Pub Date : 2010-06-21DOI: 10.1109/AIS.2010.5547036
Yun-Qian Miao, A. Khamis, M. Kamel
This paper discusses the role of three mobility models namely, fully coordinated mobility, fully random mobility and emergent mobility models in improving area coverage and detection effectiveness of a set of mobile sensors in a mobile surveillance system. A novel anti-flocking algorithm that mimics solitary animal's social behavior is described. A multiagent-based system has been implemented to examine the efficiency of the different mobility model. The simulation results show that anti-flocking mobility model outperforms the fully random mobility model. This novel model provides additional features such as scalability, robustness and adaptivity comparing with fully coordinated mobility model.
{"title":"Applying anti-flocking model in mobile surveillance systems","authors":"Yun-Qian Miao, A. Khamis, M. Kamel","doi":"10.1109/AIS.2010.5547036","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547036","url":null,"abstract":"This paper discusses the role of three mobility models namely, fully coordinated mobility, fully random mobility and emergent mobility models in improving area coverage and detection effectiveness of a set of mobile sensors in a mobile surveillance system. A novel anti-flocking algorithm that mimics solitary animal's social behavior is described. A multiagent-based system has been implemented to examine the efficiency of the different mobility model. The simulation results show that anti-flocking mobility model outperforms the fully random mobility model. This novel model provides additional features such as scalability, robustness and adaptivity comparing with fully coordinated mobility model.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90546289","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547034
A. Saleem, M. Lind, S. N. Singh
Increased interconnection and loading of the power system along with deregulation has brought new challenges for electric power system operation, control and automation. Traditional power system models used in intelligent operation and control are highly dependent on the task purpose. Thus, a model for intelligent operation and control must represent system features, so that information from measurements can be related to possible system states and to control actions. These general modeling requirements are well understood, but it is, in general, difficult to translate them into a model because of the lack of explicit principles for model construction. This paper presents research using explicit means-ends model based reasoning about complex control situations for maintaining consistent perspectives and selecting appropriate control action for goal driven agents. An example of power system operation and control is described using the modeling approach.
{"title":"Modeling control situations in power system operations","authors":"A. Saleem, M. Lind, S. N. Singh","doi":"10.1109/AIS.2010.5547034","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547034","url":null,"abstract":"Increased interconnection and loading of the power system along with deregulation has brought new challenges for electric power system operation, control and automation. Traditional power system models used in intelligent operation and control are highly dependent on the task purpose. Thus, a model for intelligent operation and control must represent system features, so that information from measurements can be related to possible system states and to control actions. These general modeling requirements are well understood, but it is, in general, difficult to translate them into a model because of the lack of explicit principles for model construction. This paper presents research using explicit means-ends model based reasoning about complex control situations for maintaining consistent perspectives and selecting appropriate control action for goal driven agents. An example of power system operation and control is described using the modeling approach.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84444781","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547043
A. Benaskeur, A. Khamis, H. Irandoust
This paper discusses distributed surveillance problems, where a set of sensors, of different modalities, can sense collaboratively and continuously a certain volume of interest. Surveillance operations in complex environments, such as littoral regions, are introduced and their main features and challenges are presented. Effective cooperation among the sensors can synergistically improve the performance of these systems and can endow them with higher-level faculties, such as dynamic task allocation, communication relaying, and cooperative target search and tracking. Different forms of cooperation in distributed surveillance systems are mentioned. The paper focuses on intra-and inter-platform target cueing and handoff as augmentative forms of cooperation in distributed surveillance.
{"title":"Cooperation in distributed surveillance","authors":"A. Benaskeur, A. Khamis, H. Irandoust","doi":"10.1109/AIS.2010.5547043","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547043","url":null,"abstract":"This paper discusses distributed surveillance problems, where a set of sensors, of different modalities, can sense collaboratively and continuously a certain volume of interest. Surveillance operations in complex environments, such as littoral regions, are introduced and their main features and challenges are presented. Effective cooperation among the sensors can synergistically improve the performance of these systems and can endow them with higher-level faculties, such as dynamic task allocation, communication relaying, and cooperative target search and tracking. Different forms of cooperation in distributed surveillance systems are mentioned. The paper focuses on intra-and inter-platform target cueing and handoff as augmentative forms of cooperation in distributed surveillance.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76595330","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547026
Ahmed Salah El-Din, M. Elsayed, A. Alsebai, N. E. Gayar, M. Elhelw
Visual Sensor Networks (VSNs) open up a new realm of smart autonomous applications based on enhanced three-dimensional sensing and collaborative reasoning. An emerging VSN application domain is pervasive healthcare delivery where gait information computed from distributed vision nodes is used for observing the wellbeing of the elderly, quantifying post-operative patient recovery and monitoring the progression of neurodegenerative diseases such as Parkinson's. The development of patient-specific gait analysis models, however, is challenging since it is unfeasible to obtain normal and impaired gait examples from the same patient before the operation in order to build supervised models for gait classification. This paper presents a novel VSN-based framework for quantification of patient-specific gait impairment and post-operative recovery by using change analysis. Real-time target extraction is first applied to VSN data and a skeletonization procedure is subsequently carried out to quantify the internal motion of moving target and compute two features; spatiotemporal cyclic motion between leg segments and head trajectory for each vision node. Change analysis is then used to measure the change, i.e. difference, between two unlabeled datasets collected pre- and post-operatively and quantify gait changes. The potential value of the proposed framework for patient gait monitoring is demonstrated and the results obtained from practical experiments are described.
