Sequential nonparametric density estimation

Using kernel estimates of the Parzen type, a naive sequential nonparametric density estimation procedure is developed. The asymptotic distribution structure of the stopping variable is examined. The stopping variable is shown to have finite moments of ail order and is shown to be dosed. The stopping variableNdepends on some preassigned errorvarepsilon, and it is shown thatNdiverges strongly toinftyasvarepsilonconverges to zero. Finally, withhat{f}_n(x)being a kernel-type estimator, it is shown thathat{f}_N(X)converges tof(x), the true density atx, with probability one asvarepsilonconverges to zero.     

A nonparametric nonstationary procedure for failure prediction

The time between failures is a very useful measurement to analyze reliability models for time-dependent systems. In many cases, the failure-generation process is assumed to be stationary, even though the process changes its statistics as time elapses. This paper presents a new estimation procedure for the probabilities of failures; it is based on estimating time-between-failures. The main characteristics of this procedure are that no probability distribution function is assumed for the failure process, and that the failure process is not assumed to be stationary. The model classifies the failures in Q different types, and estimates the probability of each type of failure s-independently from the others. This method does not use histogram techniques to estimate the probabilities of occurrence of each failure-type; rather it estimates the probabilities directly from the values of the time-instants at which the failures occur. The method assumes quasistationarity only in the interval of time between the last 2 occurrences of the same failure-type. An inherent characteristic of this method is that it assigns different sizes for the time-windows used to estimate the probabilities of each failure-type. For the failure-types with low probability, the estimator uses wide windows, while for those with high probability the estimator uses narrow windows. As an example, the model is applied to software reliability data.     

Density-free Bayes risk consistency of nonparametric pattern recognition procedures

Pattern recognition procedures based on the Cesaro mean of orthogonal series are presented and their Bayes risk consistency is established. No restrictions are put on the class conditional densities.     

Sequential CFAR detectors using a dead-zone limiter

The performances of some proposed sequential constant-false-alarm-rate (CFAR) detectors are evaluated. The observations are passed through a dead-zone limiter, the output of which is -1, 0, or +1, depending on whether the input is less than -c, between -c, and c, or greater than c , where c is a constant. The test statistic is the sum of the outputs. The test is performed on a reduced set of data (those with absolute value larger than c), with the test statistic being the sum of the signs of the reduced set of data. Both constant and linear boundaries are considered. Numerical results show a significant reduction of the average number of observations needed to achieve the same false alarm and detection probabilities as compared to a fixed-sample-size CFAR detector using the same kind of test statistic     

基于二维非参数特征分析的SAR图像目标识别

在分析传统Fisher线性鉴别分析局限性的基础上,由图像的行信息和列信息提出了两种形式的二维非参数特征分析(2DNFA)的特征提取方法,并应用于SAR图像目标的识别。直接在SAR图像矩阵上使用非参数特征分析提取特征不仅能充分发挥非参数特征分析的性能而且保留了图像矩阵的结构信息,大大降低了散度矩阵的维数,减小了运算量。使用美国MSTAR计划录取的数据对算法进行了仿真验证,实验结果显示两种形式的二维非参数特征分析在较低特征维数下的识别率均可以达到98%以上,表明所提方法的有效性和正确性。     

Sequential alignment attention model for scene text recognition

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state-of-the-art solutions usually depend on the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. Unfortunately, there often exists severe misalignment between feature areas and text labels in real-world scenarios. To address this problem, this paper proposes a sequential alignment attention model to enhance the alignment between input images and output character sequences. In this model, an attention gated recurrent unit (AGRU) is first devised to distinguish the text and background regions, and further extract the localized features focusing on sequential text regions. Furthermore, CTC guided decoding strategy is integrated into the popular attention-based decoder, which not only helps to boost the convergence of the training but also enhances the well-aligned sequence recognition. Extensive experiments on various benchmarks, including the IIIT5k, SVT, and ICDAR datasets, show that our method substantially outperforms the state-of-the-art methods.     

A robust classification procedure based on mixture classifiers and nonparametric weighted feature extraction

There are many factors to consider in carrying out a hyperspectral data classification. Perhaps chief among them are class training sample size, dimensionality, and distribution separability. The intent of this study is to design a classification procedure that is robust and maximally effective, but which provides the analyst with significant assists, thus simplifying the analyst's task. The result is a quadratic mixture classifier based on Mixed-LOOC2 regularized discriminant analysis and nonparametric weighted feature extraction. This procedure has the advantage of providing improved classification accuracy compared to typical previous methods but requires minimal need to consider the factors mentioned above. Experimental results demonstrating these properties are presented.     

