Nearest Neighbor Method

A family of Kalman-type filters that estimate the user's position indoors, using range measurements and floor plan data, is presented. The floor plan information is formulated as a set of linear constraints and is used to truncate the Gaussian posterior probability densities occurring in the Kalmantype filters. The truncated Gaussian is approximated with a new Gaussian density such that the mean and the covariance matrix coincide approximately with the mean and the covariance matrix of the truncated Gaussian. The linear constraints may be used all at once or imposed one after another. Both strategies are studied in this paper and their advantages and disadvantages are discussed. Simulation data has been processed using the Extended Kalman filter, the second order Extended Kalman filter and the Unscented Kalman filter, and the results are compared to a reference solution computed with a sequential Monte Carlo particle filter. The simulations indicate that the proposed method for using floor plan information increases the accuracy of the estimates without significantly increasing the computation or the memory requirements. The low computational demand and the absence of the training phase make it a useful addition to existing indoor positioning systems.

This article considers the problem of educational placement. Several discriminant techniques are applied to a data set from a survey project of science ability. A profile vector for each student consists of five science-educational indictors. The students are intended to be placed into three reference groups: advanced, regular, and remedial. Various discriminant techniques, including Fisher’s discriminant analysis and kernel-based nonparametric discriminant analysis, are compared. The evaluation work is based on the leaving-one-out misclassification score. Results from the five school data sets and 500 bootstrap samples reveal that the kernel-based nonparametric approach with bandwidth selected by cross validation performs reasonably well. The authors regard kernel-based nonparametric procedures as desirable competitors to Fisher’s discriminant rule for handling problems of educational placement.

Particle filtering is being investigated extensively due to its important feature of target tracking based on nonlinear and non-Gaussian model. It tracks a trajectory with a known model at a given time. It means that particle filter tracks an arbitrary trajectory only if the time instant when trajectory switches from one model to another model is known apriori. Because of this reason particle filter is not able to track any arbitrary trajectory where transition from one model to another model is not known. For real world application, trajectory is always random in nature and may follow more than one model. In this paper we propose a novel method, which overcomes the above problem. In the proposed method an interacting multiple model based approach is used along with particle filtering, which automates the model selection process for tracking an arbitrary trajectory. We have utilized nearest neighbor (NN) method for data association, which is fast and easy to implement. In the proposed approach, there is no need to have apriori information about the exact model that a target may follow.

This paper develops a ZigBee indoor positioning scheme based on the location fingerprinting approach. The proposed scheme includes four workflows: (1) creating the location fingerprint table, (2) training the locating model using neural network (NN), (3) preprocessing data through the Signal-Index-Pair method, and (4) estimating the coordinate of the mobile target instantly. Testing results show that within the error distance of 5 meters, the NN locating model with the Signal-Index-Pair data preprocess method can increase the positioning precision by 17% compared with the original NN, in terms of the cumulative error probability (CEP). It also achieves 5% CEP higher than the k (k=5) nearest neighbor method and the weighted k (k=5) nearest neighbor method. Potential applications include patient tracking in hospitals, object tracking for factory monitoring, self-navigation of autonomous robots, and visitors monitoring in military buildings, and so on.

This paper compares Models-3/Community Multiscale Air Quality (CMAQ) outputs at multiple resolutions by interpolating from coarse resolution to fine resolution and analyzing the interpolation difference. Spatial variograms provide a convenient way to investigate the spatial character of interpolation differences and, importantly, to distinguish between naive (nearest neighbor) interpolation and bilinear interpolation, which takes a weighted average of four neighboring cells. For example, when the higher resolution is three times the lower, the variogram of the difference between naive interpolation of the lower resolution output and the higher resolution output shows a depression at every third lag. This phenomenon is related to the blocky nature of naive interpolation and demonstrates the inferiority of naive interpolation to bilinear interpolation in a way that pixelwise comparisons cannot. Theoretical investigations show when one can expect to observe this periodic depression in the variogram of interpolation differences. Naive interpolation is in fact used widely in a number of settings; our results suggest that it should be routinely replaced by bilinear interpolation.

