SEARCH RESULT

In the literature, machine learning algorithms are frequently used in detecting anomalies in network traffic and in building intrusion detection systems. Deep learning is a subfield of machine learning that trains a computer-based system to perform humanitarian tasks, such as disease diagnosis, speech recognition, image recognition, fraud detection, and making predictions. In the experimental study, NSL-KDD dataset was used for evaluating the performance of the proposed deep learning based DDoS detection model. NLS-KDD dataset contains normal network traffic and 23 different DDoS attacks that consists of 41 features. In the experimental study two different experiments are carried out. Firstly, the proposed deep neural network detected the Dos attacks with 0.988 classification accuracy. In the second experiment, the number of features of NSL-KDD is reduced to 24 by examining the previous feature reduction research on NSL-KDD dataset. The proposed deep neural network classified the all cyber-attacks with 0.984 classification accuracy. The 10-fold cross validation is used for all experiments. As a result, the proposed deep learning based DDoS detection achieved good performance.

International Conference on Advanced Technologies, Computer Engineering and Science
ICATCES

Ayşegül Sungur Ünal Mehmet Hacibeyoglu

Hiding and securing information is a basic demand throughout humanity. People have used their own bodies, languages, writings, etc. to provide this need. Steganography is acknowledged by science and art that researches hiding information methods. Steganography consists of two element basically that cover and secret information. In the past, people used their bodies and poems, diaries for cover and used tattoo and acrostic methods for secret information. In recently thanks to the developments of technology, Steganography has widened its methods and study areas. There are four Steganography methods which are text, image, audio and video in computer science. All types of Steganography methods have distinctive different ways to hide information. But if we want to mention the most used ones, we can say that changing characteristic of text (like color, font size) in text Steganography and changing Least Significant Bit(LSB) way for other types of Steganography methods. The LSB is a way that we overwrite the LSB of each byte of the cover (image, video, audio) with secret information binary representation. In our project, we are developing an android mobile application that allows user to hide a secret information inside any image. The image can be captured instantly or selected from user gallery. We are using LSB image Steganography method in order to hide secret information in image. Beside this, we encrypt the secret information with an encryption algorithm before inserting it in image. At the end, user can save the result image for the future or share with somebody who able to see the secret information only with this application.

International Conference on Advanced Technologies, Computer Engineering and Science
ICATCES

Celalettin Misman Mehmet Hacibeyoglu

A well balanced dataset is crucial for the performance of the data mining classification algorithms. In medical datasets, the percentage of normal labeled classes is higher than the percentage of abnormal labeled ones, which is called as class imbalance problem in data mining. If training dataset is imbalanced, the classification algorithm generally predicts the labels of the majority class instances correctly and the minority class instances incorrectly which leads to a major problem for artificial intelligence based medical diagnosis systems. To overcome this problem, many researchers proposed over-sampling and under-sampling techniques in the literature. Over-sampling techniques increase the number of minority class instances, where the randomly chosen instances from minority class is duplicated and added to the new training dataset or synthetic instances are generated from the minority class. Under-sampling techniques decrease the number of majority class, where the randomly chosen subset of majority class is combined with the minority class instances as the new training dataset. In this study, the effect of over-sampling and under-sampling techniques in medical datasets is examined. For the experimental study, several medical benchmark datasets and well-known classification algorithms are used.

International Conference on Advanced Technologies, Computer Engineering and Science
ICATCES

Mehmet Hacibeyoglu Mohammed Hussein IBRAHIM