Pub Date : 2021-12-05DOI: 10.1109/SSCI50451.2021.9659551
Kaito Kimura, Yuan Tu, Riku Tanji, M. Mozgovoy
Interaction with opponents is a core element in video sports games. Thus, user experience in single-player matches heavily depends on the quality of AI opponents, who are expected to vary in their skill level and play styles. One way to achieve this goal is to learn game-playing behavior from real human players and to improve it if necessary with an automated optimization method. Monte-Carlo tree search (MCTS) has been successfully used for this purpose in several card and board games, such as chess and poker. We explore the possibility to apply MCTS in an action sports game of 3D tennis, and show how a dataset of pre-recorded tennis games can be used to train an MCTS-based AI system, exhibiting believable and reasonably skillful behavior.
{"title":"Creating Adjustable Human-like AI Behavior in a 3D Tennis Game with Monte-Carlo Tree Search","authors":"Kaito Kimura, Yuan Tu, Riku Tanji, M. Mozgovoy","doi":"10.1109/SSCI50451.2021.9659551","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9659551","url":null,"abstract":"Interaction with opponents is a core element in video sports games. Thus, user experience in single-player matches heavily depends on the quality of AI opponents, who are expected to vary in their skill level and play styles. One way to achieve this goal is to learn game-playing behavior from real human players and to improve it if necessary with an automated optimization method. Monte-Carlo tree search (MCTS) has been successfully used for this purpose in several card and board games, such as chess and poker. We explore the possibility to apply MCTS in an action sports game of 3D tennis, and show how a dataset of pre-recorded tennis games can be used to train an MCTS-based AI system, exhibiting believable and reasonably skillful behavior.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129557147","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660007
Hiba Arnout, Johanna Bronner, J. Kehrer, T. Runkler
In this paper, we consider the problem of translating time series of one controlled DC motor to imitate time series from another motor. Our main goal is to test different controllers and find the best performing controller for a motor operating in the field without knowing its mathematical model. By means of representation disentanglement, we present a new approach that splits the time series of each control system into two representation vectors: a first vector depicting the motor characteristics and its operating mode and a second vector describing the controller effect. We test our method on a scenario where we simulate the behavior of two different controlled DC motors. We map the behavior of a controller of a lab motor to a field motor. The experiments show that DR-TiST can recognize motor and controller characteristics and predict the right behavior.
{"title":"Translation of Time Series Data from Controlled DC Motors using Disentangled Representation Learning","authors":"Hiba Arnout, Johanna Bronner, J. Kehrer, T. Runkler","doi":"10.1109/SSCI50451.2021.9660007","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660007","url":null,"abstract":"In this paper, we consider the problem of translating time series of one controlled DC motor to imitate time series from another motor. Our main goal is to test different controllers and find the best performing controller for a motor operating in the field without knowing its mathematical model. By means of representation disentanglement, we present a new approach that splits the time series of each control system into two representation vectors: a first vector depicting the motor characteristics and its operating mode and a second vector describing the controller effect. We test our method on a scenario where we simulate the behavior of two different controlled DC motors. We map the behavior of a controller of a lab motor to a field motor. The experiments show that DR-TiST can recognize motor and controller characteristics and predict the right behavior.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126962554","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660045
A. Doboli, S. Doboli
Many modern applications require both modeling and generative capabilities, so that they can produce novel outcomes that address requirements beyond the solutions used in model training. Current AI approaches arguably emphasize modeling but pay much less attention to generative capabilities. This paper presents a new learning and response generating (LRG) agent-based model, in which interacting agents continuously learn symbolic - numeric knowledge and create new outcomes (responses) using a set of five ways to combine concepts. Each way has both fast, reactive and a slow, planned versions. Experiments present the characteristics of an agent's modeling and generating capabilities.