{"title":"Change analysis for gait impairment quantification in smart environments","authors":"Ahmed Salah El-Din, M. Elsayed, A. Alsebai, N. E. Gayar, M. Elhelw","doi":"10.1109/AIS.2010.5547026","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547026","url":null,"abstract":"Visual Sensor Networks (VSNs) open up a new realm of smart autonomous applications based on enhanced three-dimensional sensing and collaborative reasoning. An emerging VSN application domain is pervasive healthcare delivery where gait information computed from distributed vision nodes is used for observing the wellbeing of the elderly, quantifying post-operative patient recovery and monitoring the progression of neurodegenerative diseases such as Parkinson's. The development of patient-specific gait analysis models, however, is challenging since it is unfeasible to obtain normal and impaired gait examples from the same patient before the operation in order to build supervised models for gait classification. This paper presents a novel VSN-based framework for quantification of patient-specific gait impairment and post-operative recovery by using change analysis. Real-time target extraction is first applied to VSN data and a skeletonization procedure is subsequently carried out to quantify the internal motion of moving target and compute two features; spatiotemporal cyclic motion between leg segments and head trajectory for each vision node. Change analysis is then used to measure the change, i.e. difference, between two unlabeled datasets collected pre- and post-operatively and quantify gait changes. The potential value of the proposed framework for patient gait monitoring is demonstrated and the results obtained from practical experiments are described.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78263480","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547047
B. Beckman, M. Trentini, J. Pieper
The requirement for increased mobility of unmanned ground vehicles (UGVs) operating in urban settings must be addressed if robotic technology is to augment human efforts in military relevant roles and environments. In preparation for this role, Defence R&D Canada - Suffield is exploring novel mobility platforms that use intelligent mobility algorithms to improve robot mobility in unknown highly complex terrain. Robotic platforms often appear conceptually simple. Despite this appearance, the demands on these systems remain extremely ambitious while retaining the need for control systems to handle the many actuator degrees-of-freedom and numerous sensor inputs. Linear control techniques applied to a nonlinear multi degree-of-freedom vehicles are effective in controlling system behaviours in limited conditions. However, in unrestricted conditions, the nonlinear nature of the control problem and impracticality of model-based control of such a complex system have required the investigation of alternative control methods. This paper discusses linear control techniques applied to a multi degree-of-freedom robot in simulation and alternative nonlinear techniques.
{"title":"Control algorithms for stable range-of-motion behaviours of a multi degree-of-freedom robot","authors":"B. Beckman, M. Trentini, J. Pieper","doi":"10.1109/AIS.2010.5547047","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547047","url":null,"abstract":"The requirement for increased mobility of unmanned ground vehicles (UGVs) operating in urban settings must be addressed if robotic technology is to augment human efforts in military relevant roles and environments. In preparation for this role, Defence R&D Canada - Suffield is exploring novel mobility platforms that use intelligent mobility algorithms to improve robot mobility in unknown highly complex terrain. Robotic platforms often appear conceptually simple. Despite this appearance, the demands on these systems remain extremely ambitious while retaining the need for control systems to handle the many actuator degrees-of-freedom and numerous sensor inputs. Linear control techniques applied to a nonlinear multi degree-of-freedom vehicles are effective in controlling system behaviours in limited conditions. However, in unrestricted conditions, the nonlinear nature of the control problem and impracticality of model-based control of such a complex system have required the investigation of alternative control methods. This paper discusses linear control techniques applied to a multi degree-of-freedom robot in simulation and alternative nonlinear techniques.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77275655","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547024
P. Dias, J. Pinto, R. Gonçalves, J. Sousa
Operation of networks of heterogeneous vehicles and sensors imposes many technical and operational challenges. Simultaneous control of multiple vehicle types requires abstraction any device-specific details and to keep human operators in the loop by providing them a good overall picture of the current system state. In this paper, we present the Neptus command and control infrastructure for such operations, in terms of its evolution and current-day architecture. Neptus supports the various phases of vehicle and sensor operations abstracting vehicle and sensor specificities by considering vehicles as maneuver providers and using open standards for communication and data storage. Operators are kept in the loop by using adaptable interfaces which can be tailored to specific operators, vehicles or mission scenarios. Neptus has been used numerous times for field-testing unmanned vehicles and in several demonstrations of multi-vehicle operations.
