Pub Date : 2017-07-07DOI: 10.1109/CEC.2017.7969435
Caihong Mu, Huiwen Cheng, Wei Feng, Yi Liu, R. Qu
Recommender system (RS) plays an important role in helping users find the information they are interested in and providing accurate personality recommendation. It has been found that among all the users, there are some user groups called “core users” or “information core” whose historical behavior data are more reliable, objective and positive for making recommendations. Finding the information core is of great interests to greatly increase the speed of online recommendation. There is no general method to identify core users in the existing literatures. In this paper, a general method of finding information core is proposed by modelling this problem as a combinatorial optimization problem. A novel Evolutionary Algorithm with Elite Population (EA-EP) is presented to search for the information core, where an elite population with a new crossover mechanism named as ordered crossover is used to accelerate the evolution. Experiments are conducted on Movielens (100k) to validate the effectiveness of our proposed algorithm. Results show that EA-EP is able to effectively identify core users and leads to better recommendation accuracy compared to several existing greedy methods and the conventional collaborative filter (CF). In addition, EA-EP is shown to significantly reduce the time of online recommendation.
{"title":"Information core optimization using Evolutionary Algorithm with Elite Population in recommender systems","authors":"Caihong Mu, Huiwen Cheng, Wei Feng, Yi Liu, R. Qu","doi":"10.1109/CEC.2017.7969435","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969435","url":null,"abstract":"Recommender system (RS) plays an important role in helping users find the information they are interested in and providing accurate personality recommendation. It has been found that among all the users, there are some user groups called “core users” or “information core” whose historical behavior data are more reliable, objective and positive for making recommendations. Finding the information core is of great interests to greatly increase the speed of online recommendation. There is no general method to identify core users in the existing literatures. In this paper, a general method of finding information core is proposed by modelling this problem as a combinatorial optimization problem. A novel Evolutionary Algorithm with Elite Population (EA-EP) is presented to search for the information core, where an elite population with a new crossover mechanism named as ordered crossover is used to accelerate the evolution. Experiments are conducted on Movielens (100k) to validate the effectiveness of our proposed algorithm. Results show that EA-EP is able to effectively identify core users and leads to better recommendation accuracy compared to several existing greedy methods and the conventional collaborative filter (CF). In addition, EA-EP is shown to significantly reduce the time of online recommendation.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"105 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134474246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-07-07DOI: 10.1109/CEC.2017.7969541
Leticia Hernando, F. Daolio, Nadarajen Veerapen, G. Ochoa
Local Optima Networks were proposed to understand the structure of combinatorial landscapes at a coarse-grained level. We consider a compressed variant of such networks with features that are meaningful for the study of search difficulty in the context of local search. In particular, we investigate different landscapes of the Permutation Flowshop Scheduling Problem. The insert and 2-exchange neighbourhoods are considered, and two different objective functions are taken into account: the makespan and the total flow time. The aim is to analyse the network features in order to find differences between the landscape structures, giving insights about which features impact algorithm performance. We evaluate the correlation between landscape properties and the performance of an Iterated Local Search algorithm. Visualisation of the network structure is also given, where evident differences between the makespan and total flow time are observed.
