Downloads & Free Reading Options - Results

Previous research found that operators prefer to have status, alerts, and controls located on the same screen. Unfortunately, that research was done with displays that were not designed specifically for collocation. In this experiment, twelve subjects evaluated two displays specifically designed for collocating system information against a baseline that consisted of dial status displays, a separate alert area, and a controls panel. These displays differed in the amount of collocation, pattern matching, and parameter movement compared to display size. During the data runs, subjects kept a randomly moving target centered on a display using a left-handed joystick and they scanned system displays to find a problem in order to correct it using the provided checklist. Results indicate that large parameter movement aided detection and then pattern recognition is needed for diagnosis but the collocated displays centralized all the information subjects needed, which reduced workload. Therefore, the collocated display with large parameter movement may be an acceptable display after familiarization because of the possible pattern recognition developed with training and its use.

A method for real-time on-board spacecraft attitude determination is presented. The method is suitable for processing CCD or CID stellar camera and rate gyro output to determine orientation to better than five arc seconds. The system is self-calibrating; uncertain parameters such as interlock angles and gyro biases can be included in the estimation algorithms. The system is implemented in a microcomputer system which conclusively establishes that the system is compatible with on-board computation constraints. (Author)

A large memory optical recognition system has been developed and tested. The memory consists of an array of stored holographic Fourier transform matched filters in a VanderLugt type correlator. The filters are stored (on high resolution Kodak plates) and addressed using a novel holographic multi-focus lens. This element acts as a diffraction grating, splitting the input beam into 25 elements and as a lens, producing the Fourier transform of each of the 25 elements. The filters are created and addressed using a HeNe laser and a Hughes liquid crystal light valve (LCLV) to produce a coherent image from a television monitor. The LCLV also allows the filters to be addressed in real time thereby making the use of a transparency as an input scene unnecessary. This provides for real time recognition. Thus far, more than 70 matched filters have been stored on a single holographic plate and addressed in parallel and in real time with good correlations resulting. This research and the results obtained should contribute to allowing serious consideration to be given to the use of optical recognition systems for a wide variety of civilian and military applications. (Author)

The author has identified the following significant results. Uncontrolled photo mosaics of ERTS-1 imagery using MSS band 5 and 7 at a scale of 1:1,000,000 were used to make a preliminary surficial geology map in northcentral Alaska. Seven distinct geologic units were recognized, defined, and mapped directly on a photo mosaic. Results are closely correlated with published surficial geology maps. Eight MSS images were examined to test utility of ERTS data in studies of coastal processes and stream hydrology, and in the identification and interpretation of geomorphic features throughout Alaska. The feasibility of using ERTS-1 data to map structural lineaments is well illustrated on a mosaic of 8, band 5 MSS images. Along the northern edge of the Brooks Range one lineament can be followed the entire width of the mosaic, a distance of 225 miles. Two nearly parallel lineaments can be seen running along the northern and southern edges of the Schwatka Mountains. About 135 miles south of these two lineaments another series located in the Chitanana River region can be followed for 45 miles. These lineaments appear to be faults, and it is interesting to note that the Yukon River parallels these and appears to be structurally controlled.

Infrared spectra of organophosphorus compounds, including pesticides and a set of neurotoxins which have been banned from use by international agreement, along with their precursors and hydrolysis products, were obtained from a variety of sources. The data were treated to minimize spectral information related to the spectral origin. A common spectral wavelength range was selected and spectral data within this range were transduced into data vectors. Computer-assisted classification tools were used to classify the spectra as pesticides versus neurotoxins and their precursors and hydrolysis products. The performance of a k-nearest neighbor classifier for this distinction is compared with several artificial neural network classifiers.

