In this paper we propose a method for learning Bayesian belief networks from data. The method uses artificial neural networks as probability estimators, thus avoiding the need for making prior assumptions on the natur...
详细信息
ISBN:
(纸本)0262100657
In this paper we propose a method for learning Bayesian belief networks from data. The method uses artificial neural networks as probability estimators, thus avoiding the need for making prior assumptions on the nature of the probability distributions governing the relationships among the participating variables. This new method has the potential for being applied to domains containing both discrete and continuous variables arbitrarily distributed. We compare the learning performance of this new method with the performance of the method proposed by Cooper and Herskovits in [7]. The experimental results show that, although the learning scheme based on the use of ANN estimators is slower, the learning accuracy of the two methods is comparable. Category: Algorithms and Architectures.
In current practice, tapped delay line models such as the time delay neural network (TDNN) are commonly implemented using a direct form structure. In this paper, we show that the problem of high parameter sensitivity,...
详细信息
In current practice, tapped delay line models such as the time delay neural network (TDNN) are commonly implemented using a direct form structure. In this paper, we show that the problem of high parameter sensitivity, well known in linear systems, also applies to nonlinear models such as the TDNN. To overcome the consequent numerical problems, we propose a cascade form TDNN (CTDNN) and show its advantages over the commonly used direct form TDNN.
OBJECTIVES: An American medical Association survey reported that more than half of physicians are subjects of either clinical or economic profiling. This multilevel meta-analysis was designed to assess the clinical ef...
详细信息
OBJECTIVES: An American medical Association survey reported that more than half of physicians are subjects of either clinical or economic profiling. This multilevel meta-analysis was designed to assess the clinical effect of peer-comparison feedback intervention (profiles) in changing practice patterns. METHODS: Systematic computerized and manual searches were combined to retrieve articles on randomized controlled clinical trials testing profiling reports. Eligible studies were randomized, controlled clinical trials that tested peer-comparison feedback intervention and measured utilization of clinical procedures. To use all available information, data were abstracted and analyzed on three levels: (1) direction of effects, (2) p value ham the statistical comparison, and (3) odds ratio (OR). MAIN RESULTS: In the 12 eligible trials, 553 physicians were profiled. The test result was p < .05 for the vote-counting sign test of 12 studies (level 1) and p < .05 far the z-transformation test of 8 studies (level 2). There were 5 trials included In the OR analysis (level 3). The primary effect variable in two Of the 5 trials had a nonsignificant OR. However, the overall OR calculated by the Mantel-Haenszel method was significant (1.091, confidence interval: 1.045 to 1.136). CONCLUSIONS: Profiling has a statistically significant, hut minimal effect on the utilization of clinical procedures. The results of this study indicate a need for controlled clinical evaluations before subjecting large numbers of physicians to utilization management interventions.
In this paper we propose a method for learning Bayesian belief networks from data. The method uses artificial neural networks as probability estimators, thus avoiding the need for making prior assumptions on the natur...
In this paper we propose a method for learning Bayesian belief networks from data. The method uses artificial neural networks as probability estimators, thus avoiding the need for making prior assumptions on the nature of the probability distributions governing the relationships among the participating variables. This new method has the potential for being applied to domains containing both discrete and continuous variables arbitrarily distributed. We compare the learning performance of this new method with the performance of the method proposed by Cooper and Herskovits in [7]. The experimental results show that, although the learning scheme based on the use of ANN estimators is slower, the learning accuracy of the two methods is comparable.
Mathematical modeling of the dynamic behavior of physical systems, gives computers that mimic the capabilities to intelligently monitor and predict their evolutionary characteristics. In this paper, we present a syste...
详细信息
Mathematical modeling of the dynamic behavior of physical systems, gives computers that mimic the capabilities to intelligently monitor and predict their evolutionary characteristics. In this paper, we present a system for tracking the time-varying features of non-rigid objects in images of evolving scenes, using the elastic string model of planar contours, which permits the inference and prediction of the quantitative parameters that characterize evolutionary behavior. The goal of our work is to dynamically track non-rigid objects in video sequences, using object alignment techniques based on the properties of the elastic string. We present experimental results of growth cone and neurite tracking in cell growth and motion studies.
This paper describes the development of a Managed Care Workstation for implementation in a Department of Veterans Affairs hospital. Each VA hospital information system contains a wealth of information in a comprehensi...
暂无评论