Pub Date : 2021-06-05DOI: 10.1080/0952813X.2021.1928299
Ronaldo Vigo, Jay Wimsatt, Charles A. Doan, Derek E. Zeigler
ABSTRACT In the past two decades, human categorisation research has achieved significant progress via the rigorous and systematic study of concepts in terms of category structures and their families. The importance of these structure families stems from evidence suggesting that learning and categorisation performance are not only limited by low- and high-level generalisation mechanisms but by the inherent nature of the environmental and mental stimuli entertained by observers during the concept learning process. In this paper, we propose a new direction for concept learning and categorisation research based on several dual paradigmatic tensions that hinge on the inherent nature of the components of stimuli, limitations of the innate abilities of the observer to process such components, and the relationship between the two. The tensions range from the various possible properties and constraints of the dimensions underlying categories of object stimuli to various notions of supervised learning capable of significantly altering concept learnability. The substantial extant literature on concept learning research indicates that rigorous empirical investigations targeting these tensions are either non-existent or, at best, severely lacking despite their ecological significance. We shall argue that future theory building about concept learning should attempt to resolve these tensions and that without the proper empirical and theoretical focus on them, concept learning research will fail to achieve its ultimate goals anytime soon.
{"title":"Raising the Bar for Theories of Categorisation and Concept Learning: The Need to Resolve Five Basic Paradigmatic Tensions","authors":"Ronaldo Vigo, Jay Wimsatt, Charles A. Doan, Derek E. Zeigler","doi":"10.1080/0952813X.2021.1928299","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1928299","url":null,"abstract":"ABSTRACT In the past two decades, human categorisation research has achieved significant progress via the rigorous and systematic study of concepts in terms of category structures and their families. The importance of these structure families stems from evidence suggesting that learning and categorisation performance are not only limited by low- and high-level generalisation mechanisms but by the inherent nature of the environmental and mental stimuli entertained by observers during the concept learning process. In this paper, we propose a new direction for concept learning and categorisation research based on several dual paradigmatic tensions that hinge on the inherent nature of the components of stimuli, limitations of the innate abilities of the observer to process such components, and the relationship between the two. The tensions range from the various possible properties and constraints of the dimensions underlying categories of object stimuli to various notions of supervised learning capable of significantly altering concept learnability. The substantial extant literature on concept learning research indicates that rigorous empirical investigations targeting these tensions are either non-existent or, at best, severely lacking despite their ecological significance. We shall argue that future theory building about concept learning should attempt to resolve these tensions and that without the proper empirical and theoretical focus on them, concept learning research will fail to achieve its ultimate goals anytime soon.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"45 1","pages":"845 - 869"},"PeriodicalIF":2.2,"publicationDate":"2021-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86587331","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-06-02DOI: 10.1080/0952813X.2021.1924868
Subramonian Krishna Sarma
ABSTRACT Internet of Things (IoT) is a new revolution of the Internet. However, the IoT network of physical devices and objects is often vulnerable to attacks like Denial of Service (DoS) and Distributed Denial of Service (DDoS). The proposed attack detection system makes the interlinking of Development and Operations (DevOps) as it makes the relationship between development and IT operations. For this, the proposed system includes (i) Proposed Feature Extraction and (ii) Classification. The data from each application are processed under the initial stage of feature extraction, where the statistical and higher-order statistical features are concatenated. Subsequently, the extracted features are subjected to a classification process, where it determines the presence of attacks. For the classification process, this paper intends to deploy the optimised Deep Belief Network (DBN), in which the activation function is optimally tuned. A new hybrid algorithm termed Firefly Alpha Evaluated Grey Wolf Optimisation (FAE-GWO) algorithm is proposed, which is the combination of Firefly (FF) and Grey Wolf Optimisation (GWO). Finally, the performance of the proposed system model is compared over other conventional works in terms of certain performance measures.