A method for real-time on-board spacecraft attitude determination is presented. The method is suitable for processing CCD or CID stellar camera and rate gyro output to determine orientation to better than five arc seconds. The system is self-calibrating; uncertain parameters such as interlock angles and gyro biases can be included in the estimation algorithms. The system is implemented in a microcomputer system which conclusively establishes that the system is compatible with on-board computation constraints. (Author)

A method for real-time on-board spacecraft attitude determination is presented. The method is suitable for processing CCD or CID stellar camera and rate gyro output to determine orientation to better than five arc seconds. The system is self-calibrating; uncertain parameters such as interlock angles and gyro biases can be included in the estimation algorithms. The system is implemented in a microcomputer system which conclusively establishes that the system is compatible with on-board computation constraints. (Author)

This Russian patent claim is for a method of pattern recognition consisting in the modulation of a light flux with subsequent integration that is different because in order to increase the precision of recognition, the light flux is modulated according to the dispersion function on the X-axis, it is integrated on this same axis, and it is modulated according to the dispersion function on the Y-axis.

Effort on Phase I tasks continued and good progress was made during this The hardware design of the CNAPS/PC board was completed and five (5) prototype boards were fabricated. Preliminary debug of the boards has been successfully completed. Very good progress was made on the software port and debug of the CNAPS-C compiler and the source-level debugger. Progress was made, but not as much as planned, on the software port and debug of the OEMlib control software to the CNAPS/PC hardware and the DOS/Windows environment. At the time of this report, the project has expenditures and commitments totaling 66% of the funds allocated for Phase I of the contract. Pattern Recognition Image Processing

Typical multispectral/hyperspectral image (MSI/HSI) data analysis focuses on single pixel-at-a-time analysis and pattern matching to known or trained spectral signatures. Nearest neighbor and object feature classification techniques are not fully exploited through traditional MSI/HSI approaches. Inputs from multiple sensors of varying formats, spectral resolution, and spatial resolution are only beginning to be fused together.

This paper discusses some of the positions, theories, and experiments which compose the adaptive hemispheric processing area. This discussion addresses some of the basic questions of human learning and relates some of the assumptions that are critical for interpreting research findings. The paper begins by looking at some theoretical perspectives, then looks at some of the author's own research application in hemispheric processing within the area of face recognition and attentional processes. A neuro-cognitive systems model of results is proposed to interpret findings. Keywords:Brain hemispheric asymmetry; Visual recognition; Learning and cognition; Cerebral cortex; Human performance.

Spatial Light Modulators (SLMs) are key elements in real-time optical signal processors. The performance limitations of SLMs have significantly hampered the application of optical pattern recognition systems, such as optical correlators, to Real World problems. Exploration of nonlinear transformations in the Fourier plane of optical processors, such as correlators is of considerable interest. Computer simulations have been extensively reported. Several optically addressed SLMs responses, proposed for use in nonlinear transformation processes, have not been characterized. This report provides experimental data on candidate SLMs: Hughes LCLV, GEC-Marconi SLM, and the University of Colorado- Boulder ferroelectric liquid crystal SLM. Data reported are variations in the driving waveform frequency and amplitude, incident read beam intensity, and spectral variations in the read beam. Equivalent circuit models have been developed to explain modulator response. The nonlinear transformation function of optically addressed SLMs may be used to improve optical processor performance for use as correlators.

A method is proposed that is the optimal coding of objects in the classification of complex systems. The determination of the codes is reduced to the solution of a problem of linear programming. The obtained table of codes can be used, for example, for determining the oil content of beds from the data of geophysical measurements. The methods of pattern recognition are widely used not only for the identification of visual forms, but also for the classification of complex systems of different types. Criteria are either results of quantitative measurements or qualitative evaluations. It is possible to assume that each parameter has a finite number of gradations.

Wavelet transform (WT) has recently drawn the attention of the researchers due to its potential in electromyography (EMG) recognition system. However, the optimal mother wavelet selection remains a challenge to the application of WT in EMG signal processing. This paper presents a detail study for different mother wavelet function in discrete wavelet transform (DWT) and continuous wavelet transform (CWT). Additionally, the performance of different mother wavelet in DWT and CWT at different decomposition level and scale are also investigated. The mean absolute value (MAV) and wavelength (WL) features are extracted from each CWT and reconstructed DWT wavelet coefficient. A popular machine learning method, support vector machine (SVM) is employed to classify the different types of hand movements. The results showed that the most suitable mother wavelet in CWT are Mexican hat and Symlet 6 at scale 16 and 32, respectively. On the other hand, Symlet 4 and Daubechies 4 at the second decomposition level are found to be the optimal wavelet in DWT. From the analysis, we deduced that Symlet 4 at the second decomposition level in DWT is the most suitable mother wavelet for accurate classification of EMG signals of different hand movements. 

