Additive Smoothing Research Articles

Dynamic Bayesian network for semantic place classification in mobile robotics

In this paper, the problem of semantic place categorization in mobile robotics is addressed by considering a time-based probabilistic approach called dynamic Bayesian mixture model (DBMM), which is an improved variation of the dynamic Bayesian network. More specifically, multi-class semantic classification is performed by a DBMM composed of a mixture of heterogeneous base classifiers, using geometrical features computed from 2D laserscanner data, where the sensor is mounted on-board a moving robot operating indoors. Besides its capability to combine different probabilistic classifiers, the DBMM approach also incorporates time-based (dynamic) inferences in the form of previous class-conditional probabilities and priors. Extensive experiments were carried out on publicly available benchmark datasets, highlighting the influence of the number of time-slices and the effect of additive smoothing on the classification performance of the proposed approach. Reported results, under different scenarios and conditions, show the effectiveness and competitive performance of the DBMM.

ESTUDO COMPARATIVO DE MODELOS CLÁSSICOS E TÉCNICAS DE COMBINAÇÃO PARA A PREVISÃO DO VOLUME DE CAPTAÇÃO DE ÁGUA PARA USO URBANO – UM ESTUDO DE CASO

O objetivo desse estudo foi investigar o desempenho de modelos classicos etecnicas de combinacao de series temporais para modelar e prever o volume deagua captado por uma empresa de saneamento basico no Sul do Brasil. A seriedisponivel para analise continha uma base de 11 anos (2000 a 2011), com dadosregistrados mensalmente, sendo o ano de 2011 reservado para o processo devalidacao. Para atingir esse objetivo foram transcorridas algumas etapas: (i)caracterizacao da serie temporal, a qual confirmou tendencia e sazonalidade pormeio de testes nao parametricos; (ii) ajuste e comparacao dos modelos desuavizacao exponencial aditivo e multiplicativo de Holt-Winters (HW), com destaquepara o modelo multiplicativo; (iii) ajuste e comparacao dos modelos SARIMA com esem aproximacao, sendo que esse ultimo modelo foi o que se revelou mais acurado;e (iv) aplicacao de seis tipos de combinacoes dos melhores modelos HW eSARIMA, as quais apresentaram resultados mais acurados que os preditoresindividuais. Contudo, vale ressaltar que todos os modelos avaliados apresentaramresultados promissores, pois os MAPEs foram inferiores a 3%.

Robust and Blind 3D Mesh Watermarking in Spatial Domain Based on Faces Categorization and Sorting

In this paper, a 3D watermarking algorithm in spatial domain is presented with blind detection. In the proposed method, a negligible visual distortion is observed in host model. Initially, a preprocessing is applied on the 3D model to make it robust against geometric transformation attacks. Then, a number of triangle faces are determined as mark triangles using a novel systematic approach in which faces are categorized and sorted robustly. In order to enhance the capability of information retrieval by attacks, block watermarks are encoded using Reed-Solomon block error-correcting code before embedding into the mark triangles. Next, the encoded watermarks are embedded in spherical coordinates. The proposed method is robust against additive noise, mesh smoothing and quantization attacks. Also, it is stout next to geometric transformation, vertices and faces reordering attacks. Moreover, the proposed algorithm is designed so that it is robust against the cropping attack. Simulation results confirm that the watermarked models confront very low distortion if the control parameters are selected properly. Comparison with other methods demonstrates that the proposed method has good performance against the mesh smoothing attacks.

Statistical procedures for developing earthquake damage fragility curves

SummaryThis paper describes statistical procedures for developing earthquake damage fragility functions. Although fragility curves abound in earthquake engineering and risk assessment literature, the focus has generally been on the methods for obtaining the damage data (i.e., the analysis of structures), and little emphasis is placed on the process for fitting fragility curves to this data. This paper provides a synthesis of the most commonly used methods for fitting fragility curves and highlights some of their significant limitations. More novel methods are described for parametric fragility curve development (generalized linear models and cumulative link models) and non‐parametric curves (generalized additive model and Gaussian kernel smoothing). An extensive discussion of the advantages and disadvantages of each method is provided, as well as examples using both empirical and analytical data. The paper further proposes methods for treating the uncertainty in intensity measure, an issue common with empirical data. Finally, the paper describes approaches for choosing among various fragility models, based on an evaluation of prediction error for a user‐defined loss function. Copyright © 2015 John Wiley & Sons, Ltd.

