Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 8
Encontrada(s) 1 página(s)

e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

Approximate Bayesian computation by modelling summary statistics in a quasi-likelihood framework

  • Cabras, Stefano
  • Castellanos Nueda, María Eugenia
  • Ruli, Erlis
Approximate Bayesian Computation (ABC) is a useful class of methods for Bayesian inference when the likelihood function is computationally intractable. In practice, the basic ABC algorithm may be inefficient in the presence of discrepancy between prior and posterior. Therefore, more elaborate methods, such as ABC with the Markov chain Monte Carlo algorithm (ABC-MCMC), should be used. However, the elaboration of a proposal density for MCMC is a sensitive issue and very difficult in the ABC setting, where the likelihood is intractable. We discuss an automatic proposal distribution useful for ABC-MCMC algorithms. This proposal is inspired by the theory of quasi-likelihood (QL) functions and is obtained by modelling the distribution of the summary statistics as a function of the parameters. Essentially, given a real-valued vector of summary statistics, we reparametrize the model by means of a regression function of the statistics on parameters, obtained by sampling from the original model in a pilot-run simulation study. The QL theory is well established for a scalar parameter, and it is shown that when the conditional variance of the summary statistic is assumed constant, the QL has a closed-form normal density. This idea of constructing proposal distributions is extended to non constant variance and to real-valued parameter vectors. The method is illustrated by several examples and by an application to a real problem in population genetics., Maria Eugenia Castellanos was partially funded by Ministerio de Ciencia e Innovación grant MTM2013-42323. Stefano Cabras has been partially funded by Ministerio de Ciencia e Innovación grant MTM2013-42323, ECO2012-38442, RYC-2012-11455 and together with Erlis Ruli were partially funded by Ministero dell’Istruzione, dell’Univesità e della Ricerca of Italy. All authors have been partially financially supported by Regione Autonoma della Sardegna under grant CRP-59903.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

A new minimal training sample scheme for intrinsic Bayes factors in censored data

  • Cabras, Stefano
  • Castellanos, María Eugenia
  • Perra, Silvia
The problem of covariate selection for regression models with right censored data is considered. It is approached from a default Bayesian point of view with Bayes factors (BFs) and in particular with Intrinsic BF (IBF) that depends on the minimal training samples (MTSs). In the presence of censored data, the number of possible MTSs increases, due to the fact that uncensored data, relevant for training the improper prior into a proper posterior, must be combined with censored data. For this purpose, the sequential minimal training sample scheme (SMTS) accounts for such requirements but generally leads to IBF correction factors that do not have an analytical form and thus require numerical approximation. In order to obtain an analytical expression of the correction terms, a different TS scheme is introduced based on the Kaplan-Meier (KM) estimator, termed the KM minimal training sample scheme. This new tool works extremely well in the analyzed simulation setting and also in the applications; it produces results which are similar, if not better, than the IBF calculated using MTSs. The resulting new IBF, being based on analytical expressions, is much faster to compute. Evidence of these results comes from a large simulation study, theoretical arguments, and an application to a real data set. (C) 2014 Elsevier B.V. All rights reserved., Maria Eugenia Castellanos was partially supported by Ministerio de Ciencia e Innovación grant MTM2010-19528, MTM2013-42323 and the visiting professor program of the Regione Autonoma della Sardegna. Stefano Cabras has been partially supported by Ministerio de Ciencia e Innovación grant MTM2013-42323, ECO2012-38442, RYC-2012-11455, and Silvia Perra was partially supported by Ministero dell’Istruzione, dell’Univesità e della Ricerca of Italy grant CRP-59903.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

A Bayesian non-parametric modeling to estimate student response to ICT investment

  • Cabras, Stefano
  • Tena Horrillo, Juan de Dios
This paper estimates the causal impact of investment in information and communication technologies (ICT) on student performances in mathematics as measured in the Program for International Student Assessment (PISA) 2012 for Spain. To do this we apply a new methodology in this context known as Bayesian Additive Regression Trees that has important advantages over more standard parametric specifications. Results indicate that ICT has a moderate positive effect on math scores. In addition, we analyze how this effect interacts with variables related to school features and student socioeconomic status, finding that ICT investment is especially beneficial for students from a low socioeconomic background., Stefano Cabras has been supported by Ministerio de Ciencia e Innovación grant MTM2013-42323, ECO2012-38442, RYC-2012-11455, by Ministero dell'Istruzione, dell'Univesità e della Ricerca of Italy and by Regi one Autonoma della Sardegn a CRP-59903. Juan de Dios Tena Horrillo has been supported by Ministerio de Educación y Ciencia, ECO2009-08100 and ECO2012-32401.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