A model of associative memory is studied, which stores and reliably retrieves many more patterns than the number of neurons in the network. We propose a simple duality between this dense associative memory and neural networks commonly used in deep learning. On the associative memory side of this duality, a family of models that smoothly interpolates between two limiting cases can be constructed. One limit is referred to as the feature-matching mode of pattern recognition, and the other one as the prototype regime. On the deep learning side of the duality, this family corresponds to feedforward neural networks with one hidden layer and various activation functions, which transmit the activities of the visible neurons to the hidden layer. This family of activation functions includes logistics, rectified linear units, and rectified polynomials of higher degrees. The proposed duality makes it possible to apply energy-based intuition from associative memory to analyze computational properties of neural networks with unusual activation functions - the higher rectified polynomials which until now have not been used in deep learning. The utility of the dense memories is illustrated for two test cases: the logical gate XOR and the recognition of handwritten digits from the MNIST data set.

A model of associative memory is studied, which stores and reliably retrieves many more patterns than the number of neurons in the network. We propose a simple duality between this dense associative memory and neural networks commonly used in deep learning. On the associative memory side of this duality, a family of models that smoothly interpolates between two limiting cases can be constructed. One limit is referred to as the feature-matching mode of pattern recognition, and the other one as the prototype regime. On the deep learning side of the duality, this family corresponds to feedforward neural networks with one hidden layer and various activation functions, which transmit the activities of the visible neurons to the hidden layer. This family of activation functions includes logistics, rectified linear units, and rectified polynomials of higher degrees. The proposed duality makes it possible to apply energy-based intuition from associative memory to analyze computational properties of neural networks with unusual activation functions - the higher rectified polynomials which until now have not been used in deep learning. The utility of the dense memories is illustrated for two test cases: the logical gate XOR and the recognition of handwritten digits from the MNIST data set.

Thesis advisor, Tri T. Ha

Several of the techniques used in pattern recognition are reformulated as the problem of determining fixed points of a function. If x sub 0 is a fixed point of f and if f is contractive at x sub 0, then, for any y belonging to a sufficiently small neighborhood of x sub 0 the orbit of y will converge to x sub 0. Several general results regarding contractive mappings are developed with emphasis on functions.

The application of mathematical and statistical analyses techniques to imagery obtained by remote sensors is described by Principal Investigators. Scene-to-map registration, geometric rectification, and image matching are among the pattern recognition aspects discussed.

In Phase I of this project, Physical Optics Corporation (POC) accomplished the goal of the original proposal which was to develop and optimize a unique neural network (NN) algorithm that performs rapid spectral signal processing and identification. POC's NN algorithm was tested with extremely noisy Raman spectra from Lawrence Livermore National Laboratory and experimentally showed at least ten times better sensitivity and reliability than conventional spectral signal processing methods. POC built a portable demonstration system with POC's NN and successfully demonstrated real-time spectral signature identification operations. POC proposed, for Phase II implementation, a holographic optical neural network (HONN) system that is capable of rapid hyperspectral imaging through an acoustic-optic tunable filter (AOTF), real-time spectral feature identification, and mapping. The success of the Phase II project will make automatic and rapid hyperspectral image analysis and feature location possible. Neural networks, Holographic optical spectral feature identification, Portable smart spectrometer, Hyperspectral image processing.