{"title":"Local Optima Networks of the Permutation Flowshop Scheduling Problem: Makespan vs. total flow time","authors":"Leticia Hernando, F. Daolio, Nadarajen Veerapen, G. Ochoa","doi":"10.1109/CEC.2017.7969541","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969541","url":null,"abstract":"Local Optima Networks were proposed to understand the structure of combinatorial landscapes at a coarse-grained level. We consider a compressed variant of such networks with features that are meaningful for the study of search difficulty in the context of local search. In particular, we investigate different landscapes of the Permutation Flowshop Scheduling Problem. The insert and 2-exchange neighbourhoods are considered, and two different objective functions are taken into account: the makespan and the total flow time. The aim is to analyse the network features in order to find differences between the landscape structures, giving insights about which features impact algorithm performance. We evaluate the correlation between landscape properties and the performance of an Iterated Local Search algorithm. Visualisation of the network structure is also given, where evident differences between the makespan and total flow time are observed.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132438459","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-07-07DOI: 10.1109/CEC.2017.7969522
Junfeng Chen, M. Omidvar, M. Azad, Xin Yao
A proportional-integral-derivative (PID) controller is a control loop feedback mechanism widely employed in industrial control systems. The parameters tuning is a sticking point, having a great effect on the control performance of a PID system. There is no perfect rule for designing controllers, and finding an initial good guess for the parameters of a well-performing controller is difficult. In this paper, we develop a knowledge-based particle swarm optimization by incorporating the dynamic response information of PID into the optimizer. Prior knowledge not only empowers the particle swarm optimization algorithm to quickly identify the promising regions, but also helps the proposed algorithm to increase the solution precision in the limited running time. To benchmark the performance of the proposed algorithm, an electric pump drive and an automatic voltage regulator system are selected from industrial applications. The simulation results indicate that the proposed algorithm with a newly proposed performance index has a significant performance on both test cases and outperforms other algorithms in terms of overshoot, steady state error, and settling time.
{"title":"Knowledge-based particle swarm optimization for PID controller tuning","authors":"Junfeng Chen, M. Omidvar, M. Azad, Xin Yao","doi":"10.1109/CEC.2017.7969522","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969522","url":null,"abstract":"A proportional-integral-derivative (PID) controller is a control loop feedback mechanism widely employed in industrial control systems. The parameters tuning is a sticking point, having a great effect on the control performance of a PID system. There is no perfect rule for designing controllers, and finding an initial good guess for the parameters of a well-performing controller is difficult. In this paper, we develop a knowledge-based particle swarm optimization by incorporating the dynamic response information of PID into the optimizer. Prior knowledge not only empowers the particle swarm optimization algorithm to quickly identify the promising regions, but also helps the proposed algorithm to increase the solution precision in the limited running time. To benchmark the performance of the proposed algorithm, an electric pump drive and an automatic voltage regulator system are selected from industrial applications. The simulation results indicate that the proposed algorithm with a newly proposed performance index has a significant performance on both test cases and outperforms other algorithms in terms of overshoot, steady state error, and settling time.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126546255","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-07-05DOI: 10.1109/CEC.2017.7969589
Javiera Loyola Vitali, M. Riff, Elizabeth Montero
The Bus Evacuation Problem is a route planning problem, in the context of an evacuation in an emergency situation. Considering that public transport is available to support the evacuation, the objective of the problem is to determine the best route for each vehicle, to move all the people from a risk zone to open shelters located in safe zones, such that the evacuation time is minimized. In this work we present a method based on the Greedy Randomized Adaptive Search Procedure metaheuristic to solve the problem, in order to apply the solution to a real-world scenario based on a recent wildfire on Valparaíso, Chile. In computational experiments we demonstrate that our approach is effective to solve real-world size problems, and able to outperform a commercial MIP solver.
{"title":"Bus Routing for emergency evacuations: The case of the Great Fire of Valparaiso","authors":"Javiera Loyola Vitali, M. Riff, Elizabeth Montero","doi":"10.1109/CEC.2017.7969589","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969589","url":null,"abstract":"The Bus Evacuation Problem is a route planning problem, in the context of an evacuation in an emergency situation. Considering that public transport is available to support the evacuation, the objective of the problem is to determine the best route for each vehicle, to move all the people from a risk zone to open shelters located in safe zones, such that the evacuation time is minimized. In this work we present a method based on the Greedy Randomized Adaptive Search Procedure metaheuristic to solve the problem, in order to apply the solution to a real-world scenario based on a recent wildfire on Valparaíso, Chile. In computational experiments we demonstrate that our approach is effective to solve real-world size problems, and able to outperform a commercial MIP solver.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129322122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-07-05DOI: 10.1109/CEC.2017.7969483
