Print ISSN: 1811-9212

Online ISSN: 2617-3352

Keywords : Machine learning


Data Hiding by Unsupervised Machine Learning Using Clustering K-mean Technique

Hiba Hamdi Hassan; Maisa'a Abid Ali Khodher

IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING, 2021, Volume 21, Issue 4, Pages 37-49
DOI: https://doi.org/10.33103/uot.ijccce.21.4.4

Steganography includes hiding text, image, or any sentient information inside another image, video, or audio. It aims to increase individuals’ use of social media, the internet and web networks to securely transmit information between sender and receiver and an attacker will not be able to detect its information. The current article deals with steganography that can be used as machine learning method, it suggests a new method to hide data by using unsupervised machine learning (clustering k-mean algorithm). This system uses hidden data into the cover image, it consists of three steps: the first step divides the cover image into three clusterings that more contrast by using k-means cluster, the selects a text or image to be converted to binary by using ASCII code, the third step hides a binary message or binary image in the cover image by using sequential LSB method. After that, the system is implemented. The objective of the suggested system is obtained, using Unsupervised Machine Learning (K-mean technique) to securely send sensitive information without worrying about man-in-the-middle attack was proposed. Such a method is characterized by high security and capacity. Through evaluation, the system uses PSNR, MSE, Entropy, and Histogram to hide the secret message and secret image in the spatial domain in the cover image.

An Efficient Feature Engineering Method for Fraud Detection in E-commerce

Suha M. Najem; Suhad kadhem

IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING, 2021, Volume 21, Issue 3, Pages 40-52
DOI: https://doi.org/10.33103/uot.ijccce.21.3.4

With the speedy expansion of e-commerce, credit cards have also become rising vogue, and that makes online transactions sleek and suitable. In conjunction with rising in online transactions, credit card fraud also increasing, which contributes to losses incurred yearly. As a result, many deep and machine learning methods are produced to fix such as problems like Logistic Regression (LR), Support Vector Machine (SVM), Naïve Bayes (NB), and other algorithms, but the current models are still not accurate. Moreover, sometimes the used datasets still need further preprocessing, since that has been approved the important role of feature engineering in performance optimization. In this paper, effective feature engineering and feature selection methods have been produced for preprocessing the raw dataset, which was transformed with Exploratory Data Analysis (EDA). Then LightGBM, XGboost, and Random forest classifiers are used for fraud detection. Experiments show that the LightGBM and XGboost models achieved the best accuracy with 100% after applying further preprocessing on the dataset.

Impostor Detection Based Finger Veins Applying Machine Learning Methods

Ashraf Tahseen Ali; Hasanen Abdullah; Mohammed Natiq Fadhil

IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING, 2021, Volume 21, Issue 3, Pages 98-111
DOI: https://doi.org/10.33103/uot.ijccce.21.3.9

Finger veins are different from other biometric signs; it is a special characteristic of the human body. The challenge for an imposter to explore and comprehend it, since the veins are below the skin, it is impossible to tell which one is, and which one stands out because the person has more than one finger to examine. Impostor recognition based on applying three machine-learning methods will be presented in this article, and then there is a discussion at preprocessing, Linear Discriminant Analysis (LDA) for feature extraction, and k fold cross-validation as an evaluation method. These measures were carried out on two different datasets, which are the Shandong University Machine Learning and Applications - Homologous Multi-modal Traits (SDUMLA-HMT) Dataset and the University of Twente Finger Veins (UTFV) dataset. The classifier with the best results was Support Vector Machine (SVM) and Linear Regression (LR) had the lowest classifier accuracy.