Pub Date : 2016-12-06DOI: 10.1109/SSCI.2016.7849947
Cong Liu, B. V. Dongen, Nour Assy, Wil M.P. van der Aalst
Tremendous amounts of data can be recorded during software execution. This provides valuable information on software runtime analysis. Many crashes and exceptions may occur, and it is a real challenge to understand how software is behaving. Software is usually composed of various components. A component is a nearly independent part of software that full-fills a clear function. Process mining aims to discover, monitor and improve real processes by extracting knowledge from event logs. This paper presents an approach to utilize process mining as a tool to discover the real behavior of software and analyze it. The unstructured software execution data may be too complex, involving multiple interleaved components, etc. Applying existing process mining techniques results in spaghetti-like models with no clear structure and no valuable information that can be easily understood by end. In this paper, we start with the observation that software is composed of components and we use this information to decompose the problem into smaller independent ones by discovering a behavioral model per component. Through experimental analysis, we illustrate that the proposed approach facilitates the discovery of more understandable software models. All proposed approaches have been implemented in the open-source process mining toolkit ProM.
{"title":"Component behavior discovery from software execution data","authors":"Cong Liu, B. V. Dongen, Nour Assy, Wil M.P. van der Aalst","doi":"10.1109/SSCI.2016.7849947","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849947","url":null,"abstract":"Tremendous amounts of data can be recorded during software execution. This provides valuable information on software runtime analysis. Many crashes and exceptions may occur, and it is a real challenge to understand how software is behaving. Software is usually composed of various components. A component is a nearly independent part of software that full-fills a clear function. Process mining aims to discover, monitor and improve real processes by extracting knowledge from event logs. This paper presents an approach to utilize process mining as a tool to discover the real behavior of software and analyze it. The unstructured software execution data may be too complex, involving multiple interleaved components, etc. Applying existing process mining techniques results in spaghetti-like models with no clear structure and no valuable information that can be easily understood by end. In this paper, we start with the observation that software is composed of components and we use this information to decompose the problem into smaller independent ones by discovering a behavioral model per component. Through experimental analysis, we illustrate that the proposed approach facilitates the discovery of more understandable software models. All proposed approaches have been implemented in the open-source process mining toolkit ProM.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122392473","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7849983
Christos Varytimidis, Georgios Tsatiris, Konstantinos Rapantzikos, S. Kollias
There is a need for automatic processing and extracting of meaningful metadata from multimedia information, especially in the audiovisual industry. This higher level information is used in a variety of practices, such as enriching multimedia content with external links, clickable objects and useful related information in general. This paper presents a system for efficient multimedia content analysis and automatic annotation within a multimedia processing and publishing framework. This system is comprised of three modules: the first provides detection of faces and recognition of known persons; the second provides generic object detection, based on a deep convolutional neural network topology; the third provides automated location estimation and landmark recognition based on state-of-the-art technologies. The results are exported in meaningful metadata that can be utilized in various ways. The system has been developed and successfully tested in the framework of the EC Horizon 2020 Mecanex project, targeting advertising and production markets.
{"title":"A systemic approach to automatic metadata extraction from multimedia content","authors":"Christos Varytimidis, Georgios Tsatiris, Konstantinos Rapantzikos, S. Kollias","doi":"10.1109/SSCI.2016.7849983","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849983","url":null,"abstract":"There is a need for automatic processing and extracting of meaningful metadata from multimedia information, especially in the audiovisual industry. This higher level information is used in a variety of practices, such as enriching multimedia content with external links, clickable objects and useful related information in general. This paper presents a system for efficient multimedia content analysis and automatic annotation within a multimedia processing and publishing framework. This system is comprised of three modules: the first provides detection of faces and recognition of known persons; the second provides generic object detection, based on a deep convolutional neural network topology; the third provides automated location estimation and landmark recognition based on state-of-the-art technologies. The results are exported in meaningful metadata that can be utilized in various ways. The system has been developed and successfully tested in the framework of the EC Horizon 2020 Mecanex project, targeting advertising and production markets.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114054142","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7849985
Anthony Welte, L. Jaulin, M. Ceberio, V. Kreinovich
To properly process data, we need to take into account both the measurement errors and the fact that some of the observations may be outliers. This is especially important in radar-based localization problems, where some signals may reflect not from the analyzed object, but from some nearby object. There are known methods for dealing with both measurement errors and outliers in situations in which we have full information about the corresponding probability distributions. There are also known statistics-based methods for dealing with measurement errors in situations when we only have partial information about the corresponding probabilities. In this paper, we show how these methods can be extended to situations in which we also have partial information about the outliers (and even to situations when we have no information about the outliers). In some situations in which efficient semi-heuristic methods are known, our methodology leads to a justification of these efficient heuristics - which makes us confident that our new methods will be efficient in other situations as well.