We present a new method for creating a comparable document collection from two document collections in different languages. The best query keys were extracted from a Finnish source collection (articles of the newspaper Aamulehti) with the relative average term frequency (RATF) formula. The keys were translated into English with a dictionary-based query translation program. The resulting lists of words were used as queries that were run against the target collection (Los Angeles Times articles) with the nearest neighbor method. The documents were aligned with unrestricted and daterestricted alignment schemes, which were also combined. The combined scheme was found the best, when the relatedness of the document pairs was assessed with a fivedegree relevance scale. Of the 400 document pairs, roughly 40% were highly or fairly related and 75% included at least lexical similarity.

KNN (K-nearest neighbor) is an important tool in machine learning and it is used in classification and prediction problems. In recent years several modified versions of KNN search algorithm have been developed and employed to improve the efficiency of search. KNN has enormous real life applications and is widely used in data mining. Data structures like KD tree (or K dimensional tree) are used for implementing KNN effectively. A KD tree is a multidimensional binary search tree that can be balanced or unbalanced. With the increase in dimension of space the computational time of KNN-KD search goes high. Certain modifications that can help in improvising the search time has been developed in recent years.

Forest structure is a fundamental component of the forest ecosystem and significantly impacts carbon sequestration. Previous studies mainly focused on optimizing forest non-spatial attributes for restoring carbon, but the significance of stand spatial structure characteristics to carbon sequestration remains unclear. To understand the relationship between stand spatial structure (SSS) [i.e., Mingling index (M), Dominance index (D), Uniform angle index (W), Crowdedness index (C), and Hegyi competition index (Hy)] on carbon sequestration (Cs), this study was carried out based on four 50m×50 m plots of natural Larix gmelinii plots at Cuigang Forest Farm, located in the Daxing'an Mountains of Heilongjiang province, northeast China. The results revealed that the SSS of the forest exhibited minor instabilities, characterized by constant species mixing and moderate unpredictability in species dominance over 10 years (2012-2022). At the plot level, carbon stock increased significantly with an annual increment carbon (Aic) (i.e., 0.9-1.38 t ha⁻¹ per year) and relative increment carbon (Ric) significantly fluctuated (i.e., 4.48-5.84 %). A significant positive correlation was observed between SSS and Cs, particularly with M, D, and Hy, while a negative correlation was found with W and C. Hence, our findings demonstrate that SSS is crucial in regulating carbon sequestration, with dense and competitive structure contributing to maximized carbon sequestration capacity and mitigating the impact of climate change, offering critical insights for sustainable forest management.

The computation of Global Climate Models (GCMs) presents significant numerical challenges. This paper presents new algorithms based on sparse occupancy trees for learning and emulating the long wave radiation parameterization in the in the NCAR CAM climate model. This emulation occupies by far the most significant portion of the computational time in the implementation of the model. From the mathematical point of view this parameterization can be considered as a mapping R 220 → R 33 which is to be learned from scattered data samples (x i , y i), i = 1,. .. , N. Hence, the problem represents a typical application of high-dimensional statistical learning. The goal is to develop learning schemes that are not only accurate and reliable but also computationally efficient and capable of adapting to time-varying environmental states. The algorithms developed in this paper are compared with other approaches such as neural networks, nearest neighbor methods, and regression trees as to how these various goals are met.

