摘要 :
In this work we propose an audio-visual model for predicting temporal saliency in videos, that we validate and evaluate in an alternative way by employing fMRI data. We intend to bridge the gap between the large improvements achie...
展开
In this work we propose an audio-visual model for predicting temporal saliency in videos, that we validate and evaluate in an alternative way by employing fMRI data. We intend to bridge the gap between the large improvements achieved during the last years in computational modeling, especially in deep learning, and the neurobiological and behavioral research regarding human vision. The proposed audio-visual model incorporates both state-of-the-art deep architectures for visual saliency, which were trained on eye-tracking data, and behavioral findings concerning audio-visual integration in multimedia stimuli. A new fMRI database has been collected for evaluation purposes, that includes various videos and subjects. This dataset may prove useful not only for saliency but for other computer vision problems as well. The evaluation of our model using the new fMRI database under a mixed-effect analysis shows that the proposed saliency model has strong correlation with both the visual and audio brain areas, that confirms its effectiveness and appropriateness in predicting audio-visual saliency for dynamic stimuli.
收起
摘要 :
A review is presented of reusability and modularity ideas from the software engineering literature, most of which are applicable to the modeling context. Many features of structured modeling support reusability and modularity, and...
展开
A review is presented of reusability and modularity ideas from the software engineering literature, most of which are applicable to the modeling context. Many features of structured modeling support reusability and modularity, and these are noted. The main focus, however, is on achieving reusability and modularity by the integration of two or more model schemas. A five-step approach for integrating schemas written in SML (Structured Modeling Language) is proposed for this purpose. Examples are given to illustrate this approach, and the pros and cons of structured modeling for reuse are discussed at some length.
收起
摘要 :
A review is presented of reusability and modularity ideas from the software engineering literature, most of which are applicable to the modeling context. Many features of structured modeling support reusability and modularity, and...
展开
A review is presented of reusability and modularity ideas from the software engineering literature, most of which are applicable to the modeling context. Many features of structured modeling support reusability and modularity, and these are noted. The main focus, however, is on achieving reusability and modularity by the integration of two or more model schemas. A five-step approach for integrating schemas written in SML (Structured Modeling Language) is proposed for this purpose. Examples are given to illustrate this approach, and the pros and cons of structured modeling for reuse are discussed at some length.
收起
摘要 :
Database modelling is classically defining the database dictionary, database structuring and functionality within one singleton paradigm. This approach has led to sophisticated financial services, to enterprise information systems...
展开
Database modelling is classically defining the database dictionary, database structuring and functionality within one singleton paradigm. This approach has led to sophisticated financial services, to enterprise information systems and other database-backed practical solution which are easy to handle, relatively simple to change and to implement and which satisfied the needs of business in the 90ies. At the same time a number of applications have been developed that used the potential within the data for analysis, for exchange and collaboration of systems, e.g. OLTP-OLAP systems, decision support systems, scientific information systems, collaborative information systems and web information systems. These applications do not use a singleton language for data storage, data computation and data delivery. Their languages use different paradigms. We therefore need a way for specification of database applications that provide facilities for appropriate modelling depending on the needs. This paper aims in developing a framework for modelling and implementing layered database applications. It defines a database model for each of the layers and a facility for exchange of data throughout the different layers in a coherent form.
收起
摘要 :
A dynamic system model is proper for a particular application if it achieves the accuracy required by the application with minimal complexity. Because model complexity often - but not always - correlates inversely with simulation ...
展开
A dynamic system model is proper for a particular application if it achieves the accuracy required by the application with minimal complexity. Because model complexity often - but not always - correlates inversely with simulation speed, a proper model is often alternatively defined as one balancing accuracy and speed. Such balancing is crucial for applications requiring both model accuracy and speed, such as system optimization and hardware-in-the-loop simulation. Furthermore, the simplicity of proper models conduces to control system analysis and design, particularly given the ease with which lower-order controllers can be implemented compared to higher-order ones. The literature presents many algorithms for deducing proper models from simpler ones or reducing complex models until they become proper. This paper presents a broad survey of the proper modeling literature. To simplify the presentation, the algorithms are classified into frequency-, projection-, optimization-, and energy-based, based on the metrics they use for obtaining proper models. The basic mechanics, properties, advantages and limitations of the methods are discussed, along with the relationships between different techniques, with the intention of helping the modeler to identify the most suitable proper modeling method for their application.
收起
摘要 :
A dynamic system model is proper for a particular application if it achieves the accuracy required by the application with minimal complexity. Because model complexity often - but not always - correlates inversely with simulation ...
展开
A dynamic system model is proper for a particular application if it achieves the accuracy required by the application with minimal complexity. Because model complexity often - but not always - correlates inversely with simulation speed, a proper model is often alternatively defined as one balancing accuracy and speed. Such balancing is crucial for applications requiring both model accuracy and speed, such as system optimization and hardware-in-the-loop simulation. Furthermore, the simplicity of proper models conduces to control system analysis and design, particularly given the ease with which lower-order controllers can be implemented compared to higher-order ones. The literature presents many algorithms for deducing proper models from simpler ones or reducing complex models until they become proper. This paper presents a broad survey of the proper modeling literature. To simplify the presentation, the algorithms are classified into frequency-, projection-, optimization-, and energy-based, based on the metrics they use for obtaining proper models. The basic mechanics, properties, advantages and limitations of the methods are discussed, along with the relationships between different techniques, with the intention of helping the modeler to identify the most suitable proper modeling method for their application.