{"title":"Enabling a dialog – A C2I infrastructure for unmanned vehicles and sensors","authors":"P. Dias, J. Pinto, R. Gonçalves, J. Sousa","doi":"10.1109/AIS.2010.5547024","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547024","url":null,"abstract":"Operation of networks of heterogeneous vehicles and sensors imposes many technical and operational challenges. Simultaneous control of multiple vehicle types requires abstraction any device-specific details and to keep human operators in the loop by providing them a good overall picture of the current system state. In this paper, we present the Neptus command and control infrastructure for such operations, in terms of its evolution and current-day architecture. Neptus supports the various phases of vehicle and sensor operations abstracting vehicle and sensor specificities by considering vehicles as maneuver providers and using open standards for communication and data storage. Operators are kept in the loop by using adaptable interfaces which can be tailored to specific operators, vehicles or mission scenarios. Neptus has been used numerous times for field-testing unmanned vehicles and in several demonstrations of multi-vehicle operations.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80845074","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547017
S. Abdel-Mageid, R. Ramadan
Sensor deployment problem is one of the important problems in Wireless Sensor Networks (WSN) since it represents the first phase that most of the network operations depends on. Sensor deployment strategies can be classified into two classes which are deterministic and autonomous (random) deployment. In the deterministic deployment, the deployment field is assumed accessible as well as the number of sensors is small to be manually deployed in specific locations. On the other hand, with large number of sensors and in inaccessible fields, the random deployment to the sensors turns out to be the solution. However, random deployment requires sensors to be automatically located (move) for coverage and connectivity purposes. In addition, after a period of time, the sensors topology might change due to some sensor hardware failure or deplaned energy. Therefore, redeployment and/or sensors relocation process is essential. Nevertheless, mobility consumed energy as well as sensor load balancing are essential factors to be considered during the initial deployment and relocation processes. This paper proposes two deployment algorithms to manage those situations. Those algorithms achieve sensor energy balancing and small amount of deployment energy consumption. A set of simulation experiments are conducted to compare between the proposed algorithm and the existing work in terms of coverage performance, average moving distance, and message complexity.
{"title":"Efficient deployment algorithms for mobile sensor networks","authors":"S. Abdel-Mageid, R. Ramadan","doi":"10.1109/AIS.2010.5547017","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547017","url":null,"abstract":"Sensor deployment problem is one of the important problems in Wireless Sensor Networks (WSN) since it represents the first phase that most of the network operations depends on. Sensor deployment strategies can be classified into two classes which are deterministic and autonomous (random) deployment. In the deterministic deployment, the deployment field is assumed accessible as well as the number of sensors is small to be manually deployed in specific locations. On the other hand, with large number of sensors and in inaccessible fields, the random deployment to the sensors turns out to be the solution. However, random deployment requires sensors to be automatically located (move) for coverage and connectivity purposes. In addition, after a period of time, the sensors topology might change due to some sensor hardware failure or deplaned energy. Therefore, redeployment and/or sensors relocation process is essential. Nevertheless, mobility consumed energy as well as sensor load balancing are essential factors to be considered during the initial deployment and relocation processes. This paper proposes two deployment algorithms to manage those situations. Those algorithms achieve sensor energy balancing and small amount of deployment energy consumption. A set of simulation experiments are conducted to compare between the proposed algorithm and the existing work in terms of coverage performance, average moving distance, and message complexity.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78721601","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547044
M. Chitre
Autonomous underwater vehicles (AUVs) that rely on dead reckoning suffer from unbounded localization error growth at a rate dependent on the quality (and cost) of the navigational sensors. Many AUVs surface occasionally to get a GPS position update. Alternatively underwater acoustic beacons such as long baseline (LBL) arrays are used for localization, at the cost of substantial deployment effort. The idea of cooperative localization with a few vehicles with high navigation accuracy (beacon vehicles) among a team of AUVs with poor navigational sensors has recently gained interest. Autonomous surface crafts (ASCs) with GPS, or sophisticated AUVs with expensive navigational sensors may play the role of beacon vehicles. Other AUVs are able to measure their range to these acoustically, and use the resulting information for self-localization. Since a single range measurement is insufficient for unambiguous localization, multiple beacon vehicles are usually required. In this paper, we explore the use of a single beacon vehicle to support multiple AUVs. We develop path planning algorithms for the beacon vehicle that take into account and minimize the errors being accumulated by other AUVs. We show that the generated beacon vehicle path enables the other AUVs to get sufficient information to keep their localization errors bounded over time.