A Bayesian Sequential Method for Identifying Structural Shocks and Adaptive Forecasting in State Space Model

The repeated occurrences of interventions in observations make most forecasting models fail to produce appropriate forecasts. The purpose of this study is to propose the adaptive forecasting procedure based on sequential identifications of interventions and adjusting forecast to them in general linear state space model. For the detection of an intervention, the general statistic covering the various types of interventions are derived in the form of posterior odds for non-intervened versus intervened model. And assuming known as the occurrence of the intervention during a given period, the joint posterior distribution of intervening point and type is also derived for identification. The proposed procedure is tested by simulation and empirical study based on the innovational state space model of additive and non-seasonal exponential smoothing.

Bayesian Constituent Context Model for Grammar Induction

Constituent Context Model (CCM) is an effective generative model for grammar induction, the aim of which is to induce hierarchical syntactic structure from natural text. The CCM simply defines the Multinomial distribution over constituents, which leads to a severe data sparse problem because long constituents are unlikely to appear in unseen data sets. This paper proposes a Bayesian method for constituent smoothing by defining two kinds of prior distributions over constituents: the Dirichlet prior and the Pitman-Yor Process prior. The Dirichlet prior functions as an additive smoothing method, and the PYP prior functions as a back-off smoothing method. Furthermore, a modified CCM is proposed to differentiate left constituents and right constituents in binary branching trees. Experiments show that both the proposed Bayesian smoothing method and the modified CCM are effective, and combining them attains or significantly improves the state-of-the-art performance of grammar induction evaluated on standard treebanks of various languages.

Nonparametric Decomposition of Time Series Data with Inputs

The backfitting algorithm commonly used in estimating additive models is used to decompose the component shares explained by a set of predictors on a dependent variable in the presence of linear dependencies (multicollinearity) among the predictors. Simulated and actual data show that the backfitting methods are superior in terms of predictive ability as the degree of multicollinearity worsens. Furthermore, the additive smoothing splines are especially superior when the linear model yield inadequate fit to the data and the predictors exhibit extreme multicollinearity.

Asymmetry effects of shocks in Chinese stock markets volatility: A generalized additive nonparametric approach

The unique characteristics of the Chinese stock markets make it difficult to assume a particular distribution for innovations in returns and the specification form of the volatility process when modelling return volatility with the parametric GARCH family models. This paper therefore applies a generalized additive nonparametric smoothing technique to examine the volatility of the Chinese stock markets. The empirical results indicate that an asymmetric effect of negative news exists in the Chinese stock markets. Furthermore, compared with other parametric models, the generalized additive nonparametric model demonstrates a better performance for return volatility forecasts, particularly for the out-of-sample forecast. The results from this paper have important implications in risk management, portfolio selection, and hedging strategy.

Semiparametric regression models with additive nonparametric components and high dimensional parametric components

This paper concerns semiparametric regression models with additive nonparametric components and high dimensional parametric components under sparsity assumptions. To achieve simultaneous model selection for both nonparametric and parametric parts, we introduce a penalty that combines the adaptive empirical L2-norms of the nonparametric component functions and the SCAD penalty on the coefficients in the parametric part. We use the additive partial smoothing spline estimate as the initial estimate and establish its convergence rate with diverging dimensions of parametric components. Our simulation studies reveal excellent model selection performance of the proposed method. An application to an economic study on Canadian household gasoline consumption reveals interesting results.

A new Bayesian formulation for Holt's exponential smoothing

AbstractIn this paper we propose a Bayesian forecasting approach for Holt's additive exponential smoothing method. Starting from the state space formulation, a formula for the forecast is derived and reduced to a two‐dimensional integration that can be computed numerically in a straightforward way. In contrast to much of the work for exponential smoothing, this method produces the forecast density and, in addition, it considers the initial level and initial trend as part of the parameters to be evaluated. Another contribution of this paper is that we have derived a way to reduce the computation of the maximum likelihood parameter estimation procedure to that of evaluating a two‐dimensional grid, rather than applying a five‐variable optimization procedure. Simulation experiments confirm that both proposed methods give favorable performance compared to other approaches. Copyright © 2008 John Wiley & Sons, Ltd.