{"title":"Hybrid optimised deep learning-deep belief network for attack detection in the internet of things","authors":"Subramonian Krishna Sarma","doi":"10.1080/0952813X.2021.1924868","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1924868","url":null,"abstract":"ABSTRACT Internet of Things (IoT) is a new revolution of the Internet. However, the IoT network of physical devices and objects is often vulnerable to attacks like Denial of Service (DoS) and Distributed Denial of Service (DDoS). The proposed attack detection system makes the interlinking of Development and Operations (DevOps) as it makes the relationship between development and IT operations. For this, the proposed system includes (i) Proposed Feature Extraction and (ii) Classification. The data from each application are processed under the initial stage of feature extraction, where the statistical and higher-order statistical features are concatenated. Subsequently, the extracted features are subjected to a classification process, where it determines the presence of attacks. For the classification process, this paper intends to deploy the optimised Deep Belief Network (DBN), in which the activation function is optimally tuned. A new hybrid algorithm termed Firefly Alpha Evaluated Grey Wolf Optimisation (FAE-GWO) algorithm is proposed, which is the combination of Firefly (FF) and Grey Wolf Optimisation (GWO). Finally, the performance of the proposed system model is compared over other conventional works in terms of certain performance measures.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"45 1","pages":"695 - 724"},"PeriodicalIF":2.2,"publicationDate":"2021-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89433848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-05-24DOI: 10.1080/0952813X.2021.1924871
G. Szücs, Dávid Papp
ABSTRACT Given the challenge of gathering labelled training data for machine learning tasks, active learning has become popular. This paper focuses on the beginning of unsupervised active learning, where there are no labelled data at all. The aim of this zero initialised unsupervised active learning is to select the most informative examples – even from an imbalanced dataset – to be labelled manually. Our solution with proposed selection strategy, called Optimally Balanced Entropy-Based Sampling (OBEBS) reaches a balanced training set at each step to avoid imbalanced problems. Two theorems of the optimal solution for selection strategy are also presented and proved in the paper. At the beginning of the active learning, there is not enough information for supervised machine learning method, thus our selection strategy is based on unsupervised learning (clustering). The cluster membership likelihoods of the items are essential for the algorithm to connect the clusters and the classes, i.e., to find assignment between them. For the best assignment, the Hungarian algorithm is used, and single, multi, and adaptive assignment variants of OBEBS method are developed. Based on generated and real images datasets of handwritten digits, the experimental results show that our method surpasses the state-of-the-art methods.
{"title":"Zero Initialised Unsupervised Active Learning by Optimally Balanced Entropy-Based Sampling for Imbalanced Problems","authors":"G. Szücs, Dávid Papp","doi":"10.1080/0952813X.2021.1924871","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1924871","url":null,"abstract":"ABSTRACT Given the challenge of gathering labelled training data for machine learning tasks, active learning has become popular. This paper focuses on the beginning of unsupervised active learning, where there are no labelled data at all. The aim of this zero initialised unsupervised active learning is to select the most informative examples – even from an imbalanced dataset – to be labelled manually. Our solution with proposed selection strategy, called Optimally Balanced Entropy-Based Sampling (OBEBS) reaches a balanced training set at each step to avoid imbalanced problems. Two theorems of the optimal solution for selection strategy are also presented and proved in the paper. At the beginning of the active learning, there is not enough information for supervised machine learning method, thus our selection strategy is based on unsupervised learning (clustering). The cluster membership likelihoods of the items are essential for the algorithm to connect the clusters and the classes, i.e., to find assignment between them. For the best assignment, the Hungarian algorithm is used, and single, multi, and adaptive assignment variants of OBEBS method are developed. Based on generated and real images datasets of handwritten digits, the experimental results show that our method surpasses the state-of-the-art methods.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"13 1","pages":"781 - 814"},"PeriodicalIF":2.2,"publicationDate":"2021-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78136276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-05-24DOI: 10.1080/0952813X.2021.1924870
A. K. Hwaitat, R. Al-Sayyed, Imad Salah, S. Manaseer, H. Al-Bdour, Sarah Shukri
ABSTRACT PSO is a remarkable tool for solving several optimisation problems, like global optimisation and many real-life problems. It generally explores global optimal solution via exploiting the particle – swarm’s memory. Its limited properties on objective function’s continuity along with the search space and its potentiality in adapting dynamic environment make the PSO an important meta-heuristic method. PSO has an inherent tendency of trapping at local optimum which affects the convergence prematurely, when trying to solve difficult problems. This work proposed a modified version of PSO called as FPSO, where frequency-wave-sound is employed to exit from any encountered local optimum; if it is not the optimal solution. This FPSO mimics the characteristics of the waves by using three parameters, namely amplitude, frequency and wavelength. FPSO is then compared and analysed with other renowned algorithms like conventional PSO, Grey Wolf Optimisation (GOW), Multi-Verse Optimiser (MVO), Moth-Flame Optimisation (SL-PSO), Sine Cosine Algorithm (PPSO) and Butterfly Optimisation Algorithm (BOA) on 23 bench marking test bed functions. The performance is evaluated using various measures including trajectory, search history, average fitness solution and best optimisation-solution. The obtained results show that the FPSO algorithm beats other metaheuristic algorithms and confirmed its better performance on 2-dimensional test functions.
