The Experts below are selected from a list of 360 Experts worldwide ranked by ideXlab platform
Phengann Heng - One of the best experts on this subject based on the ideXlab platform.
-
dual teacher exploiting intra Domain and inter Domain Knowledge with reliable transfer for cardiac segmentation
arXiv: Image and Video Processing, 2021Co-Authors: Shujun Wang, Phengann HengAbstract:Annotation scarcity is a long-standing problem in medical image analysis area. To efficiently leverage limited annotations, abundant unlabeled data are additionally exploited in semi-supervised learning, while well-established cross-modality data are investigated in Domain adaptation. In this paper, we aim to explore the feasibility of concurrently leveraging both unlabeled data and cross-modality data for annotation-efficient cardiac segmentation. To this end, we propose a cutting-edge semi-supervised Domain adaptation framework, namely Dual-Teacher++. Besides directly learning from limited labeled target Domain data (e.g., CT) via a student model adopted by previous literature, we design novel dual teacher models, including an inter-Domain teacher model to explore cross-modality priors from source Domain (e.g., MR) and an intra-Domain teacher model to investigate the Knowledge beneath unlabeled target Domain. In this way, the dual teacher models would transfer acquired inter- and intra-Domain Knowledge to the student model for further integration and exploitation. Moreover, to encourage reliable dual-Domain Knowledge transfer, we enhance the inter-Domain Knowledge transfer on the samples with higher similarity to target Domain after appearance alignment, and also strengthen intra-Domain Knowledge transfer of unlabeled target data with higher prediction confidence. In this way, the student model can obtain reliable dual-Domain Knowledge and yield improved performance on target Domain data. We extensively evaluated the feasibility of our method on the MM-WHS 2017 challenge dataset. The experiments have demonstrated the superiority of our framework over other semi-supervised learning and Domain adaptation methods. Moreover, our performance gains could be yielded in bidirections,i.e., adapting from MR to CT, and from CT to MR.
-
dual teacher exploiting intra Domain and inter Domain Knowledge with reliable transfer for cardiac segmentation
IEEE Transactions on Medical Imaging, 2020Co-Authors: Shujun Wang, Phengann HengAbstract:Annotation scarcity is a long-standing problem in medical image analysis area. To efficiently leverage limited annotations, abundant unlabeled data are additionally exploited in semi-supervised learning, while well-established cross-modality data are investigated in Domain adaptation. In this paper, we aim to explore the feasibility of concurrently leveraging both unlabeled data and cross-modality data for annotation-efficient cardiac segmentation. To this end, we propose a cutting-edge semi-supervised Domain adaptation framework, namely Dual-Teacher++. Besides directly learning from limited labeled target Domain data (e.g., CT) via a student model adopted by previous literature, we design novel dual teacher models, including an inter-Domain teacher model to explore cross-modality priors from source Domain (e.g., MR) and an intra-Domain teacher model to investigate the Knowledge beneath unlabeled target Domain. In this way, the dual teacher models would transfer acquired inter- and intra-Domain Knowledge to the student model for further integration and exploitation. Moreover, to encourage reliable dual-Domain Knowledge transfer, we enhance the inter-Domain Knowledge transfer on the samples with higher similarity to target Domain after appearance alignment, and also strengthen intra-Domain Knowledge transfer of unlabeled target data with higher prediction confidence. In this way, the student model can obtain reliable dual-Domain Knowledge and yield improved performance on target Domain data. We extensively evaluated the feasibility of our method on the MM-WHS 2017 challenge dataset. The experiments have demonstrated the superiority of our framework over other semi-supervised learning and Domain adaptation methods. Moreover, our performance gains could be yielded in bidirections, i.e., adapting from MR to CT, and from CT to MR. Our code will be available at https://github.com/kli-lalala/Dual-Teacher-.
Maarten Van Someren - One of the best experts on this subject based on the ideXlab platform.
-
machine learning for vessel trajectories using compression alignments and Domain Knowledge
Expert Systems With Applications, 2012Co-Authors: Gerben Klaas Dirk De Vries, Maarten Van SomerenAbstract:In this paper we present a machine learning framework to analyze moving object trajectories from maritime vessels. Within this framework we perform the tasks of clustering, classification and outlier detection with vessel trajectory data. First, we apply a piecewise linear segmentation method to the trajectories to compress them. We adapt an existing technique to better retain stop and move information and show the better performance of our method with experimental results. Second, we use a similarity based approach to perform the clustering, classification and outlier detection tasks using kernel methods. We present experiments that investigate different alignment kernels and the effect of piecewise linear segmentation in the three different tasks. The experimental results show that compression does not negatively impact task performance and greatly reduces computation time for the alignment kernels. Finally, the alignment kernels allow for easy integration of geographical Domain Knowledge. In experiments we show that this added Domain Knowledge enhances performance in the clustering and classification tasks.