Bayesian hierarchical linear mixed models for additive smoothing splines

Bayesian hierarchical models have been used for smoothing splines, thin-plate splines, and L-splines. In analyzing high dimensional data sets, additive models and backfitting methods are often used. A full Bayesian analysis for such models may include a large number of random effects, many of which are not intuitive, so researchers typically use noninformative improper or nearly improper priors. We investigate propriety of the posterior for these cases. Our findings extend known results for normal linear mixed models to certain cases with Bayesian additive smoothing spline models.

An exploratory quantitative risk assessment for high molecular weight sensitizers: wheat flour

Objective: Quantitative risk assessments have been made for wheat dust and allergen exposure and wheat sensitization using classical epidemiological approaches based on simple categorizations in exposure groups. Such analyses suggest the existence of an exposure threshold level for wheat specific sensitization and were used as input in recently conducted risk assessments for wheat flour by the American Conference of Governmental Industrial Hygienists and the Dutch Expert Committee on Occupational Standards. More advanced statistical analyses were applied using generalized additive modeling and smoothed plots to evaluate the shape of the exposure response relationship in greater detail and evaluate the presence of exposure thresholds.Methods: Data were used from a recently conducted epidemiological study in bakery workers. Information was available on wheat sensitization (IgE antibodies), inhalable dust levels and wheat allergen levels. Initial analyses were based on simple exposure categorizations for inhalable dust and allergen exposure. A more detailed analysis using non-parametric generalized additive models (GAM models) and smoothing plots allowed inspection of the presence of an exposure threshold of evaluation of ‘no’ or ‘lowest observed effect levels’ (NOELs, LOELs) using exposure data on the individual level.Results: All analyses showed an increasing sensitization risk with increasing exposure. The classical epidemiological analyses gave evidence for the existence of an exposure threshold or ‘no observed effect level (NOEL)’ for specific wheat sensitization between 0.5 and 1 mg/m3 of inhalable dust. The more advanced analyses did not suggest any evidence for the existence of an exposure threshold. However, estimates of a LOEL obtained by considering an arbitrary increase in sensitization risk between 1.5 and 2 as undesirable, were close to the NOEL from the classical analyses and would therefore not lead to an essentially different exposure limit. The criterion of an increase in wheat sensitization risk was based on the risk in non-wheat dust exposed populations.Conclusion: Exposure response modeling using different classical epidemiological approaches and advanced statistical methods resulted in health based LOEL or NOEL estimates within a relatively close range. But when sensitization accompanied by asthma or rhinitis symptoms was considered as critical endpoint, steeper exposure–response relationships were observed which would lead to lower LOEL values.

An exploratory quantitative risk assessment for high molecular weight sensitizers: wheat flour

Quantitative risk assessments have been made for wheat dust and allergen exposure and wheat sensitization using classical epidemiological approaches based on simple categorizations in exposure groups. Such analyses suggest the existence of an exposure threshold level for wheat specific sensitization and were used as input in recently conducted risk assessments for wheat flour by the American Conference of Governmental Industrial Hygienists and the Dutch Expert Committee on Occupational Standards. More advanced statistical analyses were applied using generalized additive modeling and smoothed plots to evaluate the shape of the exposure response relationship in greater detail and evaluate the presence of exposure thresholds. Data were used from a recently conducted epidemiological study in bakery workers. Information was available on wheat sensitization (IgE antibodies), inhalable dust levels and wheat allergen levels. Initial analyses were based on simple exposure categorizations for inhalable dust and allergen exposure. A more detailed analysis using non-parametric generalized additive models (GAM models) and smoothing plots allowed inspection of the presence of an exposure threshold of evaluation of 'no' or 'lowest observed effect levels' (NOELs, LOELs) using exposure data on the individual level. All analyses showed an increasing sensitization risk with increasing exposure. The classical epidemiological analyses gave evidence for the existence of an exposure threshold or 'no observed effect level (NOEL)' for specific wheat sensitization between 0.5 and 1 mg/m3 of inhalable dust. The more advanced analyses did not suggest any evidence for the existence of an exposure threshold. However, estimates of a LOEL obtained by considering an arbitrary increase in sensitization risk between 1.5 and 2 as undesirable, were close to the NOEL from the classical analyses and would therefore not lead to an essentially different exposure limit. The criterion of an increase in wheat sensitization risk was based on the risk in non-wheat dust exposed populations. Exposure response modeling using different classical epidemiological approaches and advanced statistical methods resulted in health based LOEL or NOEL estimates within a relatively close range. But when sensitization accompanied by asthma or rhinitis symptoms was considered as critical endpoint, steeper exposure-response relationships were observed which would lead to lower LOEL values.