{"title":"Frequencies Wave Sound Particle Swarm Optimisation (FPSO)","authors":"A. K. Hwaitat, R. Al-Sayyed, Imad Salah, S. Manaseer, H. Al-Bdour, Sarah Shukri","doi":"10.1080/0952813X.2021.1924870","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1924870","url":null,"abstract":"ABSTRACT PSO is a remarkable tool for solving several optimisation problems, like global optimisation and many real-life problems. It generally explores global optimal solution via exploiting the particle – swarm’s memory. Its limited properties on objective function’s continuity along with the search space and its potentiality in adapting dynamic environment make the PSO an important meta-heuristic method. PSO has an inherent tendency of trapping at local optimum which affects the convergence prematurely, when trying to solve difficult problems. This work proposed a modified version of PSO called as FPSO, where frequency-wave-sound is employed to exit from any encountered local optimum; if it is not the optimal solution. This FPSO mimics the characteristics of the waves by using three parameters, namely amplitude, frequency and wavelength. FPSO is then compared and analysed with other renowned algorithms like conventional PSO, Grey Wolf Optimisation (GOW), Multi-Verse Optimiser (MVO), Moth-Flame Optimisation (SL-PSO), Sine Cosine Algorithm (PPSO) and Butterfly Optimisation Algorithm (BOA) on 23 bench marking test bed functions. The performance is evaluated using various measures including trajectory, search history, average fitness solution and best optimisation-solution. The obtained results show that the FPSO algorithm beats other metaheuristic algorithms and confirmed its better performance on 2-dimensional test functions.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"49 1","pages":"749 - 780"},"PeriodicalIF":2.2,"publicationDate":"2021-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73307977","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-05-24DOI: 10.1080/0952813X.2021.1924869
Hong Zhang, Qian Cao, Hedi Yang
ABSTRACT This paper deals with a class of delayed Nicholson-type systems involving patch structure. First, we prove that the solution of the initial value problem with respect to the addressed system exists globally and is bounded. Second, we employ the contraction fixed point theorem and analytical techniques to establish the existence of a positive asymptotically almost periodic solution and its global attractivity. Finally, an example is arranged to illustrate the effectiveness and feasibility of the obtained results.
{"title":"Dynamics analysis of delayed Nicholson-type systems involving patch structure and asymptotically almost periodic environments","authors":"Hong Zhang, Qian Cao, Hedi Yang","doi":"10.1080/0952813X.2021.1924869","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1924869","url":null,"abstract":"ABSTRACT This paper deals with a class of delayed Nicholson-type systems involving patch structure. First, we prove that the solution of the initial value problem with respect to the addressed system exists globally and is bounded. Second, we employ the contraction fixed point theorem and analytical techniques to establish the existence of a positive asymptotically almost periodic solution and its global attractivity. Finally, an example is arranged to illustrate the effectiveness and feasibility of the obtained results.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"20 1","pages":"725 - 748"},"PeriodicalIF":2.2,"publicationDate":"2021-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87199963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-05-16DOI: 10.1080/0952813X.2021.1924867
Habibeh Nazif
ABSTRACT Transportation is a significant issue due to providing people to participate in human activities. Due to an increase in population, the need for transportation has also been increased. Therefore, more traffic is visible on streets that produce more issues related to mobility like noise pollution, air pollution, and accidents. This study pays attention to an impressive transit network design in urban areas. Because of the NP-hard nature of this problem, a shark smell optimisation (SSO) algorithm based on fuzzy logic is employed. A developed system is utilised to produce, optimise, and analyse frequencies and routes of transit in the level of a network. Its target is maximising the direct travellers per unit length, i.e., subject to route length, direct traveller density, and nonlinear rate constraints (a route length ratio to the shortest road interval between the beginning and destination). Since designing an urban transport network issue is in heterogeneous environments is involved, this article provides a new method for lowering the feasible urban travel time, the urban traffic, and the feasible urban travel cost using a well-known SSO algorithm. According to the results, the proposed method has higher efficiency compared to the previous methods. In addition, the results showed that the proposed technique offers fewer transfers and travel time.
