Browsing by Subject "decision trees"

Now showing 1 - 2 of 2

Open Access
Automatic radar antenna scan analysis in electronic warfare
(2010) Eravcı, Bahaeddin
Estimation of the radar antenna scan period and recognition of the antenna scan type is usually performed by human operators in the Electronic Warfare (EW) world. In this thesis, we propose a robust algorithm to automatize these two critical processes. The proposed algorithm consists of two main parts: antenna scan period estimation and antenna scan type classification. The first part of the algorithm involves estimating the period of the signal using a time-domain approach. After this step, the signal is warped to a single vector with predetermined size (N) by resampling the data according to its period. This process ensures that the extracted features are reliable and are solely the result of the different scan types, since the effect of the different periods in the signal is removed. Four different features are extracted from the signal vector with an understanding of the phenomena behind the received signals. These features are used to train naive Bayes classifiers, decision-tree classifiers, artificial neural networks, and support vector machines. We have developed an Antenna Scan Pattern Simulator (ASPS) that simulates the position of the antenna beam with respect to time and generates synthetic data. These classifiers are trained and tested with the synthetic data and are compared by their confusion matrices, correct classification rates, robustness to noise, and computational complexity. The effect of the value of N and different signal-to-noise ratios (SNRs) on correct classification performance is investigated for each classifier. Decision-tree classifier is found to be the most suitable classifier because of its high classification rate, robustness to noise, and computational ease. Real data acquired by ASELSAN Inc. is also used to validate the algorithm. The results of the real data indicate that the algorithm is ready for deployment in the field and is capable of being robust against practical complications.
Open Access
A comparative study on human activity classification with miniature inertial and magnetic sensors
(2011) Yüksek, Murat Cihan
This study provides a comparative assessment on the different techniques of classifying human activities that are performed using body-worn miniature inertial and magnetic sensors. The classification techniques compared in this study are: naive Bayesian (NB) classifier, artificial neural networks (ANNs), dissimilarity-based classifier (DBC), various decision-tree methods, Gaussian mixture model (GMM), and support vector machines (SVM). The algorithms for these techniques are provided on two commonly used open source environments: Waikato environment for knowledge analysis (WEKA), a Java-based software; and pattern recognition toolbox (PRTools), a MATLAB toolbox. Human activities are classified using five sensor units worn on the chest, the arms, and the legs. Each sensor unit comprises a tri-axial gyroscope, a tri-axial accelerometer, and a tri-axial magnetometer. A feature set extracted from the raw sensor data using principal component analysis (PCA) is used in the classification process. Three different cross-validation techniques are employed to validate the classifiers. A performance comparison of the classification techniques is provided in terms of their correct differentiation rates, confusion matrices, and computational cost. The methods that result in the highest correct differentiation rates are found to be ANN (99.2%), SVM (99.2%), and GMM (99.1%). The magnetometer is the best type of sensor to be used in classification whereas gyroscope is the least useful. Considering the locations of the sensor units on body, the sensors worn on the legs seem to provide the most valuable information.