithm. Next, genetic algorithm optimizes the population using selection, crossover, and mutation operators. The results approve high resistance of the proposed method against well-known attacks and entropy of 7.9987 regarding standard image of size 256×256 after 30 repetitions.

Simulation of intelligent target hitting in obstructed path using physical body animation and genetic algorithm

Nowadays there are many real-time applications such as robotic motion, driver-less vehicle, intelligent target shooter(bullets and missiles), traffic routing in which human intervention is avoided. This paper proposes an exciting and generalized approach for intelligent target hitting in an obstructed path using physical body animation and genetic algorithm. This approach uses the concepts of the genetic algorithm to train the object for finding the right path to target and concepts of physical body animation to provide the motion and to react as per the collision with obstacles. Physical body animation provides a very natural feel of a real-time environment as we deal with all the external natural forces such as gravity, wind resistance the object and so on. Proposed approach deals not only with the static target but also deals with the dynamic target during the simulation.

Efficient heuristics for the minimum labeling global cut problem

Let G=(V,E,L) be an edge-labeled graph. Let V be the set of vertices of G, E the set of edges, L the set of labels (colors) such that each edge e∈E has an associated label L(e). The goal of the minimum labeling global cut problem (MLGCP) is to find a subset L′⊆L of labels such that G′=(V,E′,L\\L′) is not connected and |L′| is minimized. In this work, we generate random instances for the MLGCP to perform empirical tests. Also propose a set of heuristics using concepts of Genetic Algorithm and metaheuristic VNS, including some of their procedures, like two local search moves, and an auxiliary data structure to speed up the local search. Computational experiments show that the metaheuristic VNS outperforms other methods with respect to solution quality.

Using cloud effectively in concept based text mining using grey wolf self organizing feature map

Cloud computing is considered to be an integral aspect in all business and this is expected to change the information technology (IT) landscape. This has been based on the model that delivers services on the internet using the pay-as-you go model that has several advantages like the no up-front cost, a lower IT staff, and a lower operation cost. A technology that is made use of for retrieval of data from huge database is known as text mining. This is used by cloud for efficiently retrieving data from the data centres of cloud. In providing navigation as well as mechanisms for browsing intuitively, text document clustering has an important role. This is done by organizing huge amounts of information into smaller number of clusters. Bag of words (BoW) is a representation that is used for the clustering of these methods but in many case it is not satisfactory as relations that exist between terms that don’t co-occur are ignored. To handle this problem a document level and sentence level integration of the concepts is made. This increases the space of the feature vector and also brings down the clustering algorithm’s efficiency. In order to overcome this a self-organizing feature map (SOFM) based algorithm makes use of the concepts of genetic algorithm (GA) along with grey wolf optimization (GWO) which are considered popular in the SOFM. The goal of the SOFM-GA is to find an optimal topology of network (the number of neurons and their array dimension) along with an optimal training parameter like the scheduling of learning rate and the annealing of neighborhood width. The SOFM-GWO and the GWO-based approach to the formation of SOFM are compared with the SOM standard relating to quality and the weights and map generated. The results of the experiment show that this method achieved better results.

Seleção de pessoas por meio de algoritmos genéticos

This work has as object the process of recruitment and selection, considering the effort to find professionals for the respondent job, helping people management in their decision making. This study aimed to use the concepts of genetic algorithms and clustering through the development of codes to propose the use of artificial intelligence in recruitment and selection of people. The use of algorithms and clustering proved to be a possible tool to be used in people selection process because with its use was possible to visualize three (3) candidates clusters to the desired profile: unfit, promising and able. Since the fit would be the best candidates for the job offered, since they have the best profile and knowledge.