Background: There are too many design options for software effort estimators. How can we best explore them all? Aim: We seek aspects on general principles of effort estimation that can guide the design of effort estimators. Method: We identified the essential assumption of analogy-based effort estimation: i.e. the immediate neighbors of a project offer stable conclusions about that project. We test that assumption by generating a binary tree of clusters of effort data and comparing the variance of super-trees vs smaller sub-trees. Results: For ten data sets (from Coc81, Nasa93, Desharnais, Albrecht, ISBSG, and data from Turkish companies), we found: (a) the estimation variance of cluster sub-trees is usually larger than that of cluster super-trees; (b) if analogy is restricted to the cluster trees with lower variance then effort estimates have a significantly lower error (measured using MRE and a Wilcoxon test, 95% confidence, compared to nearest-neighbor methods that use neighborhoods of a fixed size). Conclusion: Estimation by analogy can be significantly improved by a dynamic selection of nearest neighbors, using only the project data from regions with small variance.

Changes in the configurational entropies of molecules make important contributions to free energies of reaction for processes such as protein-folding, noncovalent association, and conformational change. However, obtaining entropy from molecular simulations represents a long-standing computational challenge. Here, two recently introduced approaches, the nearest-neighbor (NN) method and the mutual-information expansion (MIE), are combined to furnish an efficient and accurate method of extracting the configurational entropy from a molecular simulation to a given order of correlations among the internal degrees of freedom. The resulting method takes advantage of the strengths of each approach. The NN method is entirely nonparametric (i.e., it makes no assumptions about the underlying probability distribution), its estimates are asymptotically unbiased and consistent, and it makes optimum use of a limited number of available data samples. The MIE, a systematic expansion of entropy in mutual information terms of increasing order, provides a wellcharacterized approximation for lowering the dimensionality of the numerical problem of calculating the entropy of a high-dimensional system. The combination of these two methods enables obtaining well-converged estimations of the configurational entropy that capture many-body correlations of higher order than is possible with the simple histogramming that was used in the MIE method originally. The combined method is tested here on two simple systems: an idealized system represented by an analytical distribution of 6 circular variables, where the full joint entropy and all the MIE terms are exactly known; and the R,S stereoisomer of tartaric acid, a molecule with 7 internalrotation degrees of freedom for which the full entropy of internal rotation has been already estimated by the NN method. For these two systems, all the expansion terms of the full MIE of the entropy are estimated by the NN method and, for comparison, the MIE approximations up to 3rd order are also estimated by simple histogramming. The results indicate that the truncation of the MIE at the 2-body level can be an accurate, computationally non-demanding approximation to the configurational entropy of anharmonic internal degrees of freedom. If needed, higher-order correlations can be

Changes in the configurational entropies of molecules make important contributions to the free energies of reaction for processes such as protein‐folding, noncovalent association, and conformational change. However, obtaining entropy from molecular simulations represents a long‐standing computational challenge. Here, two recently introduced approaches, the nearest‐neighbor (NN) method and the mutual‐information expansion (MIE), are combined to furnish an efficient and accurate method of extracting the configurational entropy from a molecular simulation to a given order of correlations among the internal degrees of freedom. The resulting method takes advantage of the strengths of each approach. The NN method is entirely nonparametric (i.e., it makes no assumptions about the underlying probability distribution), its estimates are asymptotically unbiased and consistent, and it makes optimum use of a limited number of available data samples. The MIE, a systematic expansion of entropy in...

Several effective methods have been developed recently for improving predictive performance by generating and combining multiple learned models. The general approach is to create a set of learned models either by applying an algorithm repeatedly to different versions of the training data, or by applying different learning algorithms to the same data. The predictions of the models are then combined

The assessment of an appropriate function describing the relationship between hydrological variables is a frequent problem. The usual way of estimating an overall function is a difficult task if the relationship between the variables is highly nonlinear. Nearest neighbor methods provide an alternative. In this paper different generalizations of the nearest neighbor method are suggested. Besides using different estimators the problem of finding an appropriate distance measure is discussed. Linear embedding of the observation space into an appropriate Euclidean space can improve the distance-based methods substantially. The suggested transformations and the corresponding local linear estimators are applied to two different examples: the prediction of local discharge characteristics from catchment properties and a simple 1 day flood forecasting problem.