Previous detection studies assumed that a 'target' was either present in a region of interest with a probability p, or not, with a probability (1-p). Knowing the value of p, the 'searcher' makes a measurement of a random variable that has a probability density function which depends on whether or not the target is present and attempts to make a decision regarding the presence of the target. In this paper a more general point of view is adopted in that the author allows the possibility of both target and non-target elements to be present in any region of measurement. This provides an explicit consideration of a third possibility, that of a mixture of target and non-target elements. The forms of the probability densities for the 'mix' variate are obtained for a variety of assumptions regarding the relationships among the variates involved. A sequential decision problem is extended to include the three value case.

Previous detection studies assumed that a 'target' was either present in a region of interest with a probability p, or not, with a probability (1-p). Knowing the value of p, the 'searcher' makes a measurement of a random variable that has a probability density function which depends on whether or not the target is present and attempts to make a decision regarding the presence of the target. In this paper a more general point of view is adopted in that the author allows the possibility of both target and non-target elements to be present in any region of measurement. This provides an explicit consideration of a third possibility, that of a mixture of target and non-target elements. The forms of the probability densities for the 'mix' variate are obtained for a variety of assumptions regarding the relationships among the variates involved. A sequential decision problem is extended to include the three value case.

A method for real-time on-board spacecraft attitude determination is presented. The method is suitable for processing CCD or CID stellar camera and rate gyro output to determine orientation to better than five arc seconds. The system is self-calibrating; uncertain parameters such as interlock angles and gyro biases can be included in the estimation algorithms. The system is implemented in a microcomputer system which conclusively establishes that the system is compatible with on-board computation constraints. (Author)

The pattern recognition code ARTHUR was tested on a limited cratering data base to examine its utility in studying complex continuous data. The code was found to be a convenient data analysis package but possessed no continuous data capability that is not already available to any researcher. (Author)

The pattern recognition code ARTHUR was tested on a limited cratering data base to examine its utility in studying complex continuous data. The code was found to be a convenient data analysis package but possessed no continuous data capability that is not already available to any researcher. (Author)

The pattern recognition code ARTHUR was tested on a limited cratering data base to examine its utility in studying complex continuous data. The code was found to be a convenient data analysis package but possessed no continuous data capability that is not already available to any researcher. (Author)

The program has evaluated the toxicity of two complex mixtures, the water soluble fractions (WSF) of the commercial turbine fuel Jet-A and the military fuel JP-4 using single species toxicity tests as well as the Standard Aquatic Microcosm (SAM). The WSF were not particularly toxic to the algal species tested although toxicity was observed when Daphnia magna was used as the test organism. The SAM experiments have been completed using concentrations of 0.0, 1, 5 and 15 percent WSF. Among the more interesting effects were the shifts in time of population peaks and some other variables compared to controls. Regression analysis of control to treatment groups often demonstrated only weak correlations. Multivariate nonmetric clustering (NMC) analysis, however, also demonstrated a marked separation between the 4 treatment groups for the Jet-A experiment. NMC proved to be the most powerful multivariate method of those examined for distinguishing the control and other treatment groups. An additional research effort is focused on applying multivariate methods and other mathematical techniques into the process of ecological risk assessment. Application of multivariate methods coupled with new ways of distinguishing uncertainty have the potential for revolutionizing the risk assessment process. Jet fuel, Non-metric clustering, Microcosms, Risk assessment.

For application to chemical sensor arrays, the ideal pattern recognition is accurate, fast, simple to train, robust to outliers, has low memory requirements, and has the ability to produce a measure of classification certainty. In this work, four data sets representing typical chemical sensor array data were used to compare seven pattern recognition algorithms nearest neighbor, Mahalanobis linear discriminant analysis, Bayesian linear discriminant analysis, SIMCA, back propagation neural networks, probabilistic neural networks (PNN), and learning vector quantization (LVQ) for their ability to meet the criteria. LVQ and PNN exhibited high classification accuracy and met many of the qualitative criteria for an ideal algorithm. Based on these results, a new algorithm (LVQ-PNN) that incorporates the best features of PNN and LVQ was developed. The LVQ-PNN algorithm was further improved by the addition of a faster training procedure. It was then compared with the other seven algorithms. The LVQ-PNN method achieved excellent classification performance. A general procedure for selecting the optimal rejection threshold for a PNN based algorithm using Monte Carlo simulations also was demonstrated. This outlier rejection strategy was implemented for an LVQ-PNN classifier and found consistently to reject ambiguous patterns.

