Masoumeh Zareapoor

Analysis on Credit Card Fraud Detection Techniques: Based on Certain Design Criteria

fraud is increasing significantly with the development of modern technology and the global superhighways of communication, resulting in the loss of billions of dollars worldwide each year. The companies and financial institution loose huge amounts due to fraud and fraudsters continuously try to find new rules and tactics to commit illegal actions. Thus, fraud detection systems have become essential for all credit card issuing banks to minimize their losses. The most commonly used fraud detection methods are Neural Network (NN), rule-induction techniques, fuzzy system, decision trees, Support Vector Machines (SVM), Artificial Immune System (AIS), genetic algorithms, K-Nearest Neighbor algorithms. These techniques can be used alone or in collaboration using ensemble or meta-learning techniques to build classifiers. This paper presents a survey of various techniques used in credit card fraud detection and evaluates each methodology based on certain design criteria.

Statistical-based filtering system against DDOS attacks in cloud computing

A Distributed Denial of Service (DDOS) attack can make huge damages to resources and access of the resources to genuine users. Offered defending system cannot be easily applied in cloud computing due to their relatively low competence and wide storage. In this work we presented statistical technique to detect and filter DDOS attacks. The proposed model requires small storage and ability of fast detection. The obtained results show that our model has the ability to mitigate most of TCP attacks. Detection accuracy and Time consumption were the metrics used to evaluate the performance of our proposed model. From the simulation results, it is visible our algorithms achieve high detection accuracy (97%) with fewer false alarms.

Convolutional neural network in network (CNNiN): hyperspectral image classification and dimensionality reduction

Classification is a principle technique in hyperspectral images (HSIs), where a label is assigned to each pixel based on its characteristics. However, due to lack of labelled training instances in HSIs and also its ultra-high dimensionality, deep learning approaches need a special consideration for HSI classification. As one of the first works in the HSI classification, this study proposes a novel network pipeline called convolutional neural network in network (which is deeper than the existing approaches) by jointly utilising the spatial and spectral information and produces high-level features from the original HSI. This can occur by using spatial-spectral relationships of individual pixel vector at the initial component of the proposed pipeline; the extracted features are then combined to form a joint spatial-spectral feature map. Finally, a recurrent neural network is trained on the extracted features which contain wealthy spectral and spatial properties of the HSI to predict the corresponding label of each vector. The model has been tested on two large scale hyperspectral datasets in terms of classification accuracy, training error, and computational time.

Highly Discriminative Features for Phishing Email Classification by SVD

Unstructured text documents have drawn recently more attention, because with growing amount of text documents, there is a need to classify them automatically. But an important problem in field of text categorization is the huge dimensional and very sparse dataset which hurts generalization performance of classifiers. This paper presents a Singular Value Decomposition (SVD) technique to email classification, in order to compress optimally only the kind of documents (in our experiments email classes) and to retain the most informative and discriminate features from an email document. The performance evaluation is performed on email dataset which is publicly available to demonstrate the benefit of the LSA.

Hybrid Deep Neural Networks for Face Emotion Recognition

Abstract Deep Neural Networks (DNNs) outperform traditional models in numerous optical recognition missions containing Facial Expression Recognition (FER) which is an imperative process in next-generation Human-Machine Interaction (HMI) for clinical practice and behavioral description. Existing FER methods do not have high accuracy and are not sufficient practical in real-time applications. This work proposes a Hybrid Convolution-Recurrent Neural Network method for FER in Images. The proposed network architecture consists of Convolution layers followed by Recurrent Neural Network (RNN) which the combined model extracts the relations within facial images and by using the recurrent network the temporal dependencies which exist in the images can be considered during the classification. The proposed hybrid model is evaluated based on two public datasets and Promising experimental results have been obtained as compared to the state-of-the-art methods.