C. Ortega, M. Vasile
This paper presents a non-nested algorithm for the solution of multi-objective min-max problems (MOMMP) in worst-case optimization. The algorithm has been devised for evidence-based robust optimization, where the lack of a defined probabilistic behaviour of the uncertain parameters makes it impossible to apply sample-based techniques and forces the designer to identify the worst case over the subdomains of the uncertainty space. In evidence theory, the robustness of the solutions is measured in terms of the Belief in the realization of the value of the design budgets, which acts as a lower bound to the unknown cumulative distribution function of the budget. Thus a means of finding robust solutions in preliminary design consists on applying the minimax model, where the worst-case budget over the uncertainty space is optimized over the control space. The paper proposes a novel heuristic to solve MOMMP and demonstrates its capability to approximate the worst-case Pareto front at a very reduced cost with respect to approaches based on nested optimization
{"title":"New heuristics for multi-objective worst-case optimization in evidence-based robust design","authors":"C. Ortega, M. Vasile","doi":"10.1109/CEC.2017.7969483","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969483","url":null,"abstract":"This paper presents a non-nested algorithm for the solution of multi-objective min-max problems (MOMMP) in worst-case optimization. The algorithm has been devised for evidence-based robust optimization, where the lack of a defined probabilistic behaviour of the uncertain parameters makes it impossible to apply sample-based techniques and forces the designer to identify the worst case over the subdomains of the uncertainty space. In evidence theory, the robustness of the solutions is measured in terms of the Belief in the realization of the value of the design budgets, which acts as a lower bound to the unknown cumulative distribution function of the budget. Thus a means of finding robust solutions in preliminary design consists on applying the minimax model, where the worst-case budget over the uncertainty space is optimized over the control space. The paper proposes a novel heuristic to solve MOMMP and demonstrates its capability to approximate the worst-case Pareto front at a very reduced cost with respect to approaches based on nested optimization","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115651389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-07DOI: 10.1109/CEC.2017.7969518
Julie Greensmith, Michael B. Gale
The Dendritic Cell Algorithm (DCA) has been described in a number of different ways, sometimes resulting in incorrect implementations. We believe this is due to previous, imprecise attempts to describe the algorithm. The main contribution of this paper is to remove this imprecision through a new approach inspired by purely functional programming. We use new specification to implement the deterministic DCA in Haskell - the hDCA. This functional variant will also serve to introduce the DCA to a new audience within computer science. We hope that our functional specification will help improve the quality of future DCA related research and to help others understand further its algorithmic properties.
{"title":"The Functional Dendritic Cell Algorithm: A formal specification with Haskell","authors":"Julie Greensmith, Michael B. Gale","doi":"10.1109/CEC.2017.7969518","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969518","url":null,"abstract":"The Dendritic Cell Algorithm (DCA) has been described in a number of different ways, sometimes resulting in incorrect implementations. We believe this is due to previous, imprecise attempts to describe the algorithm. The main contribution of this paper is to remove this imprecision through a new approach inspired by purely functional programming. We use new specification to implement the deterministic DCA in Haskell - the hDCA. This functional variant will also serve to introduce the DCA to a new audience within computer science. We hope that our functional specification will help improve the quality of future DCA related research and to help others understand further its algorithmic properties.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"236 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117042357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-07DOI: 10.1109/CEC.2017.7969394
O. Strub, N. Trautmann
We consider the problem of replicating the returns of a financial index as accurately as possible by selecting a subset of the assets that constitute the index and determining the portfolio weight of each selected asset subject to various constraints that are relevant in practice, including the UCITS III (Undertakings for Collective Investments in Transferable Securities) 5/10/40 concentration rule. For this problem, we present a genetic algorithm, in which the individuals correspond to subsets of the index constituents. The fitness of the individuals is determined by applying mixed-integer quadratic programming. Two main features of the presented genetic algorithm are novel. First, we use a representation of subsets which is the first that exhibits all of the four desirable properties feasibility, efficiency, locality, and heritability. The representation also allows to incorporate problem-specific knowledge in a very simple way. Second, to reduce the CPU time for the fitness evaluations, we first estimate the fitness of the individuals in an efficient way and then evaluate the fitness of promising individuals only. The results of a computational experiment based on real-world data demonstrate that in particular for large instances, the presented genetic algorithm devises very good solutions in short CPU time.