Computerized cognitive assessment batteries provide potentially valuable methods of assessing cognitive performance in field settings, but it often is difficult to follow recommendations for extensive baseline testing in these settings. The performance of male Norwegian military personnel (n = 36) on computerized tests of pattern recognition, memory, and spatial orientation tasks was studied to replicate and extend prior findings suggesting that valid measurements of cognitive functions can be obtained even when these recommendations cannot be followed. Structural equation models replicated prior findings that these cognitive tests produce performance measures with constant true score variance from the first session onward and constant error variance from the second session onward. Cross-validation of previously developed structural models indicated comparable error variances across samples, but differences in true score variance were observed which presumably represented sampling variance in the underlying cognitive abilities. These cognitive tests can produce valid results in field studies even when baseline sessions must be limited or omitted if the experimental design and analysis procedures control for other threats to validity (e.g. learning effects). (AN)

Computerized cognitive assessment batteries provide potentially valuable methods of assessing cognitive performance in field settings, but it often is difficult to follow recommendations for extensive baseline testing in these settings. The performance of male Norwegian military personnel (n = 36) on computerized tests of pattern recognition, memory, and spatial orientation tasks was studied to replicate and extend prior findings suggesting that valid measurements of cognitive functions can be obtained even when these recommendations cannot be followed. Structural equation models replicated prior findings that these cognitive tests produce performance measures with constant true score variance from the first session onward and constant error variance from the second session onward. Cross-validation of previously developed structural models indicated comparable error variances across samples, but differences in true score variance were observed which presumably represented sampling variance in the underlying cognitive abilities. These cognitive tests can produce valid results in field studies even when baseline sessions must be limited or omitted if the experimental design and analysis procedures control for other threats to validity (e.g. learning effects). (AN)

Computerized cognitive assessment batteries provide potentially valuable methods of assessing cognitive performance in field settings, but it often is difficult to follow recommendations for extensive baseline testing in these settings. The performance of male Norwegian military personnel (n = 36) on computerized tests of pattern recognition, memory, and spatial orientation tasks was studied to replicate and extend prior findings suggesting that valid measurements of cognitive functions can be obtained even when these recommendations cannot be followed. Structural equation models replicated prior findings that these cognitive tests produce performance measures with constant true score variance from the first session onward and constant error variance from the second session onward. Cross-validation of previously developed structural models indicated comparable error variances across samples, but differences in true score variance were observed which presumably represented sampling variance in the underlying cognitive abilities. These cognitive tests can produce valid results in field studies even when baseline sessions must be limited or omitted if the experimental design and analysis procedures control for other threats to validity (e.g. learning effects). (AN)

If V(R) is the vertex sequence of a symmetric cycle R in the tope graph of a simple acyclic oriented matroid M on a t-element ground set, then the set min V(R) of minimal elements in the subposet V(R) of the tope poset of M, based at the positive tope, is a critical committee for M that votes for the base tope. We consider the sequence zR of poset ranks of the elements from the vertex sequence of R as a fragment of a signal with period 2t and relate the number of members of the committee min V(R) to the magnitudes of [t/2] components, with odd indices, of the discrete Fourier transform of the distance vector zR.

Pattern recognition and machine learning are becoming integral parts of algorithms in a wide range of applications. Different algorithms and approaches for machine learning include different tradeoffs between performance and computation, so during algorithm development it is often necessary to explore a variety of different approaches to a given task. A toolbox with a unified framework across multiple pattern recognition techniques enables algorithm developers the ability to rapidly evaluate different choices prior to deployment. MATLAB is a widely used environment for algorithm development and prototyping, and although several MATLAB toolboxes for pattern recognition are currently available these are either incomplete, expensive, or restrictively licensed. In this work we describe a MATLAB toolbox for pattern recognition and machine learning known as the PRT (Pattern Recognition Toolbox), licensed under the permissive MIT license. The PRT includes many popular techniques for data preprocessing, supervised learning, clustering, regression and feature selection, as well as a methodology for combining these components using a simple, uniform syntax. The resulting algorithms can be evaluated using cross-validation and a variety of scoring metrics to ensure robust performance when the algorithm is deployed. This paper presents an overview of the PRT as well as an example of usage on Fisher's Iris dataset.