{"title":"A fuzzy logic-based method for designing an urban transport network using a shark smell optimisation algorithm","authors":"Habibeh Nazif","doi":"10.1080/0952813X.2021.1924867","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1924867","url":null,"abstract":"ABSTRACT Transportation is a significant issue due to providing people to participate in human activities. Due to an increase in population, the need for transportation has also been increased. Therefore, more traffic is visible on streets that produce more issues related to mobility like noise pollution, air pollution, and accidents. This study pays attention to an impressive transit network design in urban areas. Because of the NP-hard nature of this problem, a shark smell optimisation (SSO) algorithm based on fuzzy logic is employed. A developed system is utilised to produce, optimise, and analyse frequencies and routes of transit in the level of a network. Its target is maximising the direct travellers per unit length, i.e., subject to route length, direct traveller density, and nonlinear rate constraints (a route length ratio to the shortest road interval between the beginning and destination). Since designing an urban transport network issue is in heterogeneous environments is involved, this article provides a new method for lowering the feasible urban travel time, the urban traffic, and the feasible urban travel cost using a well-known SSO algorithm. According to the results, the proposed method has higher efficiency compared to the previous methods. In addition, the results showed that the proposed technique offers fewer transfers and travel time.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"142 1","pages":"673 - 694"},"PeriodicalIF":2.2,"publicationDate":"2021-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89943268","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-04-24DOI: 10.1080/0952813X.2021.1907792
Sandip J Modha, Prasenjit Majumder, Thomas Mandl
ABSTRACT Modeling text in a numerical representation is a prime task for any Natural Language Processing downstream task such as text classification. This paper attempts to study the effectiveness of text representation schemes on the text classification task, such as aggressive text detection, a special case of Hate speech from social media. Aggression levels are categorized into three predefined classes, namely: ‘Non-aggressive’ (NAG), ‘Overtly Aggressive’ (OAG), and ‘Covertly Aggressive’ (CAG). Various text representation schemes based on BoW techniques, word embedding, contextual word embedding, sentence embedding on traditional classifiers, and deep neural models are compared on a text classification problem. The weighted score is used as a primary evaluation metric. The results show that text representation using Googles’ universal sentence encoder (USE) performs better than word embedding and BoW techniques on traditional classifiers, such as SVM, while pre-trained word embedding models perform better on classifiers based on the deep neural models on the English dataset. Recent pre-trained transfer learning models like Elmo, ULMFi, and BERT are fine-tuned for the aggression classification task. However, results are not at par with the pre-trained word embedding model. Overall, word embedding using pre-trained fastText vectors produces the best weighted -score than Word2Vec and Glove. On the Hindi dataset, BoW techniques perform better than word embeddings on traditional classifiers such as SVM. In contrast, pre-trained word embedding models perform better on classifiers based on the deep neural nets. Statistical significance tests are employed to ensure the significance of the classification results. Deep neural models are more robust against the bias induced by the training dataset. They perform substantially better than traditional classifiers, such as SVM, logistic regression, and Naive Bayes classifiers on the Twitter test dataset.
