IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING

Feature space high dimensionality is a well-known problem in text classification and web mining domains, it is caused mainly by the large number of vocabularies contained within web documents. Several methods were applied to select the most useful and important features over the years; however, the performance of such methods is still improvable from different aspects such as the computational cost and accuracy. This research presents an enhanced cosine similarity-based hybridization of two efficient feature selection methods for higher classification performance. The reduced feature sets are generated using the Random Projection (RP) and the Principal Component Analysis (PCA) methods, individually, then hybridized based on the cosine similarity values between features’ vectors. The performance of the proposed method in terms of accuracy and F-measure was tested on a dataset of web pages based on several term weighting schemes. As compared to relevant methods, results of the proposed method show significantly higher accuracy and f-measure performance based on less feature set size.

Face recognition is one of the most important research fields of the last
two decades. This is due to the actual use of this technology in automatic detection
and monitoring systems. Face attribute and features recognition from images is
still a challenge. In this paper, face image recognition is proposed upon local face
image rather than focusing on the whole image recognition by applying preprocessing
techniques and Viola-Jones method. Principal Component Analysis
(PCA) method is used in order to extract the needed features. Experiments show
satisfied and more accurate results achieved by the proposed system comparing to
the existing systems.

Abstract - This research proposes “Integrated Network Intrusion Detection System (INIDS)” which is NIDS for wire/wireless networks. INIDN consider features of the three layers; transport and Internet layers for wire and data link layer for wireless. The proposal is a Data Mining (DM)-based INIDS, which trained over a labeled wire and wireless datasets (each transaction labeled normal, intrusion name or unknown), INIDS is a hybrid IDS (anomaly and misuse). INIDS, train and construct two separated proposed models these are, Wire-NIDS and Wireless-NIDS then integrate the two models to build the final INIDS. Wire-NIDS use NSL-KDD dataset; use Principle Component Analysis (PCA) as a feature extraction, and use Support Vector Machine (SVM) with Artificial Neural Network (ANN) as classifiers. Wireless-NIDS use proposed Wdataset dataset, use Gain Ratio (GR) as feature selection, and use Naïve Bayesian (NB) as a classifier. The results obtained from executing the proposed INIDS model showing that Wire-NIDS and Wireless-NIDS classifier accuracy and detection rate is generally higher with the subset of features obtained by PCA (8 from 41) and GR (8 from 17) than with all sets of features. Proposed confusion matrix of INIDS gives less confusion in detection rates with reduced features.

Keywords: IDS, SVM, ANN, NB, PCA, and GR.