{"title":"Robust data processing in the presence of uncertainty and outliers: Case of localization problems","authors":"Anthony Welte, L. Jaulin, M. Ceberio, V. Kreinovich","doi":"10.1109/SSCI.2016.7849985","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849985","url":null,"abstract":"To properly process data, we need to take into account both the measurement errors and the fact that some of the observations may be outliers. This is especially important in radar-based localization problems, where some signals may reflect not from the analyzed object, but from some nearby object. There are known methods for dealing with both measurement errors and outliers in situations in which we have full information about the corresponding probability distributions. There are also known statistics-based methods for dealing with measurement errors in situations when we only have partial information about the corresponding probabilities. In this paper, we show how these methods can be extended to situations in which we also have partial information about the outliers (and even to situations when we have no information about the outliers). In some situations in which efficient semi-heuristic methods are known, our methodology leads to a justification of these efficient heuristics - which makes us confident that our new methods will be efficient in other situations as well.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130820136","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7849366
A. Tijsma, Mădălina M. Drugan, M. Wiering
Balancing the ratio between exploration and exploitation is an important problem in reinforcement learning. This paper evaluates four different exploration strategies combined with Q-learning using random stochastic mazes to investigate their performances. We will compare: UCB-1, softmax, ∈-greedy, and pursuit. For this purpose we adapted the UCB-1 and pursuit strategies to be used in the Q-learning algorithm. The mazes consist of a single optimal goal state and two suboptimal goal states that lie closer to the starting position of the agent, which makes efficient exploration an important part of the learning agent. Furthermore, we evaluate two different kinds of reward functions, a normalized one with rewards between 0 and 1, and an unnormalized reward function that penalizes the agent for each step with a negative reward. We have performed an extensive grid-search to find the best parameters for each method and used the best parameters on novel randomly generated maze problems of different sizes. The results show that softmax exploration outperforms the other strategies, although it is harder to tune its temperature parameter. The worst performing exploration strategy is ∈-greedy.
平衡探索和利用之间的比例是强化学习中的一个重要问题。本文利用随机迷宫对四种结合Q-learning的探索策略进行评估，考察其性能。我们将比较:UCB-1, softmax，∈-greedy, and pursuit。为此，我们在Q-learning算法中采用了UCB-1和追击策略。迷宫由一个最优目标状态和两个靠近智能体起始位置的次优目标状态组成，这使得高效探索成为学习智能体的重要组成部分。此外，我们评估了两种不同类型的奖励函数，一种是奖励介于0和1之间的规范化奖励函数，另一种是奖励为负的非规范化奖励函数。我们进行了广泛的网格搜索，以找到每种方法的最佳参数，并将最佳参数用于不同大小的新型随机生成的迷宫问题。结果表明，softmax勘探策略优于其他策略，尽管其温度参数难以调整。表现最差的勘探策略是山贪。
{"title":"Comparing exploration strategies for Q-learning in random stochastic mazes","authors":"A. Tijsma, Mădălina M. Drugan, M. Wiering","doi":"10.1109/SSCI.2016.7849366","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849366","url":null,"abstract":"Balancing the ratio between exploration and exploitation is an important problem in reinforcement learning. This paper evaluates four different exploration strategies combined with Q-learning using random stochastic mazes to investigate their performances. We will compare: UCB-1, softmax, ∈-greedy, and pursuit. For this purpose we adapted the UCB-1 and pursuit strategies to be used in the Q-learning algorithm. The mazes consist of a single optimal goal state and two suboptimal goal states that lie closer to the starting position of the agent, which makes efficient exploration an important part of the learning agent. Furthermore, we evaluate two different kinds of reward functions, a normalized one with rewards between 0 and 1, and an unnormalized reward function that penalizes the agent for each step with a negative reward. We have performed an extensive grid-search to find the best parameters for each method and used the best parameters on novel randomly generated maze problems of different sizes. The results show that softmax exploration outperforms the other strategies, although it is harder to tune its temperature parameter. The worst performing exploration strategy is ∈-greedy.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128568415","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7850166
Milan Jelisavcic, Matteo De Carlo, E. Haasdijk, A. Eiben
This paper addresses the problem of on-line gait learning in modular robots whose shape is not known in advance. The best algorithm for this problem known to us is a reinforcement learning method, called RL PoWER. In this study we revisit the original RL PoWER algorithm and observe that in essence it is a specific evolutionary algorithm. Based on this insight we propose two modifications of the main search operators and compare the quality of the evolved gaits when either or both of these modified operators are employed. The results show that using 2-parent crossover as well as mutation with self-adaptive step-sizes can significantly improve the performance of the original algorithm.