{"title":"A Novel Learning and Response Generating Agent-based Model for Symbolic - Numeric Knowledge Modeling and Combination","authors":"A. Doboli, S. Doboli","doi":"10.1109/SSCI50451.2021.9660045","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660045","url":null,"abstract":"Many modern applications require both modeling and generative capabilities, so that they can produce novel outcomes that address requirements beyond the solutions used in model training. Current AI approaches arguably emphasize modeling but pay much less attention to generative capabilities. This paper presents a new learning and response generating (LRG) agent-based model, in which interacting agents continuously learn symbolic - numeric knowledge and create new outcomes (responses) using a set of five ways to combine concepts. Each way has both fast, reactive and a slow, planned versions. Experiments present the characteristics of an agent's modeling and generating capabilities.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"167 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123752767","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660005
José Almeida, J. Soares, F. Lezama, B. Canizes, Z. Vale
The growing number of electric vehicles (EVs) on the road and renewable energy production to meet carbon reduction targets has posed numerous electrical grid problems. The increasing use of distributed energy resources (DER) in the grid poses severe operational issues, such as grid congestion and overloading. Active management of distribution networks using the smart grid (SG) technologies and artificial intelligence (AI) techniques by multiple entities. In this case, aggregators can support the grid's operation, providing a better product for the end-user. This study proposes an effective intraday energy resource management starting with a day-ahead time frame, considering the uncertainty associated with high DER penetration. The optimization is achieved considering five different metaheuristics (DE, HyDE-DF, DEEDA, CUMDANCauchy++, and HC2RCEDUMDA). Results show that the proposed model is effective for the multiple aggregators with variations from the day-ahead around the 6 % mark, except for the final aggregator. A Wilcoxon test is also applied to compare the performance of the CUMDANCauchy++ algorithm with the remaining. CUMDANCauchy++ shows competitive results beating all algorithms in all aggregators except for DEEDA, which presents similar results.
{"title":"Evolutionary Algorithms applied to the Intraday Energy Resource Scheduling in the Context of Multiple Aggregators","authors":"José Almeida, J. Soares, F. Lezama, B. Canizes, Z. Vale","doi":"10.1109/SSCI50451.2021.9660005","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660005","url":null,"abstract":"The growing number of electric vehicles (EVs) on the road and renewable energy production to meet carbon reduction targets has posed numerous electrical grid problems. The increasing use of distributed energy resources (DER) in the grid poses severe operational issues, such as grid congestion and overloading. Active management of distribution networks using the smart grid (SG) technologies and artificial intelligence (AI) techniques by multiple entities. In this case, aggregators can support the grid's operation, providing a better product for the end-user. This study proposes an effective intraday energy resource management starting with a day-ahead time frame, considering the uncertainty associated with high DER penetration. The optimization is achieved considering five different metaheuristics (DE, HyDE-DF, DEEDA, CUMDANCauchy++, and HC2RCEDUMDA). Results show that the proposed model is effective for the multiple aggregators with variations from the day-ahead around the 6 % mark, except for the final aggregator. A Wilcoxon test is also applied to compare the performance of the CUMDANCauchy++ algorithm with the remaining. CUMDANCauchy++ shows competitive results beating all algorithms in all aggregators except for DEEDA, which presents similar results.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127289956","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9659913
Xinming Shi, Jiashi Gao, Leandro L. Minku, James J. Q. Yu, Xin Yao
Time Delay Reservoir (TDR) can exhibit effects of high dimensionality and short-term memory based on delay differential equations (DDEs), as well as having hardware-friendly characteristics. However, the predictive performance and memory capacity of the standard TDRs are still limited, and dependent on the hyperparameter of the oscillation function. In this paper, we first analyze these limitations and their corresponding reasons. We find that the reasons for such limitations are fused by two aspects, which are the trade-off between the strength of self-feedback and neighboring-feedback caused by neuron separation, as well as the unsuitable order setting of the nonlinear function in DDE. Therefore, we propose a new form of TDR with second-order time delay to overcome such limitations, incurring a more flexible time-multiplexing. Moreover, a parameter-free nonlinear function is introduced to substitute the classic Mackey-Glass oscillator, which alleviates the problem of parameter dependency. Our experiments show that the proposed approach achieves better predictive performance and memory capacity compared with the standard TDR. Our proposed model also outperforms six other existing approaches on both time series prediction and recognition tasks.
