Search results

Credit risk assessment has gained importance in recent years due to global financial crisis and credit crunch. Financial institutions therefore seek the support of credit rating agencies to predict the ability of creditors to meet financial persuasions. The purpose of this paper is to construct neural network (NN) and fuzzy support vector machine (FSVM) classifiers to discriminate good creditors from bad ones and identify a best classifier for credit risk assessment.

This study uses artificial neural network, the most popular AI technique used in the field of financial applications for classification and prediction and the new machine learning classification algorithm, FSVM to differentiate good creditors from bad. As membership value on data points influence the classification problem, this paper presents the new FSVM model. The instances membership is computed using fuzzy c-means by evolving a new membership. The FSVM model is also tested on different kernels and compared and the classifier with highest classification accuracy for a kernel is identified.

The paper identifies a standard AI model by comparing the performances of the NN model and FSVM model for a credit risk data set. This work proves that that FSVM model performs better than back propagation-neural network.

The proposed model can be used by financial institutions to accurately assess the credit risk pattern of customers and make better decisions.

This paper has developed a new membership for data points and has proposed a new FCM-based FSVM model for more accurate predictions.

In this research, neural network (NN) and genetic algorithm (GA) are used together to design optimal NN structure. The proposed approach combines the characteristics of GA and NN to reduce the computational complexity of artificial intelligence applications in design and manufacturing. Genetic input selection approach is introduced to obtain optimal NN topology. Experimental results are given to evaluate the performance of the proposed system.

The purpose of this paper is to formulate a specialized artificial neural network algorithm utilizing radial basis function (RBF) for modeling of time to failure of aircraft engine turbines.

The model uses training failure data collected from operators of turboprop aircraft working in harsh desert conditions where sand erosion is a detrimental factor in reducing turbine life. Accordingly, the model is more suited to accurate prediction of life of critical components of such engines. The used RBF employs a closest neighbor type of classifier and the hidden unit’s activation is based on the displacement between the early prototype and the input vector.

The results of the algorithm are compared to earlier work utilizing Weibull regression modeling, as well as Feed Forward Back Propagation NN. The results show that the failure rates estimated by RBF more closely match actual failure data than the estimations by both other models. The trained model showed reasonable accuracy in predicting future failure events. Moreover, the technique is shown to have comparatively higher efficiency even with reduced number of neurons in each layer of ANN. This significantly decreases computation time with minimum effect on the accuracy of results.

Using RBF technique significantly decreases the computational time with minimum effect on the accuracy of results.

Epidemics not only affect the public health but also are a threat to a nation's growth and economy as well. Early prediction of epidemic can be beneficial to take preventive measures and to reduce the impact of epidemic in an area.

A deep neural network (DNN) based context aware smart epidemic system has been proposed to prevent and monitor epidemic spread in a geographical area. Various neural networks (NNs) have been used: LSTM, RNN, BPNN to detect the level of disease, direction of spread of disease in a geographical area and marking the high-risk areas. Multiple DNNs collect and process various data points and these DNNs are decided based on type of data points. Output of one DNN is used by another DNN to reach to final prediction.

The experimental evaluation of the proposed framework achieved the accuracy of 87% for the synthetic dataset generated for Zika epidemic in Brazil in 2016.

The proposed framework is designed in a way that every data point is carefully processed and contributes to the final decision. These multiple DNNs will act as a single DNN for the end user.