The standard Super-Kamiokande analysis uses an estimator for particle identification by which it discriminates electrons (electron nutrinos) from muons (muon nutrinos). Use of this estimator has led to the claim of a significant deficiency of muons (muon nutrinos), suggesting the existence of neutrino oscillations. We investigate three areas of concern for the Super-Kamiokande estimator: the separation of the spatial part from the angular part in the probability functions, the neglect of fluctuations in the Cherenkov light in different physical processes due to the charged particles concerned, and the point-like approximation for the emission of Cherenkov light. We show that the first two factors are important for the consideration of stochastic processes in the generation of the Cherenkov light, and that the point-like assumption oversimplifies the estimation of the Cherenkov light quantities. We develop a new discrimination procedure for separating electron neutrinos from muon neutrinos, based on detailed simulations carried out with GEANT~3.21 and with newly derived mean angular distribution functions for the charged particles concerned (muons and electrons/positrons), as well as the corresponding functions for the relative fluctuations. These angular distribution functions are constructed introducing a ``moving point'' approximation. The application of our procedure between the discrimination between electron and muon to the analysis of the experimental data in SK will be made in a subsequent paper.

PASM, a Large-scale multimicroprocessor system being designed at Purdue University for image processing and pattern recognition, is described. This system can be dynamically reconfigured to operate as one or more independent SIMD and/or MIMD machines. PASM consists of a Parallel Computation Unit, which contains N processors, N memories, and an interconnection network; Q Micro Controllers, each of which controls N/Q processors; N/Q parallel secondary storage devices; a distributed Memory Management System; and a System Control Unit, to coordinate the other system components. Possible values for N and Q are 1024 and 16, respectively. The control schemes, interprocessor communications, and memory management in PASM are explored. Examples of how PASM can be used to perform image processing tasks are given. (Author)

A method for real-time on-board spacecraft attitude determination is presented. The method is suitable for processing CCD or CID stellar camera and rate gyro output to determine orientation to better than five arc seconds. The system is self-calibrating; uncertain parameters such as interlock angles and gyro biases can be included in the estimation algorithms. The system is implemented in a microcomputer system which conclusively establishes that the system is compatible with on-board computation constraints. (Author)

The goal of the work is to develop pattern-recognition and signal-processing method that will use Army-supplied human cardiovascular and psychomotor data to provide indices of responsivity to challenge. Time Series and point-process techniques will form the basis of the approach, and the assumptions that underlie the methods will be examined and tested. The detection of infrequent and brief events, and the ellucidation of their relationships, if any, to the indices, will be carried out.

The goal of the work is to develop pattern-recognition and signal-processing method that will use Army-supplied human cardiovascular and psychomotor data to provide indices of responsivity to challenge. Time Series and point-process techniques will form the basis of the approach, and the assumptions that underlie the methods will be examined and tested. The detection of infrequent and brief events, and the ellucidation of their relationships, if any, to the indices, will be carried out.

This article, on the foundation of circular harmonic filter algorithms, combines synthetic discrimination functions to put forward circular harmonic synthetic filters. It solves traditional problems of matched space filters with input target geometrical distortions. On synthetic filters, computer simulations and optical correlation experiments were carried out. Results clearly showed that circular harmonic synthetic filters possess relatively strong triple invariant optical pattern recognition capabilities. (AN)

The primary goal of this research is to study a programmable joint- transform correlator (JTC) using liquid crystal television (LCTV) panels for adaptive real-time pattern recognition applications. The technique can improve the pattern recognition and identification technology that is of interest to the U.S. Army. The technique we studied is a real-time programmable electro-optical architecture. There are several reasons for selecting the optical technique over their digital and electronic counterparts, as follows: Optical technique is capable of handling a large space-bandwidth image; optical technique is capable of performing parallel operations; optics can perform massive interconnections; optical transformation can be operated at high speed, etc. By using the LCTV, the pattern under observation can be correlated with a large number of recallable image memories. In addition, the LCTV technique is rather simple and economical to operate. The LCTV-optical correlator, in principle, can be designed into a compact portable form for insitu application. Brief outlines of the major research findings and publications are provided in this report.