{"title":"Second-order Time Delay Reservoir Computing for Nonlinear Time Series Problems","authors":"Xinming Shi, Jiashi Gao, Leandro L. Minku, James J. Q. Yu, Xin Yao","doi":"10.1109/SSCI50451.2021.9659913","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9659913","url":null,"abstract":"Time Delay Reservoir (TDR) can exhibit effects of high dimensionality and short-term memory based on delay differential equations (DDEs), as well as having hardware-friendly characteristics. However, the predictive performance and memory capacity of the standard TDRs are still limited, and dependent on the hyperparameter of the oscillation function. In this paper, we first analyze these limitations and their corresponding reasons. We find that the reasons for such limitations are fused by two aspects, which are the trade-off between the strength of self-feedback and neighboring-feedback caused by neuron separation, as well as the unsuitable order setting of the nonlinear function in DDE. Therefore, we propose a new form of TDR with second-order time delay to overcome such limitations, incurring a more flexible time-multiplexing. Moreover, a parameter-free nonlinear function is introduced to substitute the classic Mackey-Glass oscillator, which alleviates the problem of parameter dependency. Our experiments show that the proposed approach achieves better predictive performance and memory capacity compared with the standard TDR. Our proposed model also outperforms six other existing approaches on both time series prediction and recognition tasks.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115972615","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9659544
O. Odeyomi, G. Záruba
This paper discusses how clients in a federated learning system can collaborate with privacy guarantee in a fully decentralized setting without a central server. Most existing work includes a central server that aggregates the local updates from the clients and coordinates the training. Thus, the setting in this existing work is prone to communication and computational bottlenecks, especially when large number of clients are involved. Also, most existing federated learning algorithms do not cater for situations where the data distribution is time-varying such as in real-time traffic monitoring. To address these problems, this paper proposes a differentially-private online mirror descent algorithm. To provide additional privacy to the loss gradients of the clients, local differential privacy is introduced. Simulation results are based on a proposed differentially-private exponential gradient algorithm, which is a variant of differentially-private online mirror descent algorithm with entropic regularizer. The simulation shows that all the clients can converge to the global optimal vector over time. The regret bound of the proposed differentially-private exponential gradient algorithm is compared with the regret bounds of some state-of-the-art online federated learning algorithms found in the literature.
{"title":"Privacy-Preserving Online Mirror Descent for Federated Learning with Single-Sided Trust","authors":"O. Odeyomi, G. Záruba","doi":"10.1109/SSCI50451.2021.9659544","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9659544","url":null,"abstract":"This paper discusses how clients in a federated learning system can collaborate with privacy guarantee in a fully decentralized setting without a central server. Most existing work includes a central server that aggregates the local updates from the clients and coordinates the training. Thus, the setting in this existing work is prone to communication and computational bottlenecks, especially when large number of clients are involved. Also, most existing federated learning algorithms do not cater for situations where the data distribution is time-varying such as in real-time traffic monitoring. To address these problems, this paper proposes a differentially-private online mirror descent algorithm. To provide additional privacy to the loss gradients of the clients, local differential privacy is introduced. Simulation results are based on a proposed differentially-private exponential gradient algorithm, which is a variant of differentially-private online mirror descent algorithm with entropic regularizer. The simulation shows that all the clients can converge to the global optimal vector over time. The regret bound of the proposed differentially-private exponential gradient algorithm is compared with the regret bounds of some state-of-the-art online federated learning algorithms found in the literature.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"7 7","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120891697","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660078
Jaeyoon Lee, Hyuntak Lim, Ki-Seok Chung
Deep neural networks reveal their usefulness through learning from large amounts of data. However, unless the data is correctly labeled, it may be very difficult to properly train a neural network. Labeling the large set of data is a time-consuming and labor-intensive task. To overcome the risk of mislabeling, several methods that are robust against the label noise have been proposed. In this paper, we propose an effective label correction method called Curriculum Label Correction (CLC). With reference to the loss distribution from self-supervised learning, CLC identifies and corrects noisy labels utilizing curriculum learning. Our experimental results verify that CLC shows outstanding performance especially in a harshly noisy condition, 91.06% test accuracy on CIFAR-10 at a noise rate of 0.8. Code is available at https://github.com/LJY-HY/CLC.