Motivation: Membrane transport proteins play a crucial role in the import and export of ions, small molecules or macromolecules across biological membranes. Currently, there are a limited number of published computational tools which enable the systematic discovery and categorization of transporters prior to costly experimental validation. To approach this problem, we utilized a nearest neighbor method which seamlessly integrates homologous search and topological analysis into a machine-learning framework. Results: Our approach satisfactorily distinguished 484 transporter families in the Transporter Classification Database, a curated and representative database for transporters. A five-fold cross-validation on the database achieved a positive classification rate of 72.3% on average. Furthermore, this method successfully detected transporters in seven model and four non-model organisms, ranging from archaean to mammalian species. A preliminary literature-based validation has cross-va...

In oats, the year factor has a large influence on the phenotype expression. As a consequence, one year estimates of genetic distance between cultivars often have very little precision. The objectives of this work were to estimate: the coefficients of repeatability and the number of years needed for an accurate prediction of real genetic value of individuals in contrasting experiments regarding the presence or absence of fungicide application. Also, the genetic distance repeatability as well as the number of years needed for an accurate measure of genetic distance based on morphological characters was assessed. In four crop seasons, nine oat cultivars were evaluated with and without fungicide application in a random block design, with six replications. Repeatability coefficients and Mahalanobis’ distance were determined. Afterwards, the distances based on four years of evaluation in the conditions with/without fungicide application were assessed as well as their correlations. The res...

Dezoito genótipos de aveia foram testados quanto à dissimilaridade genética, com e sem o controle de moléstias da parte aérea. As variáveis avaliadas foram rendimento de grãos desaristados, peso de mil grãos, peso do hectolitro, estatura de planta e dias da emergência à floração. Foram empregados análises por variáveis canônicas e técnicas de agrupamento por meio dos métodos de otimização de Tocher e o método hierárquico do vizinho mais próximo, tendo como medida de dissimilaridade a distância generalizada de Mahalanobis. Pelos resultados, constatou-se significativa dissimilaridade genética entre os genótipos, indicando a existência de variabilidade para os caracteres avaliados. O método de Tocher e o método do vizinho mais próximo permitiram a separação dos genótipos em grupos distintos, possibilitando a identificação de futuros genitores que possam ser utilizados em cruzamentos artificiais que produzam progênies com maior heterose. Os caracteres que mais contribuíram para a dissim...

Frequent failures are becoming a serious concern to the community of high-end computing, especially when the applications and the underlying systems rapidly grow in size and complexity. In order to develop effective fault-tolerant strategies, there is a critical need to predict failure events. To this end, we have collected detailed event logs from IBM BlueGene/L, which has 128K processors, and is currently the fastest supercomputer in the world. In this study, we first show how the event records can be converted into a data set that is appropriate for running classification techniques. Then we apply classifiers on the data, including RIPPER (a rule-based classifier), Support Vector Machines (SVMs), a traditional Nearest Neighbor method, and a customized Nearest Neighbor method. We show that the customized nearest neighbor approach can outperform RIP-PER and SVMs in terms of both coverage and precision. The results suggest that the customized nearest neighbor approach can be used to alleviate the impact of failures.

Two new feature selection methods are introduced, the first based on separability criterion, the second on consistency index that includes interactions between the selected subsets of features. Comparison of accuracy was made against information-theory based selection methods on several datasets training neurofuzzy and nearest neighbor methods on various subsets of selected features. Methods based on separability seem to be most promising.

The experimental data retrieved from three-dimensional particle tracking velocimetry (3D PTV) are crucial for indoor environment engineering when designing ventilation strategies or monitoring airborne pollutants dispersion in inhabited spaces as well as in livestock compounds. The performance of the measurement algorithm, as the core part of the 3D PTV technology, has a strong influence on the final results. This study presents a method to select the better 3D PTV algorithm depending on the application targeted. This study first compared the performance of seven 3D PTV algorithms on two types of 3D laminar macro scale flows of known Navier–Stokes solutions, namely the Kovasznay and Beltrami flows. The comparison was based on the accuracy, the measuring volume coverage and the achievable trajectory length of each algorithm with respect to the particle tracking density and the total number of frames. Then, the measurement algorithms with better performances were further compared with...

