Dechang Chen

Insider Attacker Detection in Wireless Sensor Networks

Though destructive to network functions, insider attackers are not detectable with only the classic cryptography-based techniques. Many mission-critic sensor network applications demand an effective, light, flexible algorithm for internal adversary identification with only localized information available. The insider attacker detection scheme proposed in this paper meets all the requirements by exploring the spatial correlation existent among the networking behaviors of sensors in close proximity. Our work is exploratory in that the proposed algorithm considers multiple attributes simultaneously in node behavior evaluation, with no requirement on a prior knowledge about normal/malicious sensor activities. Moreover, it is application-friendly, which employs original measurements from sensors and can be employed to monitor many aspects of sensor networking behaviors. Our algorithm is purely localized, fitting well to the large-scale sensor networks. Simulation results indicate that internal adversaries can be identified with a high accuracy and a low false alarm rate when as many as 25% sensors are misbehaving.

OPFKA: Secure and efficient Ordered-Physiological-Feature-based key agreement for wireless Body Area Networks

Body Area Networks (BANs) are expected to play a major role in patient health monitoring in the near future. Providing an efficient key agreement with the prosperities of plug-n-play and transparency to support secure inter-sensor communications is critical especially during the stages of network initialization and reconfiguration. In this paper, we present a novel key agreement scheme termed Ordered-Physiological-Feature-based Key Agreement (OPFKA), which allows two sensors belonging to the same BAN to agree on a symmetric cryptographic key generated from the overlapping physiological signal features, thus avoiding the pre-distribution of keying materials among the sensors embedded in the same human body. The secret features computed from the same physiological signal at different parts of the body by different sensors exhibit some overlap but they are not completely identical. To overcome this challenge, we detail a computationally efficient protocol to securely transfer the secret features of one sensor to another such that two sensors can easily identify the overlapping ones. This protocol possesses many nice features such as the resistance against brute force attacks. Experimental results indicate that OPFKA is secure, efficient, and feasible. Compared with the state-of-the-art PSKA protocol, OPFKA achieves a higher level of security at a lower computational overhead.

Selecting Genes by Test Statistics

Gene selection is an important issue in analyzing multiclass microarray data. Among many proposed selection methods, the traditional ANOVA F test statistic has been employed to identify informative genes for both class prediction (classification) and discovery problems. However, the F test statistic assumes an equal variance. This assumption may not be realistic for gene expression data. This paper explores other alternative test statistics which can handle heterogeneity of the variances. We study five such test statistics, which include Brown-Forsythe test statistic and Welch test statistic. Their performance is evaluated and compared with that of F statistic over different classification methods applied to publicly available microarray datasets.

Gene Expression Data Classification With Kernel Principal Component Analysis

One important feature of the gene expression data is that the number of genes M far exceeds the number of samples N. Standard statistical methods do not work well when N < M. Development of new methodologies or modification of existing methodologies is needed for the analysis of the microarray data. In this paper, we propose a novel analysis procedure for classifying the gene expression data. This procedure involves dimension reduction using kernel principal component analysis (KPCA) and classification with logistic regression (discrimination). KPCA is a generalization and nonlinear version of principal component analysis. The proposed algorithm was applied to five different gene expression datasets involving human tumor samples. Comparison with other popular classification methods such as support vector machines and neural networks shows that our algorithm is very promising in classifying gene expression data.

Sense through wall human detection using UWB radar

In this article, we discuss techniques for sense through wall human detection for different types of walls. We have focused on detection of stationary human target behind wall based on breathing movements. In detecting the breathing motion, a Doppler based method is used. Also a new approach based on short time Fourier transform is discussed and an already proposed clutter reduction technique based on singular value decomposition is applied to different measurements.

Fault-tolerant target localization in sensor networks

Fault-tolerant target detection and localization is a challenging task in collaborative sensor networks. This paper introduces our exploratory work toward identifying the targets in sensor networks with faulty sensors. We explore both spatial and temporal dimensions for data aggregation to decrease the false alarm rate and improve the target position accuracy. To filter out extreme measurements, the median of all readings in a close neighborhood of a sensor is used to approximate its local observation to the targets. The sensor whose observation is a local maxima computes a position estimate at each epoch. Results from multiple epoches are combined together to further decrease the false alarm rate and improve the target localization accuracy. Our algorithms have low computation and communication overheads. Simulation study demonstrates the validity and efficiency of our design.

