دانلود رایگان مقاله لاتین ANN با برنامه کاربردی برای شناسایی تقلب و بازاریابی مستقیم از سایت الزویر
عنوان فارسی مقاله:
شبکه عصبی مصنوعی (ANN) با استفاده از برنامه های کاربردی برای تشخیص تقلب و بازاریابی مستقیم
عنوان انگلیسی مقاله:
A profit-driven Artificial Neural Network (ANN) with applications to fraud detection and direct marketing
سال انتشار : 2016
بخشی از مقاله انگلیسی:
2. Related works
A considerable amount of literature has been published on cost-sensitive learning, however, there has been relatively little literature published on maximization of the total net profit using individual profits and costs. Numerous studies have the tendency to focus on cost-sensitive learning algorithms such as over-sampling, under-sampling, meta-cost, cost-sensitive boosting, and meta heuristics [1,9,14,19–25] while some of them have worked on cost-sensitive ANN [2–4,14,21,23,26–29]. In most of these researches the authors have aimed to address the imbalance between classes and the different types of misclassification. By considering these issues resulting models are cost-sensitive classifiers which work well in imbalanced data and minimize total cost instead of minimizing total number of misclassifications [30,35]. However the central issue in these model is the classbased costs. Salchenberger et al.  developed a neural network model with variable thresholds considering statistical type I and type II errors and they used this model in thrift failure prediction. They showed the high performance of neural network in discriminating between healthy and failed institutions in comparison to other traditional models. Berardi and Zhang  studied the effect of different misclassification costs on neural network and their results represented that this scenario can be used to reach the optimal decision making based on cost of misclassification. Also they implied that most of the statistical performance metrics are not suitable to show the better performance of cost-sensitive classifiers. In the literature, some of the cost-sensitive models have been developed using different costs in cost matrix. Here, a cost matrix has been considered which contains different costs for different types of misclassifications. Table 2 illustrates a general cost matrix for these types of cost-sensitive models where there is a specific cost for misclassifying a case (positive) as non-case (negative) which is shown as C1 and another cost for misclassification of a non-case as case (C2).
A profit-driven artificial neural network (ANN) with applications to fraud ... https://www.researchgate.net/.../311478211_A_profit-driven_artificial_neural_network_... Request (PDF) | A profit-driven arti... on ResearchGate, the professional network for scientists. A profit-driven Artificial Neural Network (ANN) with ... - Semantic Scholar https://www.semanticscholar.org/...profit-driven-Artificial-Neural-Network-ANN.../50... So far, most of the research is focused on predictive modeling using statistical optimization, while profit maximization has been given less priority. It is exactly this gap that will be addressed in this study by taking a profit-driven approach to develop a profit-driven Artificial Neural Network (ANN) function which gives variable ... [PDF]accuracy driven artificial neural networks in stock ... - Semantic Scholar https://pdfs.semanticscholar.org/b3bb/1bfae3c00db46a6f46144bdba6404a4f9f4f.pdf by S Simon - Cited by 12 - Related articles Artificial Neural Networks (ANN) has been identified to be one of the dominant data mining techniques in stock market ... Also, we explore the possible research strategies in this accuracy driven ANN models. KEYWORDS. Artificial .... one can earn a portion of the company's profit called dividend. Also, by selling the shares ...