In this study, the nonparametric k-nearest neighbour method was used to describe diameter distributions of birch stands in Northwest Spain. It was applied using the following essential steps: (i) estimation of the distance between target stand and reference stands (kth-nearest neighbours); (ii) use of weight function defined by estimated distance, for each reference stand (to enable construction of a linear combination that defines the diameter distribution of target stand); and (iii) identification of the optimum number of neighbours that minimizes the error associated with statistics used to evaluate the results (bias, mean absolute error, mean square error and relative error index). The best results were obtained by measuring the distance between distributions with a function defined by arithmetic mean diameter and quadratic mean diameter, with a weight function defined by inverse of the distance between distributions and with ideal number of neighbours equal to 8.

Two new feature selection methods are introduced, the first based on separability criterion, the second on consistency index that includes interactions between the selected subsets of features. Comparison of accuracy was made against information-theory based selection methods on several datasets training neurofuzzy and nearest neighbor methods on various subsets of selected features. Methods based on separability seem to be most promising.

In the present research we study the codebook generation problem of vector quantization, using two different techniques of Genetic Algorithm (GA). We used the Simple GA (SGA) method and Ordain GA (OGA) method in vector quantization. SGA with roulette and tournament selection with elitist approach is used in the experiments. The OGA is based on the pair wise nearest neighbor method (PNN). Both these approaches were fine tuned by the inclusion of GLA. The two methods are used and compared with respect to quality of compressed image, rate of distortion and time cost. In our experiments we used images of Lena and Bridge. While using OGA we got better values of PSNR (34.5), (32.1) with less distorted image as compared to the SGA with (29.7), (29.2) PSNR values for both Lena and Bridge respectively. Although in OGA the time performance is inferior, it is thrice more time-consuming.

People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal. If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the "Taverne" license above, please follow below link for the End User Agreement:

Many indices for evaluation of features have been considered. Applied to single features they allow for filtering irrelevant attributes. Algorithms for selection of subsets of features also remove redundant features. Hashing techniques enable efficient application of feature relevance indices to selection of feature subsets. A number of such methods have been applied to artificial and real-world data. Strong influence of continuous feature discretization and very good performance of separability-based discretization has been noted.

The k-nearest-neighbor (kNN) decision rule is a simple and robust classifier for text categorization. The performance of kNN decision rule depends heavily upon the value of the neighborhood parameter k. The method categorize a test document even if the difference between the number of members of two competing categories is one. Hence, choice of k is crucial as different values of k can change the result of text categorization. Moreover, text categorization is a challenging task as the text data are generally sparse and high dimensional. Note that, assigning a document to a predefined category for an arbitrary value of k may not be accurate when there is no bound on the margin of majority voting. A method is thus proposed in spirit of the nearest-neighbor decision rule using a medoid-based weighting scheme to deal with these issues. The method puts more weightage on the training documents that are not only lie close to the test document but also lie close to the medoid of its corresponding category in decision making, unlike the standard nearest-neighbor algorithms that stress on the documents that are just close to the test document. The aim of the proposed classifier is to enrich the quality of decision making. The empirical results show that the proposed method performs better than different standard nearest-neighbor decision rules and support vector machine classifier using various well-known text collections in terms of macro-and micro-averaged f-measure.