Decomposition based Multi Objective Evolutionary Algorithms for Design of Large-Scale Water Distribution Networks

In last two decades, multiobjective evolutionary algorithms (MOEAs) have shown their merit for solving different optimization problems within the context of water resources and environmental engineering. MOEAs mainly use the concept of Pareto dominance for obtaining the trade-off solutions considering different criteria. A new alternative method for solving multiobjective problems is multiobjective evolutionary algorithm based on decomposition (MOEA/D) which uses scalarizing the objective functions. In this paper, decomposition strategies are developed for the large-scale water distribution network (WDN) design problems by integrating the concepts of harmony search (HS) and genetic algorithm (GA) within the MOEA/D framework. The proposed algorithms are then compared with two well-known non-dominance based MOEAs: NSGA2 and SPEA2 across four different WDN design problems. Experimental results show that MOEA/D outperform the Pareto dominance methods in terms of both non-domination and diversity criteria. MOEA/D-HS in particular could provide very high quality solutions with a uniform distribution along the Pareto front preserving the diversity and dominating the solutions of the other algorithms. It suggests that decomposition based multiobjective evolutionary algorithms are very promising in dealing with complicated large-scale WDN design problems.

Provide a Way to Reduction of Energy Consumption in Wireless Sensor Networks using the concepts of Fuzzy Logic and Genetic Algorithm

Wireless sensor networks are a collection of small sensor nodes that can monitor and sense of their surroundings and sending data to a main station. The limited energy of nodes is a major challenge of sensor networks that affect the survival of the network. Thus, as yet is presented several methods to optimization of energy consumption and increasing the lifetime of a sensor network. In this paper, using fuzzy system design and system optimization by genetic algorithm is presented approach to select the best cluster head in sensor networks. Using random data set has been addressed to evaluate of fuzzy-genetic system presented in this paper and finally, MSE rate or mean error of sending the messages using proposed fuzzy system in comparison with LEACH method is calculated in select the cluster head. The results of evaluations is representative of a reduction the MSE metric in proposed method in comparison with LEACH method for select the cluster head. Reduce of MSE directly is effective on energy consumption and lifetime of wireless sensor network and can cause the reduction of energy consumption and increase the lifetime of the networks.

Mobility prediction in mobile ad hoc networks using a lightweight genetic algorithm

A mobile ad hoc network is a collection of wireless mobile nodes creating a network without using any existing infrastructure. Much research has been carried out to find out an optimal routing protocol for the successful transmission of data in this network. The main hindrance is the mobility of the network. If the mobility pattern of the network can be predicted, it will help in improving the QoS of the network. This paper discusses a novel approach to mobility prediction using movement history and existing concepts of genetic algorithms, to improve the MANET routing algorithms. The proposed lightweight genetic algorithm performs outlier removal on the basis of heuristics and parent selection using the weighted roulette wheel algorithm. After performing the genetic operations a node to node adjacency matrix is obtained from which the predicted direction of each node is calculated using force directed graphs and vector calculations. The technique proposes a new approach to mobility prediction which does not depend on probabilistic methods and which is completely based on genetic algorithms.

Scenario-Based Multi-Objective Genetic Algorithm Optimization of Closed Ball-Milling Circuit of Esfordi Phosphate Plant

The main goal of this article is to demonstrate an approach based on integration of process simulation and Multi-Objective Genetic Algorithm (MOGA) concepts to solve a real grinding circuit optimization problem by finding the best operating condition under which process objectives can be achieved. Esfordi phosphate plant is located near city of Bafgh at Yazd province of Iran and produces 5 Mt of phosphate annually. The fine particles (nearly −20 µm) in hydrocyclone underflow which contain a high grade of iron are subjected to over grinding. In addition to electrical energy loss, this causes problems in the downstream process, i.e., flotation stage. The main goals of this study were to solve this problem by adjusting operating condition so that (a) hydrocyclone overflow particle size can be increased from 94.2 µm to 100 µm and (b) increase hydrocyclone underflow particle size from 205 to 500 µm. The second process objective will decrease fine particles in hydrocyclone underflow stream. First, plant sampling campaigns were carried out to calibrate ball mill and hydrocyclone models to be used for performing simulation trials. Then, full circuit simulations were done and optimized by MOGA search process to find the best operating condition that produces hydrocyclone overflow and underflow streams with predefined particle sizes simultaneously. The results indicate that there are various solutions that can be recommended for plant testing and performance improvements. The results of plant implementation of one solution for scenario No. 4 showed improved circuit performance and also validated simulator predictions.

