A Classification Capability of Reflective Neural Networks in Medical Databases
0
Citation
0
Reference
20
Related Paper
Abstract:
Medical database recorded diagnostic information based on a patient’s medical card. However, each card does not fulfill all the required information for learning algorithm. Reflective Neural Network has an outstanding classification capability, even if such records with shortage exist in a set of training cases. Reflective Neural Network is based on the network module concept. There are two kinds of network modules; an allocation module to distribute a training case and some classification modules to classify a subset of training cases. Each classification module consists of a monitor neural network and a worker neural network. The monitor neural network estimates how conformable the worker neural network is to a given training case. Moreover, the training case is distributed over different classification modules. These classification modules compete with each other in the classification task. In this paper, we report the classification capability in a medical database on the patients in ICUs.Keywords:
Economic shortage
Cite
Classification is a major problem of study that involves formulation of decision boundaries based on the training data samples. The limitations of the single neural network approaches motivate the use of multiple neural networks for solving the problem in the form of ensembles and modular neural networks. While the ensembles solve the problem redundantly, the modular neural networks divide the computation into multiple modules. The modular neural network approach is used where a Self Organizing Map (SOM) selects the module which would perform the computation of the output, whenever any input is given. In the proposed architecture, the SOM selects multiple modules for problem solving, each of which is a neural network. Then the multiple selected neural networks are used redundantly for computing the output. Each of the outputs is integrated using an integrator. The proposed model is applied to the problem of Breast Cancer diagnosis, whose database is made available from the UCI Machine Learning Repository. Experimental results show that the proposed model performs better than the conventional approaches.
Modular neural network
Cite
Citations (14)
This article aims at showing an architecture of neural networks designed for the classification of data distributed among a high number of classes. A significant gain in the global classification rate can be obtained by using our architecture. This latter is based on a set of several little neural networks, each one discriminating only two classes. The specialization of each neural network simplifies their structure and improves the classification. Moreover, the learning step automatically determines the number of hidden neurons. The discussion is illustrated by tests on databases from the UCI machine learning database repository. The experimental results show that this architecture can achieve a faster learning, simpler neural networks and an improved performance in classification.
Data set
Cite
Citations (19)
Training single neural network is a difficult task.In each training session, the network is trained for hundred of epochs and some problem may involve a large amount of variables and data.Thus, training single network is time consuming. Therefore, distributed learning approaches such as hierarchical, multi-stage, parallel neural network computing and multi-modal neural network was introduced.
Training set
Cite
Citations (0)
In this paper, a study of the effectiveness of a multiple classifier system (MCS) in a medical diagnostic task is described. A hybrid network, based on the integration of a fuzzy ARTMAP and the probabilistic neural network, is employed as the basis of the MCS. Outputs from multiple networks are combined using some decision combination method to reach a final prediction. By using a real medical database, a set of experiments has been conducted to evaluate the performance of the MSC with different network configurations. The experimental results reveal the potential of the MCS as a useful decision support tool in the medical field.
Probabilistic classification
Cite
Citations (2)
Representation
Basis (linear algebra)
Cite
Citations (31)
Although an individual neural network has proven capabilities that are powerful for pattern detection and function approximation, real-life applications of neural networks often require an entire system for the training and usage of such neural networks. We describe systems for using neural networks in decision making roles such as medical diagnosis and pattern recognition. In our medical applications, the neural network output is treated as a composite variable subject to statistical validation such as an ROC plot analysis, use of re-sampled training to measure performance variance, and avoidance of overtraining. Another system for use of neural networks lies in our approach for training on boundaries rather than individual data points in pattern classification and image analysis problems. We discuss optimizing the neural network and training using these systems.
Overtraining
Cellular neural network
Cite
Citations (0)
Self-organizing map
Cite
Citations (1)
There is always a trade off between the number of cases to be stored in the case library of a case-based expert system and the performance of retrieval efficiency. The larger the case library, the more the problem space covered, however, it would also downgrade the system performance if the number of cases grows to an unacceptably high level. In the paper, an approach to maintaining the size of a case-based expert system is proposed. The main idea is using the fuzzy class membership value of each record, determined by a trained neural network, to guide the record deletion. These fuzzy membership values are used to calculate the case density of each record, and a deletion policy can then be used to determine the percentage of record to be deleted. Using this approach, we could maintain the size of the case-base without loosing a significant amount of information. A testing case-base consisting of 214 records is used as an illustrative example of our approach, the neural network software NEURALWORKS PROFESSIONAL II/PLUS/sup (C)/ is used to develop the neural network. It was shown that it could reduce the size of the case library by 28% if we select those records that have an overall class membership of over 0.8 and case density over 0.95. Future work includes integrating adaptation rules for building deletion policy.
Cite
Citations (7)
Adaptive information systems receive and process a large amount of information of various types. An urgent task is the classification and the correct distribution of incoming data in given categories. The paper considers existing approaches to solving this problem using machine learning technologies. They are combined in a single neural network method for classifying and distributing data. The proposed method will automate the solution of problems of classification of information and files, verification of data received from users, the compliance of files with specified categories, the absence of duplication of information. The practical implementation of the method in the form of a neural network data distribution channel is considered. The experiments performed prove the accuracy and effectiveness of the proposed method.
Cite
Citations (0)
The subject of this work is a neural network, the parameters of its architecture and the data array (dataset) for its training. The aim of the work is the software implementation of part of the DLP-system (Data Leak Prevention), which allows to monitor the traffic of a corporate network and to control the transfer of confidential data over this network using a neural network. The entire development process is represented by five stages: theory, design, preparation of data for training the neural network, training the neural network and testing the implemented system. There is a brief overview of the market for such solutions in the article. The parameters used to construct the neural network architecture used to solve the problem of text data classification are described in detail. The result of the work is a functioning part of the DLP system, which allows monitoring the traffic of a corporate network via a web-interface and controlling the transfer of confidential data over this network using a one-dimensional convolutional neural network 1D CNN.
Transfer of learning
Interface (matter)
Network monitoring
Cite
Citations (1)