The similarity based decision rule computes the similarity between a new test document and the existing documents of the training set that belong to various categories. The new document is grouped to a particular category in which it has maximum number of similar documents. A document similarity based supervised decision rule for text categorization is proposed in this article. The similarity measure determine the similarity between two documents by finding their distances with all the documents of training set and it can explicitly identify two dissimilar documents. The decision rule assigns a test document to the best one among the competing categories, if the best category beats the next competing category by a previously fixed margin. Thus the proposed rule enhances the certainty of the decision. The salient feature of the decision rule is that, it never assigns a document arbitrarily to a category when the decision is not so certain. The performance of the proposed decision rule for text categorization is compared with some well known classification techniques e.g., k-nearest neighbor decision rule, support vector machine, naive bayes etc. using various TREC and Reuter corpora. The empirical results have shown that the proposed method performs significantly better than the other classifiers for text categorization.

This work proposes and evaluates a Nearest-Neighbor Method to substitute missing values in datasets formed by continuous attributes. In the substitution process, each instance containing missing values is compared with complete instances, and the closest instance is used to assign the attribute missing value. We evaluate this method in simulations performed in four datasets that are usually employed as benchmarks for data mining methods-Iris Plants, Wisconsin Breast Cancer, Pima Indians Diabetes and Wine Recognition. First, we consider the substitution process as a prediction task. In this sense, we employ two metrics (Euclidean and Manhattan) to simulate substitutions both in original and normalized datasets. The obtained results were compared to those provided by a usually employed method to perform this task, i.e. substitution by the mean value. Based on these simulations, we propose a substitution procedure for the well-known K-Means Clustering Algorithm. Then, we perform clustering simulations, comparing the results obtained in the original datasets with the substituted ones. These results indicate that the proposed method is a suitable estimator for substituting missing values, i.e. it preserves the relationships between variables in the clustering process. Therefore, the proposed Nearest-Neighbor Method is an appropriate data preparation tool for the K-Means Clustering Algorithm.

Combining outputs from different classifiers to achieve high accuracy in classification task is one of the most active research areas in ensemble method. Although many state-of-art approaches have been introduced, no method is outstanding compared with the others on numerous data sources. With the aim of introducing an effective classification model, we propose a Gaussian Mixture Model (GMM) based method that combines outputs of base classifiers (called meta-data or Level1 data) resulted from Stacking Algorithm. We further apply Genetic Algorithm (GA) to that data as feature selection strategy to explore an optimal subset of Level1 in which our GMM-based approach can achieve high accuracy. Two methods are combined in a single framework called GAGMM. Experiments implemented on 21 UCI Machine Learning Repository data files and CLEF2009 medical image database demonstrate the advantage of our framework compared with other well-known combining algorithms such as Decision Template, Multiple Response Linear Regression (MLR), SCANN and fixed combining rules as well as GMM-based approaches on original data.

In this paper a specialized method for generating Markovian random fields, with or without conditioning, is presented. Here, the prior fields are assumed to be stationary second-order Gauss-Markov random fields in N-dimensional (N-D) euclidian space. The unconditional Markov fields are generated as numerical solutions of white-noise driven PDEs, for the solution of which finite difference or finite volume discretization on a regular grid yields an algorithm commonly known as the Nearest Neighbor Method (NNM). In addition, starting with the continuous space PDE-driven field, we develop a generalization of the NNM algorithm for the generation and optimal interpolation (estimation) of conditional random fields in the Bayesian sense. This generalized NNM algorithm is called Bayesian-NNM, or B-NNM, and is essentially similar to the classical NNM-type algorithms. The B-NNM method can be used for conditional simulation of random fields, as well as the optimal estimation of spatial fields assuming Markovian prior statistics, based on a collection of measurements or data. These data may be defined either at points or, more importantly, on continuous subdomains of various sizes and shapes, such as lineaments, faults, layers, boreholes, etc. In this paper, we also present exact closed form relations for linear estimation of 1-D Markovian fields, and this without recourse to space discretization. The complete analytical results are used for testing the multi-dimensional B-NNM code in the 1-D case.