A GENETIC ALGORITHM FOR THE ONE-DIMENSIONAL CUTTING STOCK PROBLEM WITH SETUPS

This paper investigates the one-dimensional cutting stock problem considering two conflicting objective functions: minimization of both the number of objects and the number of different cutting patterns used. A new heuristic method based on the concepts of genetic algorithms is proposed to solve the problem. This heuristic is empirically analyzed by solving randomly generated instances and also practical instances from a chemical-fiber company. The computational results show that the method is efficient and obtains positive results when compared to other methods from the literature.

Neural Networks using Genetic Algorithms

Combining neural network with evolutionary algorithms leads to evolutionary artificial neural network. Evolutionary algorithms like GA to train neural nets choose their structure or design related aspects like the functions of their neurons. Along basic concepts of neural networks and genetic algorithm this paper includes a flexible method for solving travelling salesman problem using genetic algorithm. This offers a solution which includes a genetic algorithm implementation in order to give a maximal approximation of the problem with the reduction of cost.

Multi-Objective Particle Swarm Optimization for Control Laws Design

In order to overcome the difficult of large amount of calculation and to satisfy multiple design indicators in the design of control laws, an improved multi-objective particle swarm optimization (PSO) algorithm was used to design control laws of aircraft. Firstly, the hybrid concepts of genetic algorithm were introduced to particle swarm optimization (PSO) algorithm to improve the algorithm. Then based on aircraft flying quality the reference models were built, and then the tracking error, settling time and overshoot were used as the optimization goal of the control laws design. Based on this multi-objective optimize problem the attitude hold control laws were designed. The simulation results show the effectiveness of the algorithm.

An adaptive fuzzy-genetic algorithm approach for building detection using high-resolution satellite images

We propose a new approach for building detection using high-resolution satellite imagery based on an adaptive fuzzy-genetic algorithm. This novel approach improves object detection accuracy by reducing the premature convergence problem encountered when using genetic algorithms. We integrate the fundamental image processing operators with genetic algorithm concepts such as population, chromosome, gene, crossover and mutation. To initiate the approach, training samples are selected that represent the specified two feature classes, in this case “building” and “non-building”. The image processing operations are carried out on a chromosome-by-chromosome basis to reveal the attribute planes. These planes are then reduced to one hyperplane that is optimal for discriminating between the specified feature classes. For each chromosome, the fitness values are calculated through the analysis of detection and mis-detection rates. This analysis is followed by genetic algorithm operations such as selection, crossover and mutation. At the end of each generation cycle, the adaptive-fuzzy module determines the new (adjusted) probabilities of crossover and mutation. This evolutionary process repeats until a specified number of generations has been reached. To enhance the detected building patches, morphological image processing operations are applied. The approach was tested on ten different test scenes of the Batikent district of the city of Ankara, Turkey using 1m resolution pan-sharpened IKONOS imagery. The kappa statistics computed for the proposed adaptive fuzzy-genetic algorithm approach were between 0.55 and 0.88. The extraction performance of the algorithm was better for urban and suburban buildings than for buildings in rural test scenes.

A hybrid imperialist competitive algorithm for single-machine scheduling problem with linear earliness and quadratic tardiness penalties

In this paper, we study single-machine scheduling problem when each job is considered with linear earliness and quadratic tardiness penalties with no machine idle time. Here the objective is to find the best sequence of jobs in the reasonable time. This model was studied in several researches, and some algorithms were proposed to solve it such as genetic algorithm. As the size of problem increased, such algorithms were not effective and efficient. Hence, we proposed the hybrid imperialist competitive algorithm. The proposed algorithm is based upon the imperialist competitive algorithm and genetic algorithm concepts. This algorithm was tested in problems with different sizes. The results denoted that the hybrid algorithm can solve different size of problem in reasonable time. This procedure showed its efficiency in medium- and large-sized problems as compared with other available methods.