{"title":"Path planning for cooperative underwater range-only navigation using a single beacon","authors":"M. Chitre","doi":"10.1109/AIS.2010.5547044","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547044","url":null,"abstract":"Autonomous underwater vehicles (AUVs) that rely on dead reckoning suffer from unbounded localization error growth at a rate dependent on the quality (and cost) of the navigational sensors. Many AUVs surface occasionally to get a GPS position update. Alternatively underwater acoustic beacons such as long baseline (LBL) arrays are used for localization, at the cost of substantial deployment effort. The idea of cooperative localization with a few vehicles with high navigation accuracy (beacon vehicles) among a team of AUVs with poor navigational sensors has recently gained interest. Autonomous surface crafts (ASCs) with GPS, or sophisticated AUVs with expensive navigational sensors may play the role of beacon vehicles. Other AUVs are able to measure their range to these acoustically, and use the resulting information for self-localization. Since a single range measurement is insufficient for unambiguous localization, multiple beacon vehicles are usually required. In this paper, we explore the use of a single beacon vehicle to support multiple AUVs. We develop path planning algorithms for the beacon vehicle that take into account and minimize the errors being accumulated by other AUVs. We show that the generated beacon vehicle path enables the other AUVs to get sufficient information to keep their localization errors bounded over time.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74191881","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547077
J. Cruz, D. Kulić, W. Owen
High performance control of robotic systems, including the new generation of humanoid, assistive and entertainment robots, requires adequate knowledge of the dynamics of the system. This can be problematic in the presence of modeling uncertainties as the performance of classical, modelbased controllers is highly dependant upon accurate knowledge of the system. In addition, future robotic systems such as humanoids are likely to be redundant, requiring a mechanism for redundancy resolution when performing lower degree-of-freedom tasks. In this paper, a learning approach to estimating the inverse dynamic equations is presented. Locally Weighted Projection Regression (LWPR) is used to learn the inverse dynamics of a manipulator in both joint and task space and the resulting controllers are used to drive a 3 and 4 DOF robot in simulation. The performance of the learning controllers is compared to a traditional model based control method and is also shown to be a viable control method for a redundant system.
{"title":"Learning inverse dynamics for redundant manipulator control","authors":"J. Cruz, D. Kulić, W. Owen","doi":"10.1109/AIS.2010.5547077","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547077","url":null,"abstract":"High performance control of robotic systems, including the new generation of humanoid, assistive and entertainment robots, requires adequate knowledge of the dynamics of the system. This can be problematic in the presence of modeling uncertainties as the performance of classical, modelbased controllers is highly dependant upon accurate knowledge of the system. In addition, future robotic systems such as humanoids are likely to be redundant, requiring a mechanism for redundancy resolution when performing lower degree-of-freedom tasks. In this paper, a learning approach to estimating the inverse dynamic equations is presented. Locally Weighted Projection Regression (LWPR) is used to learn the inverse dynamics of a manipulator in both joint and task space and the resulting controllers are used to drive a 3 and 4 DOF robot in simulation. The performance of the learning controllers is compared to a traditional model based control method and is also shown to be a viable control method for a redundant system.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73076754","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 : 2010-06-21DOI: 10.1109/AIS.2010.5547038
K. Abida, F. Karray
We present a review of the most significant advances in the field of system combination towards reducing word error rates (WER) in large vocabulary continuous speech recognition (LVCSR). We have mainly focused on Recognizer Output Voting Error Reduction (ROVER), confusion network (CN) and minimum frame word error rate (fWER) based combination along with the latest improvements. Despite lot of progress witnessed in this field, some challenges still remain in enhancing the performance of LVCSR. We suggest in this paper some directions that may lead to a lower WER within the framework of system combination.
{"title":"Systems combination in large vocabulary continuous speech recognition","authors":"K. Abida, F. Karray","doi":"10.1109/AIS.2010.5547038","DOIUrl":"https://doi.org/10.1109/AIS.2010.5547038","url":null,"abstract":"We present a review of the most significant advances in the field of system combination towards reducing word error rates (WER) in large vocabulary continuous speech recognition (LVCSR). We have mainly focused on Recognizer Output Voting Error Reduction (ROVER), confusion network (CN) and minimum frame word error rate (fWER) based combination along with the latest improvements. Despite lot of progress witnessed in this field, some challenges still remain in enhancing the performance of LVCSR. We suggest in this paper some directions that may lead to a lower WER within the framework of system combination.","PeriodicalId":71187,"journal":{"name":"自主智能系统(英文)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2010-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83960846","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}