The primary goal of this research is to study a programmable joint- transform correlator (JTC) using liquid crystal television (LCTV) panels for adaptive real-time pattern recognition applications. The technique can improve the pattern recognition and identification technology that is of interest to the U.S. Army. The technique we studied is a real-time programmable electro-optical architecture. There are several reasons for selecting the optical technique over their digital and electronic counterparts, as follows: Optical technique is capable of handling a large space-bandwidth image; optical technique is capable of performing parallel operations; optics can perform massive interconnections; optical transformation can be operated at high speed, etc. By using the LCTV, the pattern under observation can be correlated with a large number of recallable image memories. In addition, the LCTV technique is rather simple and economical to operate. The LCTV-optical correlator, in principle, can be designed into a compact portable form for insitu application. Brief outlines of the major research findings and publications are provided in this report.

The primary goal of this research is to study a programmable joint- transform correlator (JTC) using liquid crystal television (LCTV) panels for adaptive real-time pattern recognition applications. The technique can improve the pattern recognition and identification technology that is of interest to the U.S. Army. The technique we studied is a real-time programmable electro-optical architecture. There are several reasons for selecting the optical technique over their digital and electronic counterparts, as follows: Optical technique is capable of handling a large space-bandwidth image; optical technique is capable of performing parallel operations; optics can perform massive interconnections; optical transformation can be operated at high speed, etc. By using the LCTV, the pattern under observation can be correlated with a large number of recallable image memories. In addition, the LCTV technique is rather simple and economical to operate. The LCTV-optical correlator, in principle, can be designed into a compact portable form for insitu application. Brief outlines of the major research findings and publications are provided in this report.

The primary goal of this research is to study a programmable joint- transform correlator (JTC) using liquid crystal television (LCTV) panels for adaptive real-time pattern recognition applications. The technique can improve the pattern recognition and identification technology that is of interest to the U.S. Army. The technique we studied is a real-time programmable electro-optical architecture. There are several reasons for selecting the optical technique over their digital and electronic counterparts, as follows: Optical technique is capable of handling a large space-bandwidth image; optical technique is capable of performing parallel operations; optics can perform massive interconnections; optical transformation can be operated at high speed, etc. By using the LCTV, the pattern under observation can be correlated with a large number of recallable image memories. In addition, the LCTV technique is rather simple and economical to operate. The LCTV-optical correlator, in principle, can be designed into a compact portable form for insitu application. Brief outlines of the major research findings and publications are provided in this report.

This paper deals with the entity extraction task (named entity recognition) of a text mining process that aims at unveiling non-trivial semantic structures, such as relationships and interaction between entities or communities. In this paper we present a simple and efficient named entity extraction algorithm. The method, named PAMPO (PAttern Matching and POs tagging based algorithm for NER), relies on flexible pattern matching, part-of-speech tagging and lexical-based rules. It was developed to process texts written in Portuguese, however it is potentially applicable to other languages as well. We compare our approach with current alternatives that support Named Entity Recognition (NER) for content written in Portuguese. These are Alchemy, Zemanta and Rembrandt. Evaluation of the efficacy of the entity extraction method on several texts written in Portuguese indicates a considerable improvement on $recall$ and $F_1$ measures.