{"title":"An empirical evaluation of text representation schemes to filter the social media stream","authors":"Sandip J Modha, Prasenjit Majumder, Thomas Mandl","doi":"10.1080/0952813X.2021.1907792","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1907792","url":null,"abstract":"ABSTRACT Modeling text in a numerical representation is a prime task for any Natural Language Processing downstream task such as text classification. This paper attempts to study the effectiveness of text representation schemes on the text classification task, such as aggressive text detection, a special case of Hate speech from social media. Aggression levels are categorized into three predefined classes, namely: ‘Non-aggressive’ (NAG), ‘Overtly Aggressive’ (OAG), and ‘Covertly Aggressive’ (CAG). Various text representation schemes based on BoW techniques, word embedding, contextual word embedding, sentence embedding on traditional classifiers, and deep neural models are compared on a text classification problem. The weighted score is used as a primary evaluation metric. The results show that text representation using Googles’ universal sentence encoder (USE) performs better than word embedding and BoW techniques on traditional classifiers, such as SVM, while pre-trained word embedding models perform better on classifiers based on the deep neural models on the English dataset. Recent pre-trained transfer learning models like Elmo, ULMFi, and BERT are fine-tuned for the aggression classification task. However, results are not at par with the pre-trained word embedding model. Overall, word embedding using pre-trained fastText vectors produces the best weighted -score than Word2Vec and Glove. On the Hindi dataset, BoW techniques perform better than word embeddings on traditional classifiers such as SVM. In contrast, pre-trained word embedding models perform better on classifiers based on the deep neural nets. Statistical significance tests are employed to ensure the significance of the classification results. Deep neural models are more robust against the bias induced by the training dataset. They perform substantially better than traditional classifiers, such as SVM, logistic regression, and Naive Bayes classifiers on the Twitter test dataset.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"32 1","pages":"499 - 525"},"PeriodicalIF":2.2,"publicationDate":"2021-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77833498","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-04-14DOI: 10.1080/0952813X.2021.1908430
Fethi Demim, S. Benmansour, A. Nemra, A. Rouigueb, M. Hamerlain, A. Bazoula
ABSTRACT Localisation technology is one of the most important challenges of underwater vehicle applications that accomplish any scheduled mission in the complex underwater environment. Currently, the Simultaneous Localisation and Mapping (SLAM) of the Autonomous Underwater Vehicle (AUV) is becoming a hotspot research. AUVs have, only recently, received more attention and underwater platforms continue to dominate the research. To ensure the success of an accurate AUV localisation mission, the problem of drift on the estimated trajectory must be overcome. In order to improve the positioning accuracy of the AUV localisation, a new filter referred to as the Adaptive Smooth Variable Structure Filter (ASVSF) based SLAM positioning algorithm is proposed. To verify the improvement of this filter, the combined SVSF and the Extended Kalman Filter (EKF) are presented. Experimental results based on dataset for underwater SLAM algorithm show the accuracy and stability of the ASVSF AUV localisation position. Several experiments were tested under real-life conditions with an autonomous underwater vehicle based on different filters. The results of these filters have been compared based on Root Mean Squared Error (RMSE) and in terms of localisation and map building errors. It is shown that the adaptive SVSF-SLAM strategy obtains the best performance compared to other algorithms.
{"title":"Simultaneous localisation and mapping for autonomous underwater vehicle using a combined smooth variable structure filter and extended kalman filter","authors":"Fethi Demim, S. Benmansour, A. Nemra, A. Rouigueb, M. Hamerlain, A. Bazoula","doi":"10.1080/0952813X.2021.1908430","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1908430","url":null,"abstract":"ABSTRACT Localisation technology is one of the most important challenges of underwater vehicle applications that accomplish any scheduled mission in the complex underwater environment. Currently, the Simultaneous Localisation and Mapping (SLAM) of the Autonomous Underwater Vehicle (AUV) is becoming a hotspot research. AUVs have, only recently, received more attention and underwater platforms continue to dominate the research. To ensure the success of an accurate AUV localisation mission, the problem of drift on the estimated trajectory must be overcome. In order to improve the positioning accuracy of the AUV localisation, a new filter referred to as the Adaptive Smooth Variable Structure Filter (ASVSF) based SLAM positioning algorithm is proposed. To verify the improvement of this filter, the combined SVSF and the Extended Kalman Filter (EKF) are presented. Experimental results based on dataset for underwater SLAM algorithm show the accuracy and stability of the ASVSF AUV localisation position. Several experiments were tested under real-life conditions with an autonomous underwater vehicle based on different filters. The results of these filters have been compared based on Root Mean Squared Error (RMSE) and in terms of localisation and map building errors. It is shown that the adaptive SVSF-SLAM strategy obtains the best performance compared to other algorithms.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"8 1","pages":"621 - 650"},"PeriodicalIF":2.2,"publicationDate":"2021-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79670795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-04-13DOI: 10.1080/0952813X.2021.1908431
B. Aksoy, O. Salman
ABSTRACT Today, health is the most important value of human life pandemics at different time intervals in the world history. Finally, the COVID-19 outbreak that occurred in Wuhan, China in December 2019, spread to the whole world in a really short time and caused a pandemic. In order to prevent this pandemic, early detection of the COVID-19 is very important. In this study, chest x-ray images of 1019 patients with open-source dataset were taken from four different sources. The images were analysed using Capsule Networks (CapsNet) model, which is one of the deep learning methods, whose popularity has increased in recent years. With the designed CapsNet model, individuals with COVID-19 disease were tried to be identified. The designed CapsNet model can detect COVID-19 disease with an accuracy rate of 98.02%. The obtained model cloud computing application was developed in order to use the work performed faster and easier.