{"title":"A genetic algorithm for the UCITS-constrained index-tracking problem","authors":"O. Strub, N. Trautmann","doi":"10.1109/CEC.2017.7969394","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969394","url":null,"abstract":"We consider the problem of replicating the returns of a financial index as accurately as possible by selecting a subset of the assets that constitute the index and determining the portfolio weight of each selected asset subject to various constraints that are relevant in practice, including the UCITS III (Undertakings for Collective Investments in Transferable Securities) 5/10/40 concentration rule. For this problem, we present a genetic algorithm, in which the individuals correspond to subsets of the index constituents. The fitness of the individuals is determined by applying mixed-integer quadratic programming. Two main features of the presented genetic algorithm are novel. First, we use a representation of subsets which is the first that exhibits all of the four desirable properties feasibility, efficiency, locality, and heritability. The representation also allows to incorporate problem-specific knowledge in a very simple way. Second, to reduce the CPU time for the fitness evaluations, we first estimate the fitness of the individuals in an efficient way and then evaluate the fitness of promising individuals only. The results of a computational experiment based on real-world data demonstrate that in particular for large instances, the presented genetic algorithm devises very good solutions in short CPU time.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134429971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-06DOI: 10.1109/CEC.2017.7969477
R. Tyasnurita, E. Özcan, R. John
A selection hyper-heuristic is a search method that controls a prefixed set of low-level heuristics for solving a given computationally difficult problem. This study investigates a learning-via demonstrations approach generating a selection hyper-heuristic for Open Vehicle Routing Problem (OVRP). As a chosen ‘expert’ hyper-heuristic is run on a small set of training problem instances, data is collected to learn from the expert regarding how to decide which low-level heuristic to select and apply to the solution in hand during the search process. In this study, a Time Delay Neural Network (TDNN) is used to extract hidden patterns within the collected data in the form of a classifier, i.e an ‘apprentice’ hyper-heuristic, which is then used to solve the ‘unseen’ problem instances. Firstly, the parameters of TDNN are tuned using Taguchi orthogonal array as a design of experiments method. Then the influence of extending and enriching the information collected from the expert and fed into TDNN is explored on the behaviour of the generated apprentice hyper-heuristic. The empirical results show that the use of distance between solutions as an additional information collected from the expert generates an apprentice which outperforms the expert algorithm on a benchmark of OVRP instances.
{"title":"Learning heuristic selection using a Time Delay Neural Network for Open Vehicle Routing","authors":"R. Tyasnurita, E. Özcan, R. John","doi":"10.1109/CEC.2017.7969477","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969477","url":null,"abstract":"A selection hyper-heuristic is a search method that controls a prefixed set of low-level heuristics for solving a given computationally difficult problem. This study investigates a learning-via demonstrations approach generating a selection hyper-heuristic for Open Vehicle Routing Problem (OVRP). As a chosen ‘expert’ hyper-heuristic is run on a small set of training problem instances, data is collected to learn from the expert regarding how to decide which low-level heuristic to select and apply to the solution in hand during the search process. In this study, a Time Delay Neural Network (TDNN) is used to extract hidden patterns within the collected data in the form of a classifier, i.e an ‘apprentice’ hyper-heuristic, which is then used to solve the ‘unseen’ problem instances. Firstly, the parameters of TDNN are tuned using Taguchi orthogonal array as a design of experiments method. Then the influence of extending and enriching the information collected from the expert and fed into TDNN is explored on the behaviour of the generated apprentice hyper-heuristic. The empirical results show that the use of distance between solutions as an additional information collected from the expert generates an apprentice which outperforms the expert algorithm on a benchmark of OVRP instances.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132959029","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-05DOI: 10.1109/CEC.2017.7969532
Rui Wu, Jose T. Painumkal, J. Volk, Siming Liu, S. Louis, S. Tyler, S. Dascalu, F. Harris
We attack the problem of predicting nitrate concentrations in a stream by using a genetic algorithm to minimize the difference between observed and predicted concentrations on hydrologic nitrate concentration model based on a US Geological Survey collected data set. Nitrate plays a significant role in maintaining ecological balance in aquatic ecosystems and any advances in nitrate prediction accuracy will improve our understanding of the non-linear interplay between the factors that impact aquatic ecosystem health. We compare the genetic algorithm tuned model against the LOADEST estimation tool in current use by hydrologists, and against a random forest, generalized linear regression, decision tree, and gradient booted tree and show that the genetic algorithm does statistically significantly better. These results indicate that genetic algorithms are a viable approach to tuning such non-linear, hydrologic models.
