Detecting network intruders and malicious software is a significant problem for network administrators and security experts. New threats are emerging at an increasing rate, and current signature and statistics-based techniques are failing to keep pace. Intelligent systems that can adapt to new threats are needed to mitigate these new strains of malware as they are released. This research develops a system that uses contextual relationships and information across different layers of abstraction to detect malware based on its qualia, or essence. By looking for the underlying concepts that make a piece of software malicious, this system avoids the pitfalls of static solutions that focus on predefined signatures or anomaly thresholds. This type of qualia-based system provides a framework for developing intelligent classification and decision-making systems for any number of application areas.

Identity capabilities are being modernized through biometric technology and systems. With maturing biometrics in full, rapid development, a higher accuracy of identity verification is required. Improvements to the security of biometric verification systems is provided through higher accuracy; ultimately reducing fraud, theft, and loss of resources from unauthorized personnel. With trivial biometric systems, a higher acceptance threshold to obtain higher accuracy rates increased false rejection rates and user unacceptability. However, maintaining the higher accuracy rate enhances the security of the system. Through the methods presented in this paper, higher accuracy rates are obtained without lowering the acceptance threshold, therefore improving the security level, false rejection rates, and user acceptability. An area of biometrics with a paucity of research is template aging, specifically in regards to facial aging. This paper presents methods of modeling and predicting facial template aging based on matching score analysis. A novel foundational framework for facial template aging is presented and provides a methodological framework. The groundwork discusses new techniques used in the template aging framework, to include the ldquoerror score matrixrdquo and ldquodecay error estimaterdquo concepts. The matching scores are calculated using commercially available facial matching algorithms/SDKs against publicly available facial databases. Improved performance error rates while maintaining or improving upon the overall matching and/or rejection levels is accomplished with the new framework. Using such scores, prediction of a timeframe if or when an individual needs to be re-enrolled with a new template is feasible.

LADAR provides 3D shape information that has yet to be fully exploited for object recognition or classification. This is partly due to the operating conditions, but mostly due to a representational gap in computational intelligence. This paper briefly explores some of the hurdles of object classification using LADAR data and proposes a theoretical framework, based on the biological inspiration of qualia, we believe will allow us to address these operating conditions and, most importantly, this representational gap. Our framework works on concepts instead of parts, and iterates a top-down and bottom-up solution that updates the hypothesis with the accrual of evidence. This creates a system that we believe will generalize concepts, learn from experience, and even recognize the need for the addition of new classes based on its current world view.

A physiologically motivated human color visual system model which represents visual information with one brightness component (A) and two chromatic components (C/sub 1/ and C/sub 2/) is used to create stimuli for testing the color perception of deuteranomalous trichromats. Two experiments are performed. Using simple ramp patterns, the first experiment finds that three deuteranomalous trichromat test subjects can distinguish variations only in the C/sub 2/ component of the color vision model. This finding is further tested in the second experiment: a set of paired comparison preference tests. Two altered versions of each of three natural color images are prepared by setting either one of the color components to a constant over the full image. Pairs of an original and a distorted image are presented to the test subjects, and they are asked to indicate which image they prefer. The experimental results indicate that the C/sub 1/ channel is severely attenuated in the deuteranomalous trichromat test subjects, and that nearly all their color sensation is mediated by the C/sub 2/ channel of the color vision model.

This paper presents the first physiologically motivated pulse coupled neural network (PCNN)-based image fusion network for object detection. Primate vision processing principles, such as expectation driven filtering, state dependent modulation, temporal synchronization, and multiple processing paths are applied to create a physiologically motivated image fusion network. PCNN's are used to fuse the results of several object detection techniques to improve object detection accuracy. Image processing techniques (wavelets, morphological, etc.) are used to extract target features and PCNN's are used to focus attention by segmenting and fusing the information. The object detection property of the resulting image fusion network is demonstrated on mammograms and Forward Looking Infrared Radar (FLIR) images. The network removed 94% of the false detections without removing any true detections in the FLIR images and removed 46% of the false detections while removing only 7% of the true detections in the mammograms. The model exceeded the accuracy obtained by any individual filtering methods or by logical ANDing the individual object detection technique results.