Over the course of the previous two decades, there has been a rise in the quantity of text documents stored digitally. The ability to organize and categorize those documents in an automated mechanism, is known as text categorization which is used to classify them into a set of predefined categories so they may be preserved and sorted more efficiently. Identifying appropriate structures, architectures, and methods for text classification presents a challenge for researchers. This is due to the significant impact this concept has on content management, contextual search, opinion mining, product review analysis, spam filtering, and text sentiment mining. This study analyzes the generic categorization strategy and examines supervised machine learning approaches and their ability to comprehend complex models and nonlinear data interactions. Among these methods are k-nearest neighbors (KNN), support vector machine (SVM), and ensemble learning algorithms employing various evaluation techniques. Thereafter, an evaluation is conducted on the constraints of every technique and how they can be applied to real-life situations.

We describe a new decision list induction algorithm called the Greedy Prepend Algorithm (GPA). GPA improves on other decision list algorithms by introducing a new objective function for rule selection and a set of novel search algorithms that allow application to large scale real world problems. GPA achieves state-of-the-art classification accuracy on the protein secondary structure prediction problem in bioinformatics and the English part of speech tagging problem in computational linguistics. For both domains GPA produces a rule set that human experts find easy to interpret, a marked advantage in decision support environments. In addition, we compare GPA to other decision list induction algorithms as well as support vector machines, C4.5, naive Bayes, and a nearest neighbor method on a number of standard data sets from the UCI machine learning repository.

Airborne LiDAR techniques can provide accurate measurements of tree height, from which estimates of stem volume and forest woody biomass can be obtained. These techniques, however, are still expensive to apply repeatedly over large areas. The current paper presents a methodology which first transforms mean stand heights obtained from LiDAR over small strips into relevant stem volume estimates. These are then extended over an entire forest by applying two estimation methods (k-NN and locally calibrated regression) to Landsat ETM+ images. The methodology is tested over a coastal area covered by pine forest in the Regional Park of San Rossore (Central Italy). The results are evaluated by comparison with the ground stem volumes of a recent forest inventory, taking into consideration the effect of stand size. In general, the accuracies of two estimation methods are dependent on the size of the forest stands and are satisfactory only when considering stands larger than 5-10 ha. The outputs of the parametric regression procedure are slightly more stable than those of k-NN and more faithfully reproduce the spatial patterns of the ground data.

Background: There are too many design options for software effort estimators. How can we best explore them all? Aim: We seek aspects on general principles of effort estimation that can guide the design of effort estimators. Method: We identified the essential assumption of analogy-based effort estimation: i.e. the immediate neighbors of a project offer stable conclusions about that project. We test that assumption by generating a binary tree of clusters of effort data and comparing the variance of super-trees vs smaller sub-trees. Results: For ten data sets (from Coc81, Nasa93, Desharnais, Albrecht, ISBSG, and data from Turkish companies), we found: (a) the estimation variance of cluster sub-trees is usually larger than that of cluster super-trees; (b) if analogy is restricted to the cluster trees with lower variance then effort estimates have a significantly lower error (measured using MRE and a Wilcoxon test, 95% confidence, compared to nearest-neighbor methods that use neighborhoods of a fixed size). Conclusion: Estimation by analogy can be significantly improved by a dynamic selection of nearest neighbors, using only the project data from regions with small variance.

Up-to-date information of forest resources is required at a variety of scales in order to support forest management practices ranging from strategic to operational levels. The rate of change in Scottish forests is significant and may occur over short periods due to active silvicultural management and the impact of natural disturbances (e.g. windthrow). Field-based sample inventories collect information on the location, condition and extent of current forest resources. However, Bottai et al. (2002) note that in many cases the sampling intensities for operational inventories are too coarse and cannot provide accurate representations of the forest resources.

PTTRNFNDR is an unsupervised statistical learning algorithm that detects patterns in DNA sequences, protein sequences, or any natural language texts that can be decomposed into letters of a finite alphabet. PTTRNFNDR performs complex mathematical computations, and its processing time increases when input texts become large. To achieve better speed performance, several strategies were applied in the implementation of the program, including parallel operations of binary search trees. A standard binary search tree is not thread-safe due to its dynamic insertions and deletions. Here, we adjusted the standard binary search tree for parallelized operations to achieve improved performance of the PTTRNFNDR algorithm. The method can be applied to other software platforms to quicken data searching through parallel operations of binary search trees when several conditions are met.