Sequential image deep learning-based Wi-Fi human activity recognition method

For the problems existing in most of the researches,such as weak anti-noise ability,incompatible signal size and insufficient feature extraction of deep-learning-based Wi-Fi human activity recognition,a kind of sequential image deep learning-based recognition method was proposed.Based on the idea of sequential image deep learning,a series of image frames were reconstructed from time-varied Wi-Fi signal to ensure the consistency of input size.In addition,a low-rank decomposition method was innovatively designed to separate low-rank activity information merged in noises.Finally,a deep model combining temporal stream and spatial stream was proposed to automatically capture the spatiotemporal features from length-varied image sequences.The proposed method was extensively tested in WiAR dataset and self collected dataset.The experimental results show the proposed method could achieve the accuracy of 0.94 and 0.96,which indicate its high-accuracy performance and robustness in pervasive environments.     

On adaptive nonparametric detection systems using dependent samples

A procedure is obtained for modifying given sampled-data parametric detectors to make them asymptotically nonparametric. Unlike standard nonparametric devices, these detectors do not require the assumption of independent samples but only a knowledge of the input spectral shapes. As examples of this technique, two types of conventional array detectors are modified to produce nonparametric systems.     

Automatic CRP mapping using nonparametric machine learning approaches

This paper studies an uneven two-class unsupervised classification problem of satellite imagery, i.e., the mapping of U.S. Department of Agriculture's (USDA) Conservation Reserve Program (CRP) tracts. CRP is a nationwide program that encourages farmers to plant long-term, resource conserving covers to improve soil, water, and wildlife resources. With recent payments of nearly US $1.6 billion for new enrollments (2002 signup), it is imperative to obtain accurate digital CRP maps for management and evaluation purposes. CRP mapping is a complex classification problem where both CRP and non-CRP areas are composed of various cover types. Two nonparametric machine learning approaches, i.e., decision tree classifier (DTC) and support vector machine (SVMs) are implemented in this work. Specifically, considering the importance of CRP classification sensitivity, a new DTC pruning method is proposed to increase recall. We also study two SVM relaxation approaches to increase recall. Moreover, a localized and parallel framework is suggested in order to efficiently deal with the large-scale CRP mapping need. Simulation results validate the applicability of the suggested framework and proposed techniques.     

Sequential generators using read-only memories

A sequential generator for the chainig for logic or arithmetic functions is described which uses low-cost and fast-access-time read-only memories. The generator proposed enables memory capacity to be considerably reduced by use of a multiplexer and an adder, and is claimed to give a considerable time saving over classical methods.     

Sequential transmission using noiseless feedback

The presence of a feedback channel makes possible a variety of sequential transmission procedures, each of which can be classified as either a block-transmission or a continuous-transmission scheme according to the way in which information is encoded for transmission over a noisy forward channel. A sequential continuous-transmission system employing a binary symmetric forward channel (but which is suitable for use with any discrete memoryless forward channel) and a noiseless feedback channel is described. Its error exponent is shown to be substantially greater than the optimum block-code error exponent at all transmission rates less than channel capacity. The average value and the first-order probability distribution of the effective constraint length, found by simulating the system on an IBM 709 computer, are also given.     

Signal and image processing using three-dimensional binary ranking transforms

Because binary mathematical morphology permits fast local neighborhood operations by flash conversion, it is used extensively in high-speed pattern-recognition computer systems. Further, since anyN-dimensional integer function may be represented by an (N + 1)-dimensional binary (bilevel) function, ordinary two-dimensional graylevel images become three-dimensional binary images. Thus these images may be processed by high-speed flash-conversion computers assuming that a sufficiently compact three-dimensional kernel can be devised. The tetradekahedron of the face-centered-cubic tessellation forms a perfect kernel in three-dimensions. Its neighborhood is compact. It has total symmetry with all 12 neighbors equidistant from the central element. Using this kernel a variety of useful three-dimensional morphological operations may be performed for target track detection, shaded graphics, data clustering, automated focusing, and spatial filtering.This research was supported by the National Cancer Institute (Grant CA45047), the National Science Foundation (Grant DCR8611863), the Office of Naval Research (Contract Number N001488K-0435-N143), and the Department of Defense (delivery order 00055, San Diego State University Foundation, under Contract 85-D0203 from the Naval Ocean Systems Center).     

FIR system identification based on a nonparametric Bayesian model using the Indian buffet process

In this paper, we propose a nonparametric Bayesian model combined with the Indian buffet process (IBP) for a finite impulse response (FIR) system. We develop an FIR system identification method that can simultaneously estimate the number of FIR taps and coefficients. In the proposed model, each FIR tap consists of a coefficient and a gain, and the gain is a binary value. An infinite-dimensional binary vector is composed of binary values, and we assume that this binary vector is generated by the IBP. To identify the FIR system, we specify the likelihood function and prior distributions of the parameters and derive their posterior distributions. We can simultaneously estimate the number of FIR taps and coefficients by sampling from posterior distributions using the Gibbs sampler. Our simulations demonstrate that although the number of FIR taps is unknown, the identification performance of the proposed method in a high signal-to-noise ratio environment is similar to or better than that of the conventional least square solution.     

