There exist numerous schemes and methods to determine the output of an ensemble of classifiers. The most common approach being the majority vote. Furthermore, we might expect that an improvement can be achieved if the...
详细信息
There exist numerous schemes and methods to determine the output of an ensemble of classifiers. The most common approach being the majority vote. Furthermore, we might expect that an improvement can be achieved if there is a method by which we may weigh the members of the ensemble according to their individual performance. The feature based approach presented an architecture that tries to approach this target. However, if there is a way that the final classification may influence these weights we should expect an increased performance in the overall classification task. In this paper we present a new training algorithm that utilizes a feedback mechanism to iteratively improve the classification capability of the feature based approach. This approach is compared with the standard training method as well as standard aggregation schemes for combining classifier ensembles. Empirical results show that this architecture improved on classification accuracy.
This paper presents a multi-agent system for coordinating the deployment of multiple sensors in a modeled environment. The sensing task is the maximal sensor coverage of one or more targets in a scene and the position...
详细信息
This paper presents a multi-agent system for coordinating the deployment of multiple sensors in a modeled environment. The sensing task is the maximal sensor coverage of one or more targets in a scene and the position of each sensor is controlled by an autonomous agent. The agents rely on negotiation to achieve the level of coordination necessary to accomplish the given sensing task. Currently, the system focuses on the use of cameras for visual inspection tasks in which a single camera may be inadequate due to occluding objects in the scene, the number of targets to be observed or the sheer size of the target. The paper presents the negotiation mechanism developed for the multi-agent planning of the camera positions and illustrates its effectiveness by an example. Results show that the agents were able to autonomously position the cameras so as to allow for an acceptable level of coverage of the targets being observed.
Among existing ocean data assimilation methodologies, reduced-state Kalman filters are a widely-studied compromise between resolution, optimality, error specification, and computational feasibility. In such reduced-st...
详细信息
Among existing ocean data assimilation methodologies, reduced-state Kalman filters are a widely-studied compromise between resolution, optimality, error specification, and computational feasibility. In such reduced-state filters, the measurement update takes place on a coarser grid than that of the general circulation model (GCM); therefore, these filters require mapping operators from the GCM grid to the reduced state and vice-versa. The general requirements are that the state-reduction and interpolation operators be pseudo-inverses of each other, that the coarse state defines a closed dynamical system, that the mapping operations be insensitive to noise, and that they be appropriate for regions with irregular coastlines and bathymetry. In this paper we investigate a variety of approaches, including computing the pseudoinverse by brute force, using the FFT, subsampling methods, implicit methods, and finally develop a novel iterative approach. We also evaluate the mapping performance of eleven interpolation kernels; surprisingly, common kernels such as bilinear, exponential, Gaussian, and sinc, performed only moderately well. This comprehensive study greatly reduces the computational bottleneck and guesswork of pseudo-inverse algorithms, making possible the application of reduced-state filters to global problems at state-of-the-art resolution.
Complex products and systems, like an aircraft, ship, or machinery plant, involve a large number of components which are arranged under spatial constraints/relationships in the design space. Current practices approxim...
详细信息
Complex products and systems, like an aircraft, ship, or machinery plant, involve a large number of components which are arranged under spatial constraints/relationships in the design space. Current practices approximate these relationships. with simplified "dimensional constraints" aiming at formulating the design problem as a system of (in)equalities to be solved automatically, e.g., by a geometric-constraint solver This research proposes informationally-complete models for design-constraints based on an analysis of geometric and non-geometric properties of the related space volumes. Also, an extended product model is proposed describing the system's structure and components as well as related procedures and constraints to be used as a system life-cycle model.
We propose an evolutionary neural network-training algorithm for beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure. Generally the...
详细信息
We propose an evolutionary neural network-training algorithm for beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure. Generally the network resulting from learning applied to a predetermined architecture is either insufficient or over-complicated. This paper describes a hierarchical genetic learning model of the BBFNN. In order to examine the performance of the proposed algorithm, they were used for the approximation problems. The results obtained are very satisfactory with respect to the relative error.
This paper presents a new approach to understanding mental models for complex industrial systems. Using the Abstraction Hierarchy (AH) framework as a way to describe mental models, we propose an approach to assess eco...
详细信息
This paper presents a new approach to understanding mental models for complex industrial systems. Using the Abstraction Hierarchy (AH) framework as a way to describe mental models, we propose an approach to assess ecological compatibility between an operator's internal mental model and a model of the environment. In order to do so, we have combined and built upon the previous work of Rasmussen (1979) and Moray (1996). We believe that this new approach can assess ecological compatibility and ultimately provide the operator with a more accurate and complete internal mental model that reflects the reality of the environment. We argue that Ecological interfacedesign (EID) is a way to link mental models and the environment. Future work will be needed in order to assess ecological compatibility. This includes capturing and examining operators' internal mental models, and comparing them with models of the environment.
This paper presents a new approach to understanding mental models for complex industrial systems. Using the Abstraction Hierarchy (AH) framework as a way to describe mental models, we propose an approach to assess eco...
This paper presents a new approach to understanding mental models for complex industrial systems. Using the Abstraction Hierarchy (AH) framework as a way to describe mental models, we propose an approach to assess ecological compatibility between an operator's internal mental model and a model of the environment. In order to do so, we have combined and built upon the previous work of Rasmussen (1979) and Moray (1996). We believe that this new approach can assess ecological compatibility and ultimately provide the operator with a more accurate and complete internal mental model that reflects the reality of the environment. We argue that Ecological interfacedesign (EID) is a way to link mental models and the environment. Future work will be needed in order to assess ecological compatibility. This includes capturing and examining operators' internal mental models, and comparing them with models of the environment.
Work domain analysis is an approach that is moving from research to practical usage. Being highly analytical and time-intensive in its early stages, clients will search for early indications that the analysis is on tr...
详细信息
Work domain analysis is an approach that is moving from research to practical usage. Being highly analytical and time-intensive in its early stages, clients will search for early indications that the analysis is on track and relevant. This paper presents one approach for validating a work domain analysis by mapping the action patterns of operators against the work domain space developed by the analysis. This approach provides some validation the analysis, shows the use of the work domain over time in three dimensions, and provides a substantial link between the work analysis approach and traditional task-based approaches.
Work domain analysis is an approach that is moving from research to practical usage. Being highly analytical and time-intensive in its early stages, clients will search for early indications that the analysis is on tr...
Work domain analysis is an approach that is moving from research to practical usage. Being highly analytical and time-intensive in its early stages, clients will search for early indications that the analysis is on track and relevant. This paper presents one approach for validating a work domain analysis by mapping the action patterns of operators against the work domain space developed by the analysis. This approach provides some validation the analysis, shows the use of the work domain over time in three dimensions, and provides a substantial link between the work analysis approach and traditional task-based approaches.
An ensemble of neural networks offers several advantages over classical single classifier systems when applied to complex pattern classification problems. However, the performance of the ensemble as a unit depends not...
详细信息
ISBN:
(纸本)0780370449
An ensemble of neural networks offers several advantages over classical single classifier systems when applied to complex pattern classification problems. However, the performance of the ensemble as a unit depends not only on the effective aggregation of the modules decisions, but also on the accuracy of the individual classification decisions of each module. The accuracy at the modular level is a result of the quality of training received by each module. This paper presents an adaptive training algorithm that can be used to direct the training of the individual modules so as to improve the classification accuracy and training efficiency of the ensemble.
暂无评论