{"title":"Improving RL power for on-line evolution of gaits in modular robots","authors":"Milan Jelisavcic, Matteo De Carlo, E. Haasdijk, A. Eiben","doi":"10.1109/SSCI.2016.7850166","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850166","url":null,"abstract":"This paper addresses the problem of on-line gait learning in modular robots whose shape is not known in advance. The best algorithm for this problem known to us is a reinforcement learning method, called RL PoWER. In this study we revisit the original RL PoWER algorithm and observe that in essence it is a specific evolutionary algorithm. Based on this insight we propose two modifications of the main search operators and compare the quality of the evolved gaits when either or both of these modified operators are employed. The results show that using 2-parent crossover as well as mutation with self-adaptive step-sizes can significantly improve the performance of the original algorithm.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114239960","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7850091
Rémi Delassus, R. Giot, Raphael Cherrier, Gabriele Barbieri, Guy Melançon
It seems necessary to detect a broken bike rooted at a station in near realtime as the number of bikes within bikeshare systems has reached more than a million in 2015. Indeed, a bike that cannot be moved is not cost effective in terms of number of trips. This brings frustration to users who were expecting to find a bike at that station without knowing that it is actually defective. We thus propose a methodology from feature extraction to anomaly detection on a distributed cloud infrastructure in order to detect bicycles requiring a repair. Through a first step of K-means clustering, and a second step consisting of spotting samples that do not clearly belong to any cluster, we separate anomalies from normal behaviors. The proposal is validated on a publicly available dataset provided by Motivate, the operator of the New-York bikeshare system. The number of distinct bikes that have been classified by this algorithm as broken at least once during a month is close to the number of repairs given in monthly reports of Motivate.
{"title":"Broken bikes detection using CitiBike bikeshare system open data","authors":"Rémi Delassus, R. Giot, Raphael Cherrier, Gabriele Barbieri, Guy Melançon","doi":"10.1109/SSCI.2016.7850091","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850091","url":null,"abstract":"It seems necessary to detect a broken bike rooted at a station in near realtime as the number of bikes within bikeshare systems has reached more than a million in 2015. Indeed, a bike that cannot be moved is not cost effective in terms of number of trips. This brings frustration to users who were expecting to find a bike at that station without knowing that it is actually defective. We thus propose a methodology from feature extraction to anomaly detection on a distributed cloud infrastructure in order to detect bicycles requiring a repair. Through a first step of K-means clustering, and a second step consisting of spotting samples that do not clearly belong to any cluster, we separate anomalies from normal behaviors. The proposal is validated on a publicly available dataset provided by Motivate, the operator of the New-York bikeshare system. The number of distinct bikes that have been classified by this algorithm as broken at least once during a month is close to the number of repairs given in monthly reports of Motivate.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120989163","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 : 2016-12-06DOI: 10.1109/SSCI.2016.7849368
Mathijs Pieters, M. Wiering
Most research in reinforcement learning has focused on stationary environments. In this paper, we propose several adaptations of Q-learning for a dynamic environment, for both single and multiple agents. The environment consists of a grid of random rewards, where every reward is removed after a visit. We focus on experience replay, a technique that receives a lot of attention nowadays, and combine this method with Q-learning. We compare two variations of experience replay, where experiences are reused based on time or based on the obtained reward. For multi-agent reinforcement learning we compare two variations of policy representation. In the first variation the agents share a Q-function, while in the second variation both agents have a separate Q-function. Furthermore, in both variations we test the effect of reward sharing between the agents. This leads to four different multi-agent reinforcement learning algorithms, from which sharing a Q-function and sharing the rewards is the most cooperative method. The results show that in the single-agent environment both experience replay algorithms significantly outperform standard Q-learning and a greedy benchmark agent. In the multi-agent environment the highest maximum reward sum in a trial is achieved by using one Q-function and reward sharing. The highest mean reward sum is obtained with separate Q-functions and separate rewards.