收起
摘要 :
In higher education, understanding, modelling, and forecasting student performance presents considerable hurdles in terms of developing accurate and effective diagnostic models. Several researchers have attempted to construct inte...
展开
In higher education, understanding, modelling, and forecasting student performance presents considerable hurdles in terms of developing accurate and effective diagnostic models. Several researchers have attempted to construct intelligent classifiers for predicting student accomplishment, but they failed to recognise the necessity of identifying the main components that contribute to the obtained performance. Such recognition is vital for programme directors to evaluate the strengths and limitations of their academic programmes and, as a result, make the required corrections to improve student accomplishment. Consequently, there is room to present a model that improves the accuracy of predicting student academic performance, as measured by future grades in various courses, and, also an optimized multi-label classifier that predicts qualitative values for the influence of various factors on the obtained student performance, primarily using a feed-forward neural network. The goal of the experimental findings are to demonstrate significant gains over single baseline models (e.g., linear regression, matrix factorization), indicating the applicability of the suggested technique in identifying several factors influencing student performance. This study underlines the need of predicting student achievement of learning outcomes in future studies. Along with this, another model that proved to be precise and accurate in predicting the outcome in a similar but not identical scenario, was the KNN classifier, which used data to classify the graduation chances of students. Using the classifiers in the data sets, there are several analyses that can be covered in terms of question pattern, format, and more. This study aims to push a new service that can be established in several organizations and locations to benefit educational professionals and students. The datasets are naturally created in the course of the university’s classes. This allows for a lot of flexibility in terms of output variables and classifiers. Along with this, using PyCharm or another Python based IDE and Python’s built in packages and libraries to perform the analyses and build the neural models. This allows for widespread application and use.
收起
摘要 :
In railway systems, timetables and infrastructure are key components for satisfying transport needs. In order to find a feasible configuration that satisfies the complex set of requirements, numerous timetables and track topologie...
展开
In railway systems, timetables and infrastructure are key components for satisfying transport needs. In order to find a feasible configuration that satisfies the complex set of requirements, numerous timetables and track topologies have to be evaluated. Therefore, there is an immense need for automatic timetable and track topology design. This paper discusses the concept of creating abstraction layers based on a data model to support and simplify these designs. It uses different state-of-the-art models and combines them so that all aspects of railway operation are included. Using this method, a layer model has been developed that was inspired by the OSI Reference Model and takes current developments such as the UIC RailTopoModel into consideration. Possible applications of such a layer model are the automatic generation of timetables and track topologies, as well as entrance data for synchronous railway simulations for capacity evaluation, and a universal representation of a railway network.
收起
摘要 :
In railway systems, timetables and infrastructure are key components for satisfying transport needs. In order to find a feasible configuration that satisfies the complex set of requirements, numerous timetables and track topologie...
展开
In railway systems, timetables and infrastructure are key components for satisfying transport needs. In order to find a feasible configuration that satisfies the complex set of requirements, numerous timetables and track topologies have to be evaluated. Therefore, there is an immense need for automatic timetable and track topology design. This paper discusses the concept of creating abstraction layers based on a data model to support and simplify these designs. It uses different state-of-the-art models and combines them so that all aspects of railway operation are included. Using this method, a layer model has been developed that was inspired by the OSI Reference Model and takes current developments such as the UIC RailTopoModel into consideration. Possible applications of such a layer model are the automatic generation of timetables and track topologies, as well as entrance data for synchronous railway simulations for capacity evaluation, and a universal representation of a railway network.
收起
摘要 :
Dynamic models of physical systems with physically meaningful states and parameters have become increasingly important, for design, control and even procurement decisions. The successful use of models in these contexts requires th...
展开
Dynamic models of physical systems with physically meaningful states and parameters have become increasingly important, for design, control and even procurement decisions. The successful use of models in these contexts requires that the models be of sufficient quality. However, while algorithms have been developed to help formulate and integrate physical system models, as well as to generate minimum complexity physical system models, algorithms to assess the "quality" of dynamic system models have not been produced. This is true even if the attributes of model are limited to accuracy and validity. The objective of this paper is to introduce a new methodology that systematically quantifies the accuracy of a predicted system response and determines the validity of the physical system model used to predict the system response. The accuracy and validity of the model are evaluated using statistical properties of measured system response. The new algorithm is called Accuracy & Validation Algorithm for Simulation (AVASIM), and is a time-domain perspective comparing the model's time trajectories at user-defined points of interest as well as over the entire simulation horizon. To illustrate AVASIM, the quality of a handling model of a DaimlerChrysler Grand Cherokee is compared to the measurements obtained from that vehicle subjected to known steering inputs. Results demonstrate that the accuracy and validity of the Grand Cherokee model can be systematically assessed using the proposed methodology, and, thus, AVASIM appears to be a powerful tool for assessing the quality of system models.
收起
原文获取