-
comparing vessel trajectories using geographical Domain Knowledge and alignments
International Conference on Data Mining, 2010Co-Authors: Gerben Klaas Dirk De Vries, Willem Robert Van Hage, Maarten Van SomerenAbstract:This paper presents a similarity measure that combines low-level trajectory information with geographical Domain Knowledge to compare vessel trajectories. The similarity measure is largely based on alignment techniques. In a clustering experiment we show how the measure can be used to discover behavior concepts in vessel trajectory data that are dependent both on the low-level trajectories and the Domain Knowledge. We also apply this measure in a classification task to predict the type of vessel. In this task the combined measure performs better than similarities based on Domain Knowledge or low-level information alone.
Motoshi Saeki - One of the best experts on this subject based on the ideXlab platform.
-
RE - Using Domain Ontology as Domain Knowledge for Requirements Elicitation
14th IEEE International Requirements Engineering Conference (RE'06), 2006Co-Authors: Haruhiko Kaiya, Motoshi SaekiAbstract:Domain Knowledge is one of crucial factors to get a great success in requirements elicitation of high quality, and only Domain experts, not requirements analysts, have it. We propose a new requirements elicitation method ORE (Ontology based Requirements Elicitation), where a Domain ontology can be used as Domain Knowledge. In our method, a Domain ontology plays a role on semantic Domain which gives meanings to requirements statements by using a semantic function. By using inference rules on the ontology and a quality metrics on the semantic function, an analyst can be navigated which requirements should be added for improving completeness of the current version of the requirements and/or which requirements should be deleted from the current version for keeping consistency. We define this process as a method and evaluate it by an experimental case study of software music players.
Shujun Wang - One of the best experts on this subject based on the ideXlab platform.
-
dual teacher exploiting intra Domain and inter Domain Knowledge with reliable transfer for cardiac segmentation
arXiv: Image and Video Processing, 2021Co-Authors: Shujun Wang, Phengann HengAbstract:Annotation scarcity is a long-standing problem in medical image analysis area. To efficiently leverage limited annotations, abundant unlabeled data are additionally exploited in semi-supervised learning, while well-established cross-modality data are investigated in Domain adaptation. In this paper, we aim to explore the feasibility of concurrently leveraging both unlabeled data and cross-modality data for annotation-efficient cardiac segmentation. To this end, we propose a cutting-edge semi-supervised Domain adaptation framework, namely Dual-Teacher++. Besides directly learning from limited labeled target Domain data (e.g., CT) via a student model adopted by previous literature, we design novel dual teacher models, including an inter-Domain teacher model to explore cross-modality priors from source Domain (e.g., MR) and an intra-Domain teacher model to investigate the Knowledge beneath unlabeled target Domain. In this way, the dual teacher models would transfer acquired inter- and intra-Domain Knowledge to the student model for further integration and exploitation. Moreover, to encourage reliable dual-Domain Knowledge transfer, we enhance the inter-Domain Knowledge transfer on the samples with higher similarity to target Domain after appearance alignment, and also strengthen intra-Domain Knowledge transfer of unlabeled target data with higher prediction confidence. In this way, the student model can obtain reliable dual-Domain Knowledge and yield improved performance on target Domain data. We extensively evaluated the feasibility of our method on the MM-WHS 2017 challenge dataset. The experiments have demonstrated the superiority of our framework over other semi-supervised learning and Domain adaptation methods. Moreover, our performance gains could be yielded in bidirections,i.e., adapting from MR to CT, and from CT to MR.
-
dual teacher exploiting intra Domain and inter Domain Knowledge with reliable transfer for cardiac segmentation
IEEE Transactions on Medical Imaging, 2020Co-Authors: Shujun Wang, Phengann HengAbstract:Annotation scarcity is a long-standing problem in medical image analysis area. To efficiently leverage limited annotations, abundant unlabeled data are additionally exploited in semi-supervised learning, while well-established cross-modality data are investigated in Domain adaptation. In this paper, we aim to explore the feasibility of concurrently leveraging both unlabeled data and cross-modality data for annotation-efficient cardiac segmentation. To this end, we propose a cutting-edge semi-supervised Domain adaptation framework, namely Dual-Teacher++. Besides directly learning from limited labeled target Domain data (e.g., CT) via a student model adopted by previous literature, we design novel dual teacher models, including an inter-Domain teacher model to explore cross-modality priors from source Domain (e.g., MR) and an intra-Domain teacher model to investigate the Knowledge beneath unlabeled target Domain. In this way, the dual teacher models would transfer acquired inter- and intra-Domain Knowledge to the student model for further integration and exploitation. Moreover, to encourage reliable dual-Domain Knowledge transfer, we enhance the inter-Domain Knowledge transfer on the samples with higher similarity to target Domain after appearance alignment, and also strengthen intra-Domain Knowledge transfer of unlabeled target data with higher prediction confidence. In this way, the student model can obtain reliable dual-Domain Knowledge and yield improved performance on target Domain data. We extensively evaluated the feasibility of our method on the MM-WHS 2017 challenge dataset. The experiments have demonstrated the superiority of our framework over other semi-supervised learning and Domain adaptation methods. Moreover, our performance gains could be yielded in bidirections, i.e., adapting from MR to CT, and from CT to MR. Our code will be available at https://github.com/kli-lalala/Dual-Teacher-.