The application of artificial neural networks (ANNs) for the estimation of HI gas mass fraction (\fgas) is investigated, based on a sample of 13,674 galaxies in the Sloan Digital Sky Survey (SDSS) with HI detections or upper limits from the Arecibo Legacy Fast Arecibo L-band Feed Array (ALFALFA). We show that, for an example set of fixed input parameters ($g-r$ colour and $i$-band surface brightness), a multidimensional quadratic model yields \fgas\ scaling relations with a smaller scatter (0.22 dex) than traditional linear fits (0.32 dex), demonstrating that non-linear methods can lead to an improved performance over traditional approaches. A more extensive ANN analysis is performed using 15 galaxy parameters that capture variation in stellar mass, internal structure, environment and star formation. Of the 15 parameters investigated, we find that $g-r$ colour, followed by stellar mass surface density, bulge fraction and specific star formation rate have the best connection with \fgas. By combining two control parameters, that indicate how well a given galaxy in SDSS is represented by the ALFALFA training set (\pr) and the scatter in the training procedure (\sigf), we develop a strategy for quantifying which SDSS galaxies our ANN can be adequately applied to, and the associated errors in the \fgas\ estimation. In contrast to previous works, our \fgas\ estimation has no systematic trend with galactic parameters such as M$_{\star}$, $g-r$ and SFR. We present a catalog of \fgas\ estimates for more than half a million galaxies in the SDSS, of which $\sim$ 150,000 galaxies have a secure selection parameter with average scatter in the \fgas\ estimation of 0.22 dex.

Coupled oscillator-based networks are an attractive approach for implementing hardware neural networks based on emerging nanotechnologies. However, the readout of the state of a coupled oscillator network is a difficult challenge in hardware implementations, as it necessitates complex signal processing to evaluate the degree of synchronization between oscillators, possibly more complicated than the coupled oscillator network itself. In this work, we focus on a coupled oscillator network particularly adapted to emerging technologies, and evaluate two schemes for reading synchronization patterns that can be readily implemented with basic CMOS circuits. Through simulation of a simple generic coupled oscillator network, we compare the operation of these readout techniques with a previously proposed full statistics evaluation scheme. Our approaches provide results nearly identical to the mathematical method, but also show better resilience to moderate noise, which is a major concern for hardware implementations. These results open the door to widespread realization of hardware coupled oscillator-based neural systems.

Thesis (MS)?Naval Postgraduate School, 1964

Spatial light modulators (SLMs) are being used in correlation-based optical pattern recognition systems to implement the Fourier domain filters. Currently available SLMs have certain limitations with respect to the realizability of these filters. Therefore, it is necessary to incorporate the SLM constraints in the design of the filters. The design of a SLM-constrained minimum average correlation energy (SLM-MACE) filter using the simulated annealing-based optimization technique was investigated. The SLM-MACE filter was synthesized for three different types of constraints. The performance of the filter was evaluated in terms of its recognition (discrimination) capabilities using computer simulations. The correlation plane characteristics of the SLM-MACE filter were found to be reasonably good. The SLM-MACE filter yielded far better results than the analytical MACE filter implemented on practical SLMs using the constrained magnitude technique. Further, the filter performance was evaluated in the presence of noise in the input test images. This work demonstrated the need to include the SLM constraints in the filter design. Finally, a method is suggested to reduce the computation time required for the synthesis of the SLM-MACE filter.

Spatial light modulators (SLMs) are being used in correlation-based optical pattern recognition systems to implement the Fourier domain filters. Currently available SLMs have certain limitations with respect to the realizability of these filters. Therefore, it is necessary to incorporate the SLM constraints in the design of the filters. The design of a SLM-constrained minimum average correlation energy (SLM-MACE) filter using the simulated annealing-based optimization technique was investigated. The SLM-MACE filter was synthesized for three different types of constraints. The performance of the filter was evaluated in terms of its recognition (discrimination) capabilities using computer simulations. The correlation plane characteristics of the SLM-MACE filter were found to be reasonably good. The SLM-MACE filter yielded far better results than the analytical MACE filter implemented on practical SLMs using the constrained magnitude technique. Further, the filter performance was evaluated in the presence of noise in the input test images. This work demonstrated the need to include the SLM constraints in the filter design. Finally, a method is suggested to reduce the computation time required for the synthesis of the SLM-MACE filter.