{"title":"Parameter estimation of nonlinear nitrate prediction model using genetic algorithm","authors":"Rui Wu, Jose T. Painumkal, J. Volk, Siming Liu, S. Louis, S. Tyler, S. Dascalu, F. Harris","doi":"10.1109/CEC.2017.7969532","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969532","url":null,"abstract":"We attack the problem of predicting nitrate concentrations in a stream by using a genetic algorithm to minimize the difference between observed and predicted concentrations on hydrologic nitrate concentration model based on a US Geological Survey collected data set. Nitrate plays a significant role in maintaining ecological balance in aquatic ecosystems and any advances in nitrate prediction accuracy will improve our understanding of the non-linear interplay between the factors that impact aquatic ecosystem health. We compare the genetic algorithm tuned model against the LOADEST estimation tool in current use by hydrologists, and against a random forest, generalized linear regression, decision tree, and gradient booted tree and show that the genetic algorithm does statistically significantly better. These results indicate that genetic algorithms are a viable approach to tuning such non-linear, hydrologic models.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125839659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-05DOI: 10.1109/CEC.2017.7969501
Alejandro Martín, Félix Fuentes-Hurtado, V. Naranjo, David Camacho
Deep Neural Networks (DNN) have become a powerful, widely used, and successful mechanism to solve problems of different nature and varied complexity. Their ability to build models adapted to complex non-linear problems, have made them a technique widely applied and studied. One of the fields where this technique is currently being applied is in the malware classification problem. The malware classification problem has an increasing complexity, due to the growing number of features needed to represent the behaviour of the application as exhaustively as possible. Although other classification methods, as those based on SVM, have been traditionally used, the DNN pose a promising tool in this field. However, the parameters and architecture setting of these DNNs present a serious restriction, due to the necessary time to find the most appropriate configuration. This paper proposes a new genetic algorithm designed to evolve the parameters, and the architecture, of a DNN with the goal of maximising the malware classification accuracy, and minimizing the complexity of the model. This model is tested against a dataset of malware samples, which are represented using a set of static features, so the DNN has been trained to perform a static malware classification task. The experiments carried out using this dataset show that the genetic algorithm is able to select the parameters and the DNN architecture settings, achieving a 91% accuracy.
{"title":"Evolving Deep Neural Networks architectures for Android malware classification","authors":"Alejandro Martín, Félix Fuentes-Hurtado, V. Naranjo, David Camacho","doi":"10.1109/CEC.2017.7969501","DOIUrl":"https://doi.org/10.1109/CEC.2017.7969501","url":null,"abstract":"Deep Neural Networks (DNN) have become a powerful, widely used, and successful mechanism to solve problems of different nature and varied complexity. Their ability to build models adapted to complex non-linear problems, have made them a technique widely applied and studied. One of the fields where this technique is currently being applied is in the malware classification problem. The malware classification problem has an increasing complexity, due to the growing number of features needed to represent the behaviour of the application as exhaustively as possible. Although other classification methods, as those based on SVM, have been traditionally used, the DNN pose a promising tool in this field. However, the parameters and architecture setting of these DNNs present a serious restriction, due to the necessary time to find the most appropriate configuration. This paper proposes a new genetic algorithm designed to evolve the parameters, and the architecture, of a DNN with the goal of maximising the malware classification accuracy, and minimizing the complexity of the model. This model is tested against a dataset of malware samples, which are represented using a set of static features, so the DNN has been trained to perform a static malware classification task. The experiments carried out using this dataset show that the genetic algorithm is able to select the parameters and the DNN architecture settings, achieving a 91% accuracy.","PeriodicalId":335123,"journal":{"name":"2017 IEEE Congress on Evolutionary Computation (CEC)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123707858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：