Vibration perception thresholds in workers with long term exposure to lead

OBJECTIVESTo evaluate the impact of long term occupational exposure to lead on function of the peripheral nervous system as reflected by vibration perception threshold (VPT), measured with a portable vibrameter.METHODS217...

Sparse Data and Smoothing in Statistical Part-of-Speech Tagging*

This article reports on a series of experiments in statistical part-of-speech tagging of Swedish texts, with different probabilistic models and different smoothing schemes for both lexical and contextual probabilities. The most important conclusions are that lexical and contextual probabilities require different smoothing methods and that smoothing is only crucial for lexical probabilities. Of the particular smoothing methods tested in the experiments, Good-Turing estimation achieves the best results for the lexical model, while a simple additive smoothing scheme gives the best performance for the contextual model.

Joint analysis of HIV and AIDS surveillance data in back-calculation.

AIDS surveillance data are the main source of information to perform back-calculation of HIV incidence. We propose a method to incorporate additional information gained by linkage with an HIV surveillance system, containing data on the time of first positive HIV test. In this paper we generalize an earlier method that was developed to use HIV testing data available only for AIDS cases. The new method also makes use of cases with an HIV positive test who have not yet developed AIDS, typically a substantial proportion of the HIV-infected population. Furthermore, we use a more realistic model for the HIV testing rate, incorporating dependence on both time since infection and calendar time. The method makes use of an EM algorithm with generalized additive model smoothing, and is applied to data from Veneto, a region of northern Italy. Our results show that HIV incidence in Veneto peaked in the late 1980s, and decreased thereafter. Importantly, the HIV incidence estimates based on joint analysis of HIV and AIDS surveillance data are more efficient than estimates based on AIDS surveillance data alone. Our estimates also show a decreasing trend in the HIV testing rate over time, which leads to the conclusion that the interval between HIV infection and first positive test has lengthened over time. Furthermore, it is found that for infected individuals, the probability of seeking on HIV test is highest soon after infection.

Parameter space of the Holt-winters' model

In the additive Holt-Winters' seasonal exponential smoothing model, it is theoretically possible for smoothing parameters in the usual (0, 1) interval to produce ‘non-invertible’ models. A sample of 406 monthly series reveal that this is a real concern. In the multiplicative model, reasonable estimation procedures produce smoothing constants outside the additive invertible region. When this occurs, the impact on forecasts of values in the distant past is much larger than for recent values and forecasts are poor. Results from the 406 series show that the problem can be avoided and forecasts improved if a subset of the additive invertible region is used as the parameter space.

Forecast Functions Implied by Autoregressive Integrated Moving Average Models and Other Related Forecast Procedures

Summary The simplifying operators in ARIMA (autrogressive integrated moving average) models determine the form of the corresponding forecast functions. For example, regular differences imply polynomial trends and seasonal differences certain periodic functions. The same functions also arise in the context of many other forecast procedures, such as regressions on time, exponential smoothing and Kalman filtering. In this paper we describe how the various methods update the coefficients in these forecast functions and discuss their similarities and differences. In addition, we compare the forecasts from seasonal ARIMA models and the forecasts from Winters' additive and multiplicative smoothing methods.