{"title":"Q-learning with experience replay in a dynamic environment","authors":"Mathijs Pieters, M. Wiering","doi":"10.1109/SSCI.2016.7849368","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849368","url":null,"abstract":"Most research in reinforcement learning has focused on stationary environments. In this paper, we propose several adaptations of Q-learning for a dynamic environment, for both single and multiple agents. The environment consists of a grid of random rewards, where every reward is removed after a visit. We focus on experience replay, a technique that receives a lot of attention nowadays, and combine this method with Q-learning. We compare two variations of experience replay, where experiences are reused based on time or based on the obtained reward. For multi-agent reinforcement learning we compare two variations of policy representation. In the first variation the agents share a Q-function, while in the second variation both agents have a separate Q-function. Furthermore, in both variations we test the effect of reward sharing between the agents. This leads to four different multi-agent reinforcement learning algorithms, from which sharing a Q-function and sharing the rewards is the most cooperative method. The results show that in the single-agent environment both experience replay algorithms significantly outperform standard Q-learning and a greedy benchmark agent. In the multi-agent environment the highest maximum reward sum in a trial is achieved by using one Q-function and reward sharing. The highest mean reward sum is obtained with separate Q-functions and separate rewards.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128600951","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7849914
Ananth A. Jillepalli, D. Leon, Stuart Steiner, Frederick T. Sheldon
In this article, we describe the characteristics, structure, and uses of HERMES. HERMES is a high-level security policy description language. Its characteristics are: (1) enable the specification of organizational domain knowledge in a hierarchical manner; (2) enable the specification of security policies at desired granularity levels within the organizational IT and OT infrastructure; (3) enable security policies to be automatically instantiated into security configurations; (4) it is human-centered and designed for ease of use; (5) it is application and device independent. We show an example of using HERMES to write a high-level policy and show examples of how such policy can be instantiated into a domain and device, user and role, application and action specific security configuration. We also describe the integration of HERMES within the HiFiPol:Browser policy management system. We believe HERMES is a necessary step toward securing the client side of the web ecosystem and prevent or mitigate the current onslaught of web browser-based attacks, such as phishing.
{"title":"HERMES: A high-level policy language for high-granularity enterprise-wide secure browser configuration management","authors":"Ananth A. Jillepalli, D. Leon, Stuart Steiner, Frederick T. Sheldon","doi":"10.1109/SSCI.2016.7849914","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849914","url":null,"abstract":"In this article, we describe the characteristics, structure, and uses of HERMES. HERMES is a high-level security policy description language. Its characteristics are: (1) enable the specification of organizational domain knowledge in a hierarchical manner; (2) enable the specification of security policies at desired granularity levels within the organizational IT and OT infrastructure; (3) enable security policies to be automatically instantiated into security configurations; (4) it is human-centered and designed for ease of use; (5) it is application and device independent. We show an example of using HERMES to write a high-level policy and show examples of how such policy can be instantiated into a domain and device, user and role, application and action specific security configuration. We also describe the integration of HERMES within the HiFiPol:Browser policy management system. We believe HERMES is a necessary step toward securing the client side of the web ecosystem and prevent or mitigate the current onslaught of web browser-based attacks, such as phishing.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121040772","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7849963
Shaaban A. Sahmoud, H. Topcuoglu
Detecting changes in a landscape is a critical issue for several evolutionary dynamic optimization techniques. This is because most of these techniques have steps to be taken as a response to the environmental changes. It may not be feasible for most of the real world problems to know a priori when a change occurs; therefore, explicit schemes should be proposed to detect the points in time when a change occurs. Although there are both sensor-based and population-based detection schemes presented in the literature for single objective dynamic optimization problems, there are no such efforts for the dynamic multi-objective optimization problems (DMOPs). This paper proposes a set of novel sensor-based change detection schemes for DMOPs. An empirical study is presented for validating the performance of the proposed detection schemes by using eight test problems which have different types and characteristics. Additionally, the proposed change detection schemes are incorporated into a dynamic multi-objective evolutionary algorithm to validate the effectiveness of the proposed change detection schemes.