{"title":"CLC: Noisy Label Correction via Curriculum Learning","authors":"Jaeyoon Lee, Hyuntak Lim, Ki-Seok Chung","doi":"10.1109/SSCI50451.2021.9660078","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660078","url":null,"abstract":"Deep neural networks reveal their usefulness through learning from large amounts of data. However, unless the data is correctly labeled, it may be very difficult to properly train a neural network. Labeling the large set of data is a time-consuming and labor-intensive task. To overcome the risk of mislabeling, several methods that are robust against the label noise have been proposed. In this paper, we propose an effective label correction method called Curriculum Label Correction (CLC). With reference to the loss distribution from self-supervised learning, CLC identifies and corrects noisy labels utilizing curriculum learning. Our experimental results verify that CLC shows outstanding performance especially in a harshly noisy condition, 91.06% test accuracy on CIFAR-10 at a noise rate of 0.8. Code is available at https://github.com/LJY-HY/CLC.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"39 7","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121011295","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660186
Mustafa Atay, Hailey Gipson, Tony Gwyn, K. Roy
The prevalent commercial deployment of automated facial analysis systems such as face recognition as a robust authentication method has increasingly fueled scientific attention. Current machine learning algorithms allow for a relatively reliable detection, recognition, and categorization of face images comprised of age, race, and gender. Algorithms with such biased data are bound to produce skewed results. It leads to a significant decrease in the performance of state-of-the-art models when applied to images of gender or ethnicity groups. In this paper, we study the gender bias in facial recognition with gender balanced and imbalanced training sets using five traditional machine learning algorithms. We aim to report the machine learning classifiers which are inclined towards gender bias and the ones which mitigate it. Miss rates metric is effective in finding out potential bias in predictions. Our study utilizes miss rates metric along with a standard metric such as accuracy, precision or recall to evaluate possible gender bias effectively.
{"title":"Evaluation of Gender Bias in Facial Recognition with Traditional Machine Learning Algorithms","authors":"Mustafa Atay, Hailey Gipson, Tony Gwyn, K. Roy","doi":"10.1109/SSCI50451.2021.9660186","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660186","url":null,"abstract":"The prevalent commercial deployment of automated facial analysis systems such as face recognition as a robust authentication method has increasingly fueled scientific attention. Current machine learning algorithms allow for a relatively reliable detection, recognition, and categorization of face images comprised of age, race, and gender. Algorithms with such biased data are bound to produce skewed results. It leads to a significant decrease in the performance of state-of-the-art models when applied to images of gender or ethnicity groups. In this paper, we study the gender bias in facial recognition with gender balanced and imbalanced training sets using five traditional machine learning algorithms. We aim to report the machine learning classifiers which are inclined towards gender bias and the ones which mitigate it. Miss rates metric is effective in finding out potential bias in predictions. Our study utilizes miss rates metric along with a standard metric such as accuracy, precision or recall to evaluate possible gender bias effectively.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122285661","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660134
Yuxuan Huang, Luiz Fernando Capretz, D. Ho
Application of machine learning for stock prediction is attracting a lot of attention in recent years. A large amount of research has been conducted in this area and multiple existing results have shown that machine learning methods could be successfully used toward stock predicting using stocks' historical data. Most of these existing approaches have focused on short term prediction using stocks' historical price and technical indicators. In this paper, we prepared 22 years' worth of stock quarterly financial data and investigated three machine learning algorithms: Feed-forward Neural Network (FNN), Random Forest (RF) and Adaptive Neural Fuzzy Inference System (ANFIS) for stock prediction based on fundamental analysis. In addition, we applied RF based feature selection and bootstrap aggregation in order to improve model performance and aggregate predictions from different models. Our results show that RF model achieves the best prediction results, and feature selection is able to improve test performance of FNN and ANFIS. Moreover, the aggregated model outperforms all baseline models as well as the benchmark DJIA index by an acceptable margin for the test period. Our findings demonstrate that machine learning models could be used to aid fundamental analysts with decision-making regarding stock investment.