A model of the human color vision system is used to construct constant brightness color Mach band stimuli along two chromatic axes for display on a standard workstation color monitor. Chromatic (not brightness) Mach bands appear in the displayed stimuli as over- and under-shoots along the axis of the varied component. The stimuli are displayed on several different but similar monitors without affecting the appearance of the Mach bands. Anomalous trichromats see no change in color when presented with a stimulus containing variations along only one of the two chromatic axes.

Three hierarchical multiresolution image fusion techniques are implemented and tested using image data from the Airborne Visual/Infrared Imaging Spectrometer (AVIRIS) hyperspectral sensor. The methods presented focus on combining multiple images from the AVIRIS sensor into a smaller subset of images white maintaining the visual information necessary for human analysis. Two of the techniques are published algorithms that were originally designed to combine images from multiple sensors, but are shown to work well on multiple images from the same sensor. The third method presented was developed specifically to fuse hyperspectral images for visual analysis. This new method uses the spatial frequency response (contrast sensitivity) of the human visual system to determine which features in the input images need to be preserved in the composite image(s) thus ensuring the composite image maintains the visually relevant features from each input image. The image fusion algorithms are analyzed using test images with known image characteristics and image data from the AVIRIS hyperspectral sensor. After analyzing the signal-to-noise ratios and visual aesthetics of the fused images, contrast sensitivity based fusion is shown to provide excellent fusion results and, in every case, outperformed the other two methods.

A new model-based vision (MBV) algorithm is developed to find regions of interest (ROI's) corresponding to masses in digitized mammograms and to classify the masses as malignant/benign. The MBV algorithm is comprised of five modules to structurally identify suspicious ROI's, eliminate false positives, and classify the remaining as malignant or benign. The focus of attention module uses a difference of Gaussians (DoG) filter to highlight suspicious regions in the mammogram. The index module uses tests to reduce the number of nonmalignant regions from 8.39 to 2.36 per full breast image. Size, shape, contrast, and Laws texture features are used to develop the prediction module's mass models. Derivative-based feature saliency techniques are used to determine the best features for classification. Nine features are chosen to define the malignant/benign models. The feature extraction module obtains these features from all suspicious ROI's. The matching module classifies the regions using a multilayer perceptron neural network architecture to obtain an overall classification accuracy of 100% for the segmented malignant masses with a false-positive rate of 1.8 per full breast image. This system has a sensitivity of 92% for locating malignant ROI's. The database contains 272 images (12 b, 100 microm) with 36 malignant and 53 benign mass images. The results demonstrate that the MBV approach provides a structured order of integrating complex stages into a system for radiologists.

Automatic speaker recognition systems are maturing and databases have been designed to specifically compare algorithms and results to target error rates. The LDC YOHO speaker verification database was designed to test error rates at the 1% false rejection and 0.1% false acceptance level. This work examines the use of speaker-dependent (SD) monophone models to meet these requirements. By representing each speaker with 22 monophones, both closed-set speaker identification and global-threshold verification was performed. Using four combination lock phrases, speaker identification error rates are obtained at 0.19% for males and 0.31% for females. By defining a test hypothesis, a critical error analysis for speaker verification is developed and new results reported for YOHO. A new Bhattacharyya distance is developed for cohort selection. This method, based on the second order statistics of the enrolment Viterbi log-likelihoods, determines the optimal cohorts and achieves an equal error rate of 0.282%.

In this paper, we present an integrated approach to feature and architecture selection for single hidden layer-feedforward neural networks trained via backpropagation. In our approach, we adopt a statistical model building perspective in which we analyze neural networks within a nonlinear regression framework. The algorithm presented in this paper employs a likelihood-ratio test statistic as a model selection criterion. This criterion is used in a sequential procedure aimed at selecting the best neural network given an initial architecture as determined by heuristic rules. Application results for an object recognition problem demonstrate the selection algorithm's effectiveness in identifying reduced neural networks with equivalent prediction accuracy.