{"title":"Sensor-based change detection schemes for dynamic multi-objective optimization problems","authors":"Shaaban A. Sahmoud, H. Topcuoglu","doi":"10.1109/SSCI.2016.7849963","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849963","url":null,"abstract":"Detecting changes in a landscape is a critical issue for several evolutionary dynamic optimization techniques. This is because most of these techniques have steps to be taken as a response to the environmental changes. It may not be feasible for most of the real world problems to know a priori when a change occurs; therefore, explicit schemes should be proposed to detect the points in time when a change occurs. Although there are both sensor-based and population-based detection schemes presented in the literature for single objective dynamic optimization problems, there are no such efforts for the dynamic multi-objective optimization problems (DMOPs). This paper proposes a set of novel sensor-based change detection schemes for DMOPs. An empirical study is presented for validating the performance of the proposed detection schemes by using eight test problems which have different types and characteristics. Additionally, the proposed change detection schemes are incorporated into a dynamic multi-objective evolutionary algorithm to validate the effectiveness of the proposed change detection schemes.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125022630","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850194
N. Naik
Multi-cloud systems have been gaining popularity due to the several benefits of the multi-cloud infrastructure such as lower level of vendor lock-in and minimize the risk of widespread data loss or downtime. Thus, the multi-cloud infrastructure enhances the dependability of the cloud-based system. However, it also poses many challenges such as nonstandard and inherent complexity due to different technologies, interfaces, and services. Consequently, it is a challenging task to design multi-cloud dependable systems. Virtualization is the key technology employed in the development of cloud-based systems. Docker has recently introduced its container-based virtualization technology for the development of software systems. It has newly launched a distributed system development tool called Swarm, which allows the development of a cluster of multiple Swarm nodes on multiple clouds. Docker Swarm has also incorporated several dependability attributes to support the development of a multi-cloud dependable system. However, making Swarm cluster always available requires minimum three active manager nodes which can safeguard one failure. This essential condition for the dependability is one of the main limitations because if two manager nodes fail suddenly due to the failure of their hosts, then Swarm cluster cannot be made available for routine operations. Therefore, this paper proposes an intuitive approach based on Computational Intelligence (CI) for enhancing its dependability. The proposed CI-based approach predicts the possible failure of the host of a manager node by observing its abnormal behaviour. Thus, this indication can automatically trigger the process of creating a new manager node or promoting an existing node as a manager for enhancing the dependability of Docker Swarm.
{"title":"Applying Computational Intelligence for enhancing the dependability of multi-cloud systems using Docker Swarm","authors":"N. Naik","doi":"10.1109/SSCI.2016.7850194","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850194","url":null,"abstract":"Multi-cloud systems have been gaining popularity due to the several benefits of the multi-cloud infrastructure such as lower level of vendor lock-in and minimize the risk of widespread data loss or downtime. Thus, the multi-cloud infrastructure enhances the dependability of the cloud-based system. However, it also poses many challenges such as nonstandard and inherent complexity due to different technologies, interfaces, and services. Consequently, it is a challenging task to design multi-cloud dependable systems. Virtualization is the key technology employed in the development of cloud-based systems. Docker has recently introduced its container-based virtualization technology for the development of software systems. It has newly launched a distributed system development tool called Swarm, which allows the development of a cluster of multiple Swarm nodes on multiple clouds. Docker Swarm has also incorporated several dependability attributes to support the development of a multi-cloud dependable system. However, making Swarm cluster always available requires minimum three active manager nodes which can safeguard one failure. This essential condition for the dependability is one of the main limitations because if two manager nodes fail suddenly due to the failure of their hosts, then Swarm cluster cannot be made available for routine operations. Therefore, this paper proposes an intuitive approach based on Computational Intelligence (CI) for enhancing its dependability. The proposed CI-based approach predicts the possible failure of the host of a manager node by observing its abnormal behaviour. Thus, this indication can automatically trigger the process of creating a new manager node or promoting an existing node as a manager for enhancing the dependability of Docker Swarm.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125741643","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：