Iris Vessey - One of the best experts on this subject based on the ideXlab platform.
-
understanding the role of is and application Domain Knowledge on conceptual schema problem solving a verbal protocol study
Journal of the Association for Information Systems, 2016Co-Authors: Vijay Khatri, Iris VesseyAbstract:One of the most neglected areas of information systems research is the role of the Domain to which researchers apply IS methods, tools, and techniques; that is, the application Domain. For example, little prior information systems (IS) or related research has examined how IS and application Domain Knowledge (ISDK and ADK, respectively) influence how individuals solve conceptual schema problem-solving tasks. In this research, we investigate the effects of both ISDK and ADK on two types of conceptual schema problem-solving tasks: schema based and inferential. We used verbal protocol analysis to explore the roles that ISDK and ADK play in the problem-solving processes participants use when addressing these tasks. We found that, for the two types of conceptual schema problem-solving tasks, ADK and ISDK have similar effects on problem-solving processes. That is, we found that, for schema-based problem-solving tasks, participants used focused (depth-first) processes when the application Domain was familiar as did participants with greater IS Domain Knowledge. We also found that, for inferential problem-solving tasks, participants used exploratory (breadth-first) processes when the application Domain was familiar as did participants with greater IS Domain Knowledge. We then show how cognitive psychology literature on problem solving can help explain the effects of ISDK and ADK and, thus, provide the theoretical foundation for analyzing the roles of each type of Knowledge in the process of IS problem solving.
-
information search process for a well structured is problem the role of is and application Domain Knowledge
International Conference on Information Systems, 2008Co-Authors: Vijay Khatri, Iris VesseyAbstract:Prior research has shown that the effect of information systems (IS) Domain Knowledge and application Domain Knowledge on problem solving is contingent on task type (Khatri et al. 2006). We build on this study by engaging in an in-depth analysis of how both these types of Knowledge influence one type of task referred to as “schema-based problem solving” task. Our theoretical analysis is based on the fact that conceptual schema understanding is a well-structured problem area and that, in such a setting, participants engage in depth-first rather than the breadth-first search that is evident in the more-frequently studied ill-structured problem areas. We used protocol analysis to explore the search process in the context of varying levels of both IS and application Domain Knowledge. We found that Knowledge of the IS and application Domains result in similar effects on the search process: both high IS Knowledge and familiarity with the application Domain result in deeper (more focused) search.
-
understanding conceptual schemas exploring the role of application and is Domain Knowledge
Information Systems Research, 2006Co-Authors: Vijay Khatri, Iris Vessey, V Ramesh, Paul F Clay, Sungjin ParkAbstract:Although information systems (IS) problem solving involves Knowledge of both the IS and application Domains, little attention has been paid to the role of application Domain Knowledge. In this study, which is set in the context of conceptual modeling, we examine the effects of both IS and application Domain Knowledge on different types of schema understanding tasks: syntactic and semantic comprehension tasks and schema-based problem-solving tasks. Our thesis was that while IS Domain Knowledge is important in solving all such tasks, the role of application Domain Knowledge is contingent upon the type of understanding task under investigation. We use the theory of cognitive fit to establish theoretical differences in the role of application Domain Knowledge among the different types of schema understanding tasks. We hypothesize that application Domain Knowledge does not influence the solution of syntactic and semantic comprehension tasks for which cognitive fit exists, but does influence the solution of schema-based problem-solving tasks for which cognitive fit does not exist. To assess performance on different types of conceptual schema understanding tasks, we conducted a laboratory experiment in which participants with high- and low-IS Domain Knowledge responded to two equivalent conceptual schemas that represented high and low levels of application Knowledge (familiar and unfamiliar application Domains). As expected, we found that IS Domain Knowledge is important in the solution of all types of conceptual schema understanding tasks in both familiar and unfamiliar applications Domains, and that the effect of application Domain Knowledge is contingent on task type. Our findings for the EER model were similar to those for the ER model. Given the differential effects of application Domain Knowledge on different types of tasks, this study highlights the importance of considering more than one application Domain in designing future studies on conceptual modeling.