Histologic grade remains a prognostic factor for breast cancer regardless of the number of positive lymph nodes and tumor size: A study of 161,708 cases of breast cancer from the SEER program

CONTEXT The appropriate staging of breast cancers includes an evaluation of tumor size and nodal status. Histologic grade in breast cancer, though important and assessed for all tumors, is not integrated within tumor staging. OBJECTIVE To determine whether the histologic grade remains a prognostic factor for breast cancer regardless of tumor size and the number of involved axillary lymph nodes. DESIGN By using a new clustering algorithm, the 10-year survival for every combination of T, N, and the histologic grade was determined for cases of breast cancer obtained from the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute. There were 36 combinations of TN, defined according to the American Joint Committee on Cancer, and grade. RESULTS For each combination of T and N, a categorical increase in the histologic grade was associated with a progressive decrease in 10-year survival regardless of the number of involved axillary lymph nodes or size of the primary tumor. Absolute survival differences between high and low grade persisted despite larger tumor sizes and greater nodal involvement, though trends were apparent with increasing breast cancer stage. Statistical significance depended on the number of cases for each combination. CONCLUSIONS Histologic grade continues to be of prognostic importance for overall survival despite tumor size and nodal status. Furthermore, these results seem to indicate that the assignment of the histologic grade has been consistent among pathologists when evaluated in a large data set of patients with breast cancer. The incorporation of histologic grade in TNM staging for breast cancer provides important prognostic information.

Developing Prognostic Systems of Cancer Patients by Ensemble Clustering

Accurate prediction of survival rates of cancer patients is often key to stratify patients for prognosis and treatment. Survival prediction is often accomplished by the TNM system that involves only three factors: tumor extent, lymph node involvement, and metastasis. This prediction from the TNM has been limited, because other potential prognostic factors are not used in the system. Based on availability of large cancer datasets, it is possible to establish powerful prediction systems by using machine learning procedures and statistical methods. In this paper, we present an ensemble clustering-based approach to develop prognostic systems of cancer patients. Our method starts with grouping combinations that are formed using levels of factors recorded in the data. The dissimilarity measure between combinations is obtained through a sequence of data partitions produced by multiple use of PAM algorithm. This dissimilarity measure is then used with a hierarchical clustering method in order to find clusters of combinations. Prediction of survival is made simply by using the survival function derived from each cluster. Our approach admits multiple factors and provides a practical and useful tool in outcome prediction of cancer patients. A demonstration of use of the proposed method is given for lung cancer patients.

Maximizing capacity with power control under physical interference model in duplex mode

This paper addresses the joint selection and power assignment of a largest set of given links which can communicate successfully at the same time under the physical interference model in the duplex (i.e. bidirectional) mode. For the special setting in which all nodes have unlimited maximum transmission power, Halldorsson and Mitra [5] developed an approximation algorithm with a huge constant approximation bound. For the general setting in which all nodes have bounded maximum transmission power, the existence of constant approximation algorithm remains open. In this paper, we resolve this open problem by developing an approximation algorithm which not only works for the general setting of bounded maximum transmission power, but also has a much smaller constant approximation bound.

Pattern recognition and string matching

Foreword. Correcting the Training Data R. Barandela, et al. Context Free Grammars and Semantic Networks for Flexible Assembly Recognition C. Bauckhage, G. Sagerer. Stochastic Recognition of Occluded Objects B. Bhanu, et al. Approximate String Matching for Angular String Elements with Applications to On-Line and Off-line Handwriting Recognition S.-H. Cha, S.N. Srihari. Uniform, Fast Convergence of Arbitrarily Tight Upper and Lower Bounds on the Bayes Error D. Chen, et al. Building RBF Networks for Time Series Classification by Boosting J.R. Diez, C.J.A. Gonzalez. Similarity Measures and Clustering of String Patterns A. Fred. Pattern Recognition for Intrusion Detection in Computer Networks G. Giacinto, F. Roli. Model-Based Pattern Recognition M. Haindl. Structural Pattern Recognition in Graphs L. Holder, et al. Deriving Pseudo-Probabilities of Correctness Given Scores (DPPS) K. Ianakiev, V. Govindaraju. Weighed Mean and Generalized Median of Strings Y. Jiang, H. Bunke. A Region-Based Algorithm for Classifier-Independent Feature Selection M. Kudo. Inference of K-Piecewise Testable Tree Languages D. Lopez, et al. Mining Partially Periodic Patterns With Unknown Periods From Event tream S. Ma, J.L. Hellerstein. Combination of Classifiers for Supervised Learning: A Survey S. Ma, C. Ji. Image Segmentation and Pattern Recognition: A Novel Concept, the Historgram of Connected Elements D. Maravell, M.A. Patricio. Prototype Extraction for k-NN Classifiers using Median Srings C.D. Martinez-Hinarejos, et al. Cyclic String Matching: Efficient Exact and Approximate Algorithms A. Marzal, et al. Homogeneity, Autocorrelation and Anisotropy in Patterns A. Molina. Robust Structural Indexing through Quasi-Invariant Shape Signatures and FeatureGeneration H. Nishida. Energy Minimisation Methods for Static and Dynamic Curve Matching E. Nyssen, et al. Recent Feature Selection Methods in Statistical Pattern Recognition P. Pudil, et al. Fast Image Segmentation under Noise R.M. Romano, D. Vitulano. Set Analysis of Coincident Errors and Its Applications for Combining Classifiers D. Ruta, B. Gabrys. Enhanced Neighbourhood Specifications for Pattern Classification J.S. San nchez, A.I. Marques. Algorithmic Synthesis in Neural Network Training for Pattern Recognition K. Sirlantzis. Binary Strings and multi-class learning problems T. Windeatt, R. Ghaderi.