The major goal of this research was to investigate speech coding techniques in an attempt to achieve high quality speech transmittable at 4800 bits per second. The approach taken to achieve this goal was to code the frequency domain representation of speech. Speech was represented by a sparse set of frequency components. Four frequency selections schemes were implemented, and the resulting frequency coefficients (magnitude and phase) were coded in an efficient manner for transmission. Specific techniques involved in the speech coder included: (1) a recurrent neural architecture to make a periodic/noiselike decision, (2) the use of variable length windows for analysis and synthesis, and (3) a representation of noiselike speech using frequency banded energy information. The quality of the reconstructed speech was tested using listening tests which compared the different frequency selection schemes, along with original and sampled speech. The system did not achieve 'toll quality' speech; however, the resulting speech was highly intelligible. Specific quality degradation was noted at window transitions.

The relationship between backpropagation and extended Kalman filtering for training multilayer perceptrons is examined. These two techniques are compared theoretically and empirically using sensor imagery. Backpropagation is a technique from neural networks for assigning weights in a multilayer perceptron. An extended Kalman filter can also be used for this purpose. A brief review of the multilayer perceptron and these two training methods is provided. Then, it is shown that backpropagation is a degenerate form of the extended Kalman filter. The training rules are compared in two examples: an image classification problem using laser radar Doppler imagery and a target detection problem using absolute range images. In both examples, the backpropagation training algorithm is shown to be three orders of magnitude less costly than the extended Kalman filter algorithm in terms of a number of floating-point operations.>

A concept is proposed for utilizing artificial neural networks to enhance the high-speed tracking accuracy of robotic manipulators. Tracking accuracy is a function of the controller's ability to compensate for disturbances produced by dynamical interactions between the links. A model-based control algorithm uses a nominal model of those dynamical interactions to reduce the disturbances. The problem is how to provide accurate dynamics information to the controller in the presence of payload uncertainty and modeling error. Neural network payload estimation uses a series of artificial neural networks to recognize the payload variation associated with a degradation in tracking performance. The network outputs are combined with a knowledge of nominal dynamics to produce a computationally efficient direct form of adaptive control. The concept is validated through experimentation and analysis on the first three links of a PUMA-560 manipulator. A multilayer perceptron architecture with two hidden layers is used. Integration of the principles of neural network pattern recognition and model-based control produces a tracking algorithm with enhanced robustness to incomplete dynamic information. Tracking efficacy and applicability to robust control algorithms are discussed.

The problem of fusing information from multiple sensors to perform pattern recognition is addressed. A technique is proposed for performing target/nontarget discrimination using information from absolute range and forward looking infrared (FLIR) sensors. A multilayer perceptron is used to perform the feature-level fusion and is compared to ak-nearest-neighbor classifier as well as a nonparametric Bayesian classifier using Parzen windows. All three classifiers show statistically significant improvement from the fusion process; however, the multilayer perceptron uses far fewer free parameters, which should improve generalization capability. Also, the multilayer perceptron requires much less computation than the traditional classifiers

The multilayer perceptron, when trained as a classifier using backpropagation, is shown to approximate the Bayes optimal discriminant function. The result is demonstrated for both the two-class problem and multiple classes. It is shown that the outputs of the multilayer perceptron approximate the a posteriori probability functions of the classes being trained. The proof applies to any number of layers and any type of unit activation function, linear or nonlinear.

Classical optical information processing and classical neural networks can be adapted and combined to create optical neural networks which offer significant and fundamental advantages over electronic neural networks in various well-defined cases. A systematic morphology of optical neural networks is presented. Special problems they create are discussed. The state of the art of their implementation is indicated, and some supportable speculations on their future are given.>