{"title":"Machine Learning for Stock Prediction Based on Fundamental Analysis","authors":"Yuxuan Huang, Luiz Fernando Capretz, D. Ho","doi":"10.1109/SSCI50451.2021.9660134","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660134","url":null,"abstract":"Application of machine learning for stock prediction is attracting a lot of attention in recent years. A large amount of research has been conducted in this area and multiple existing results have shown that machine learning methods could be successfully used toward stock predicting using stocks' historical data. Most of these existing approaches have focused on short term prediction using stocks' historical price and technical indicators. In this paper, we prepared 22 years' worth of stock quarterly financial data and investigated three machine learning algorithms: Feed-forward Neural Network (FNN), Random Forest (RF) and Adaptive Neural Fuzzy Inference System (ANFIS) for stock prediction based on fundamental analysis. In addition, we applied RF based feature selection and bootstrap aggregation in order to improve model performance and aggregate predictions from different models. Our results show that RF model achieves the best prediction results, and feature selection is able to improve test performance of FNN and ANFIS. Moreover, the aggregated model outperforms all baseline models as well as the benchmark DJIA index by an acceptable margin for the test period. Our findings demonstrate that machine learning models could be used to aid fundamental analysts with decision-making regarding stock investment.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"119 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116317314","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 : 2021-12-05DOI: 10.1109/SSCI50451.2021.9660097
Cheng Gong, Lie Meng Pang, H. Ishibuchi
A good initial population generation method is of necessity to improve the performance of evolutionary multiobjective optimization (EMO) algorithms. However, until now only a few methods for generating an initial population have been proposed for EMO algorithms. In this paper, we propose a simple idea of generating an initial population for a popular decomposition-based algorithm, i.e., MOEA/D with the penalty-based boundary intersection (PBI) function, and demonstrate its effectiveness. The basic idea is to generate more initial solutions than the population size and to assign an appropriate solution to each weight vector. Firstly, we modify the initialization phase of MOEA/D through two different strategies based on this idea. Then, the modified MOEA/D algorithms are compared with the original MOEA/D on frequently-used many-objective test problems: DTLZ1, DTLZ3 and DTLZ4. Our experimental results clearly show that the proposed initial population generation method can significantly improve the performance of the original MOEA/D.
{"title":"Initial Population Generation Method and its Effects on MOEA/D","authors":"Cheng Gong, Lie Meng Pang, H. Ishibuchi","doi":"10.1109/SSCI50451.2021.9660097","DOIUrl":"https://doi.org/10.1109/SSCI50451.2021.9660097","url":null,"abstract":"A good initial population generation method is of necessity to improve the performance of evolutionary multiobjective optimization (EMO) algorithms. However, until now only a few methods for generating an initial population have been proposed for EMO algorithms. In this paper, we propose a simple idea of generating an initial population for a popular decomposition-based algorithm, i.e., MOEA/D with the penalty-based boundary intersection (PBI) function, and demonstrate its effectiveness. The basic idea is to generate more initial solutions than the population size and to assign an appropriate solution to each weight vector. Firstly, we modify the initialization phase of MOEA/D through two different strategies based on this idea. Then, the modified MOEA/D algorithms are compared with the original MOEA/D on frequently-used many-objective test problems: DTLZ1, DTLZ3 and DTLZ4. Our experimental results clearly show that the proposed initial population generation method can significantly improve the performance of the original MOEA/D.","PeriodicalId":255763,"journal":{"name":"2021 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125518876","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：