RESUMO-Mais de 200 linhagens avançadas de aveia (Avena retive L.) e trigo (Triticuin aeflivuzn L.), selecionadas em 1983, foram avaliadas em dois experimentos, conduzidos em Guaíba, RS, durante o ano de 1984. O objetivo foi testar rendimento de grãos, peso do hectolitro e de 100 grãos, rendimento biol6gico, ciclo de desenvolvimento e estatura de planta. Os resultados foram significativos para todas as variáveis, indicando que a seleção tem Lido eficiente na aveia;por outro lado, em trigo, somente uma linhagem mostrou superioridade em rendimento de grãos sobre a melhor cultivar-padrão. Termos para indexação: desempenho genotipico. PRELIMINAR PERFORMANCE OF NEW OATANDWHEATGENOT'YPE$ IN CENTRAL DEPRESSION OF RS ABSTRACT. More than two hundred oat and wheat advanced-Iines selected in 1983 were grown in fieid experiments at Guaiba, RS, Brazil, in 1984. The objective was to quantify, grain yieId, 100 grain weight, hectoliter weight, biological yield, heading time, and plant stature. The results were significant for ali traits, indicating that selection was efficient in oats; on the other hand, only one wheat-lirre showed supetiority to ihe best check'variety. Index terms: genotype performance.

Forest policy makers increasingly desire the use of quantitative descriptions to define desirable forest characteristics as a target for forest management. A framework for quantitative, multivariate target definition and assessment is described. The framework uses the joint distribution of multiple forest structure attributes to describe a set of desired forest structures and to identify a target region. The target region contains the most likely attribute values and its extent is controlled by choosing a probability of acceptance or acceptance level. Nonparametric procedures implementing the target definition and assessment framework have been developed and are described. The implemented procedures were used with a real data set representing 129 riparian stands in western Washington State, U.S.A. to define a threedimensional target for riparian forest management in the region using stand density, quadratic mean diameter, and average tree height. A bootstrap simulation and a 50-50 split representative sample were used to evaluate the consistency of the implemented procedures by testing the null hypothesis that attribute value distributions for a target data set and an observation data set, both randomly drawn from a common distribution, were statistically indistinguishable. Chi-squared goodness of fit tests with a = 0.05 were used to compare observed mean acceptance percentages from the bootstrap simulation and observed acceptance percentages from the 50-50 split representative sample to the targeted acceptance levels of 95%, 90%, 80%, and 50%. Evaluation results indicated that the target definition and assessment procedures were consistent by failing to reject the null hypothesis for each evaluation method, with p-values of p = 0.963 for the bootstrap simulation and p = 0.866 for the 50-50 split representative sample.

When employing nearest neighbor classifiers scaling of input variables is often useful. In this paper we propose a small modification in usual data preprocessing: scaling of variables should be done by use of pooled variances instead of overall ones. Thus prediction accuracy is distinctly improved in some situations.

Day by day, human-agent negotiation becomes more and more vital to reach a socially beneficial agreement when stakeholders need to make a joint decision together. Developing agents who understand not only human preferences but also attitudes is a significant prerequisite for this kind of interaction. Studies on opponent modeling are predominantly based on automated negotiation and may yield good predictions after exchanging hundreds of offers. However, this is not the case in human-agent negotiation in which the total number of rounds does not usually exceed tens. For this reason, an opponent model technique is needed to extract the maximum information gained with limited interaction. This study presents a conflict-based opponent modeling technique and compares its prediction performance with the well-known approaches in human-agent and automated negotiation experimental settings. According to the results of human-agent studies, the proposed model outpr erforms them despite the dive...