In this paper we present an unsupervised approach for intrusion detection based on spectral clustering (SC). Recently spectral clustering has gained wider application because of its promising results on several challenging clustering problems [1]. SC uses spectral graph theory to form a Laplacian matrix where the first k eigenvectors of this matrix are clustered using k-means to form representative clusters. The representative clusters are labeled normal or anomalous according to an assignment heuristic. We have provided different techniques to detect intrusions (or anomalies) which are scattered uniformly and form small clusters of anomalous data. To improve the clustering results, the scattered anomalies are detected and removed before representative clusters are formed using SC. For evaluation, a synthetic and real data set (KDD Cup 1999) are used and our results show that the application of SC is a promising approach to the development of an intrusion detection system. From the experiments we demonstrate that the application of SC yields a detection rate (DR) in the range of 91%-100% with the false positive rate (FPR) being less than 4.5%.

This paper presents a robust extended Kalman filter (REKF) applied to the simultaneous localization and mapping (SLAM) problem. Conventional Kalman filter methods suffer from the assumption of Gaussian noise statistics, which often lead to failures when these assumptions do not hold. Additionally, the linearization errors associated with the implementation of the standard EKF can also severely degrade the performance of the localization estimate. Currently, stochastic mapping provides a framework for the concurrent mapping of landmarks and localization of the vehicle with respect to the landmarks. However, the stochastic map is essentially an augmented EKF with the limitations thereof. This research addresses the linearization and Guassian assumption errors as they relate to the SLAM problem by proposing a new method, robust stochastic mapping. The robust stochastic map uses a Robust EKF (REKF) in order to address these limitations through the implementation of the bounded Hinfin norm. Experimental data are presented to illustrate the advantage of the localization using the proposed estimation procedure

This paper presents a data driven fault detection and identification (FDI) method using Support Vector Machines (SVM) and the Wavelet Packet Transform (WPT). The primary focus of this paper is to present a robust data driven fault diagnosis scheme. The investigated scheme has the capability to detect and identify faulty components of a given system through examination of its output due to a specified input. The use of the wavelet packet transformation serves to draw out those features of the output response which best characterize each of the fault classes for the various components. Support vector machines are used as the diagnosis phase to detect and isolate faults of a given system.

This paper presents a six degree-of-freedom nonlinear H-/sub /spl infin// controller for autonomous underwater vehicles. The controller makes use of Euler parameters, or unit quaternion, which have been used extensively with spacecraft and aircraft as opposed to the more popular Euler angle representation. Euler angles suffer as with other 3-parameter representations from singular points. One of the main challenges associated with designing an underwater control is the unknowns of the model. The nonlinear H-/sub /spl infin// controller attempts to address this problem such that the closed loop system has L/sub 2/ - gain /spl les/ /spl gamma/ from torque disturbances to a penalty variable for a fixed positive constant /spl gamma/.