Foveal automatic target recognition using a multiresolution neuralnetwork

This paper presents a method for detecting and classifying a target from its foveal (graded resolution) imagery using a multiresolution neural network. Target identification decisions are based on minimizing an energy function. This energy function is evaluated by comparing a candidate blob with a library of target models at several levels of resolution simultaneously available in the current foveal image. For this purpose, a concurrent (top-down-and-bottom-up) matching procedure is implemented via a novel multilayer Hopfield (1985) neural network. The associated energy function supports not only interactions between cells at the same resolution level, but also between sets of nodes at distinct resolution levels. This permits features at different resolution levels to corroborate or refute one another contributing to an efficient evaluation of potential matches. Gaze control, refoveation to more salient regions of the available image space, is implemented as a search for high resolution features which will disambiguate the candidate blob. Tests using real two-dimensional (2-D) objects and their simulated foveal imagery are provided.     

An improved sliding-load calibration procedure using asemiparametric circle-fitting procedure

Circle-fitting problems often occur in microwave engineering when dealing with variable delays, e.g., during calibration using a sliding load. This paper proposes an efficient semiparametric circle-fitting procedure, which takes into account the phase relationships over the frequencies. It produces more accurate results than the standard sliding-load calibration, requires only three positions on the sliding load for the whole frequency band, and is more robust to the settings of the positions of the sliding load. The proposed method also has the ability to detect whether or not the sliding load is defective or out of its specifications. This can be done by using only three positions on the sliding load. Optimal-position settings are then proposed. The performance of the proposed method is illustrated on sliding-load measurements up to 50 GHz, demonstrating the ability of detecting modeling errors and showing that the accuracy of the proposed method using three positions is comparable to the standard method with six positions     

Generalized Gaussian mixture models as a nonparametric Bayesian approach for clustering using class-specific visual features

Recently, there has been a growing interest in the problem of learning mixture models from data. The reasons and motivations behind this interest are clear, since finite mixture models offer a formal approach to the important problems of clustering and data modeling. In this paper, we address the problem of modeling non-Gaussian data which are largely present, and occur naturally, in several computer vision and image processing applications via the learning of a generative infinite generalized Gaussian mixture model. The proposed model, which can be viewed as a Dirichlet process mixture of generalized Gaussian distributions, takes into account the feature selection problem, also, by determining a set of relevant features for each data cluster which provides better interpretability and generalization capabilities. We propose then an efficient algorithm to learn this infinite model parameters by estimating its posterior distributions using Markov Chain Monte Carlo (MCMC) simulations. We show how the model can be used, while comparing it with other models popular in the literature, in several challenging applications involving photographic and painting images categorization, image and video segmentation, and infrared facial expression recognition.     

Automatic benthic imagery recognition using a hierarchical two-stage approach

The main objective of this work is to establish an automated classification system of seabed images. A novel two-stage approach to solving the image region classification task is presented. The first stage is based on information characterizing geometry, colour and texture of the region being analysed. Random forests and support vector machines are considered as classifiers in this work. In the second stage, additional information characterizing image regions surrounding the region being analysed is used. The reliability of decisions made in the first stage regarding the surrounding regions is taken into account when constructing a feature vector for the second stage. The proposed technique was tested in an image region recognition task including five benthic classes: red algae, sponge, sand, lithothamnium and kelp. The task was solved with the average accuracy of 90.11% using a data set consisting of 4589 image regions and the tenfold cross-validation to assess the performance. The two-stage approach allowed increasing the classification accuracy for all the five classes, more than 27% for the “difficult” to recognize “kelp” class.     

PN code acquisition using signed-rank-based nonparametric detectorsin DS/SS systems

In this paper, signed-rank based nonparametric detectors are used for pseudonoise (PN) code acquisition in direct-sequence spread-spectrum (DS/SS) systems. We first derive the locally optimum rank (LOR) detector and then propose the locally suboptimum rank (LSR) and modified signed-rank (MSR) detectors using approximate score functions. We compare the single-dwell scheme without the verification mode using the proposed LSR and MSR detectors with that using the conventional squared-sum (SS) and modified sign (MS) detectors. From the simulation results, it is shown that the proposed LSR and MSR detectors perform better than the MS detector by about 2-3 dB and are nearly optimum