{"title":"Detection of COVID-19 Disease in Chest X-Ray Images with capsul networks: application with cloud computing","authors":"B. Aksoy, O. Salman","doi":"10.1080/0952813X.2021.1908431","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1908431","url":null,"abstract":"ABSTRACT Today, health is the most important value of human life pandemics at different time intervals in the world history. Finally, the COVID-19 outbreak that occurred in Wuhan, China in December 2019, spread to the whole world in a really short time and caused a pandemic. In order to prevent this pandemic, early detection of the COVID-19 is very important. In this study, chest x-ray images of 1019 patients with open-source dataset were taken from four different sources. The images were analysed using Capsule Networks (CapsNet) model, which is one of the deep learning methods, whose popularity has increased in recent years. With the designed CapsNet model, individuals with COVID-19 disease were tried to be identified. The designed CapsNet model can detect COVID-19 disease with an accuracy rate of 98.02%. The obtained model cloud computing application was developed in order to use the work performed faster and easier.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"14 1","pages":"527 - 541"},"PeriodicalIF":2.2,"publicationDate":"2021-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90633759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-04-12DOI: 10.1080/0952813X.2021.1907794
Decui Liang, Xin He, Zeshui Xu, Jiahong Li
ABSTRACT In the study of two-sided matching decision problems, preference ordinal information is a key factor. However, in real life, it is often difficult to ascertain complete preference ordinal information, and in most cases we can only obtain an interval-valued preference ordinal information. In this paper, a strict two-sided matching based on multi-attribute interval-valued preference ordinal information is discussed. As a generalised decision model, the strict two-sided matching adequately considers the requirement of satisfaction degree of two-sided agents. Firstly, the ranking method of probability degree is introduced to deal with the information of various interval numbers. Then, in the case of multiple attributes, we propose two methods for strict two-sided matching problem. The one is to aggregate multi-attribute satisfaction degree and then construct the decision model. The another is to separately deal with the interval-valued preference ordinal information of each attribute and then design the corresponding model. Finally, in the context of Internet finance, we adopt an example of the venture capital two-sided matching problem to illustrate our proposed methods and confirm the effectiveness.
{"title":"Multi-attribute strict two-sided matching methods with interval-valued preference ordinal information","authors":"Decui Liang, Xin He, Zeshui Xu, Jiahong Li","doi":"10.1080/0952813X.2021.1907794","DOIUrl":"https://doi.org/10.1080/0952813X.2021.1907794","url":null,"abstract":"ABSTRACT In the study of two-sided matching decision problems, preference ordinal information is a key factor. However, in real life, it is often difficult to ascertain complete preference ordinal information, and in most cases we can only obtain an interval-valued preference ordinal information. In this paper, a strict two-sided matching based on multi-attribute interval-valued preference ordinal information is discussed. As a generalised decision model, the strict two-sided matching adequately considers the requirement of satisfaction degree of two-sided agents. Firstly, the ranking method of probability degree is introduced to deal with the information of various interval numbers. Then, in the case of multiple attributes, we propose two methods for strict two-sided matching problem. The one is to aggregate multi-attribute satisfaction degree and then construct the decision model. The another is to separately deal with the interval-valued preference ordinal information of each attribute and then design the corresponding model. Finally, in the context of Internet finance, we adopt an example of the venture capital two-sided matching problem to illustrate our proposed methods and confirm the effectiveness.","PeriodicalId":15677,"journal":{"name":"Journal of Experimental & Theoretical Artificial Intelligence","volume":"40 1","pages":"545 - 569"},"PeriodicalIF":2.2,"publicationDate":"2021-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74588474","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","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：