A Markov chain representation of the multiple testing problem

  • Cabras, Stefano
The problem of multiple hypothesis testing can be represented as a Markov process where a new alternative hypothesis is accepted in accordance with its relative evidence to the currently accepted one. This virtual and not formally observed process provides the most probable set of non null hypotheses given the data; it plays the same role as Markov Chain Monte Carlo in approximating a posterior distribution. To apply this representation and obtain the posterior probabilities over all alternative hypotheses, it is enough to have, for each test, barely defined Bayes Factors, e.g. Bayes Factors obtained up to an unknown constant. Such Bayes Factors may either arise from using default and improper priors or from calibrating p-values with respect to their corresponding Bayes Factor lower bound. Both sources of evidence are used to form a Markov transition kernel on the space of hypotheses. The approach leads to easy interpretable results and involves very simple formulas suitable to analyze large datasets as those arising from gene expression data (microarray or RNA-seq experiments)., The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author has been funded by Ministerio de Ciencia e Innovación grant MTM2013-42323, ECO2012-38442, RYC-2012-11455, and ACOMP/2015/202, by Ministero dell'Istruzione, dell'Univesità e della Ricerca of Italy, and Regione Autonoma della Sardegna CRP-59903.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

A random forest application to contact-state classification for robot programming by human demonstration

  • Cabras, Stefano
  • Castellanos, M. E.
  • Staffetti, E.
This paper addresses the non-parametric estimation of the stochastic process related to the classification problem that arises in robot programming by demonstration of compliant motion tasks. Robot programming by demonstration is a robot programming paradigm in which a human operator demonstrates the task to be performed by the robot. In such demonstration, several observable variables, such as velocities and forces can be modeled, non-parametrically, in order to classify the current state of a contact between an object manipulated by the robot and the environment in which it operates. Essential actions in compliant motion tasks are the contacts that take place, and therefore, it is important to understand the sequence of contact states made during a demonstration, called contact classification. We propose a contact classification algorithm based on the random forest algorithm. The main advantage of this approach is that it does not depend on the geometric model of the objects involved in the demonstration. Moreover, it does not rely on the kinestatic model of the contact interactions. The comparison with state-of-the-art contact classifiers shows that random forest classifier is more accurate., Stefano Cabras and María Eugenia Castellanos have been partially supported by Ministerio de Ciencia e Innovación grants MTM2013-42323, ECO2012-38442, RYC-2012-11455, by Ministero dell'Istruzione, dell'Univesità e della Ricerca of Italy and by Regione Autonoma della SardegnaCRP-59903. Ernesto Staffetti have been partially supported by the project TRA2013-47619-C2-2-R of theSpanish Ministerio de Economía y Competitividad (2014-2016).
Repositorio Digital de la UPF
Publication Artículo científico (JournalArticle)

Geostatistical modelling of chemical residues on archaeological floors in the presence of barriers

  • Lancelotti, Carla
  • Muñoz, Facundo
  • Negre Pérez, Joan
Maps representing the distribution of chemical residues over anthropogenic floors are the main diagnostic tools used by archaeologists for addressing the identification of geochemical signatures of past actions. Geostatistics allows producing these maps from a sample of locations by modelling the spatial autocorrelation structure of these kind of phenomena. However, the homogeneity of the prediction regions is a strong assumption in the model. The presence of barriers, such as the inner walls of domestic units, introduces discontinuities in prediction areas. In this paper, we investigate how to incorporate information of a geographical nature into the process of geostatistical prediction. We propose the use of cost-based distances to quantify the correlation between locations, a solution which has proved to be a practical alternative approach for archaeological intrasite analysis. The cost-based approach produces more reliable results avoiding the unrealistic assumption of the homogeneity of the study area. As a working example, a case study of the distribution of two specific chemical signatures in domestic floors is presented within a controlled ethnographical context in Northern Gujarat (India). On a broad disciplinary scale, the benefits of using our approach include improved estimates in regions with complex geometry and lower uncertainty in the kriging predictions., This work was supported by SimulPast Consolider Ingenio project (CSD2010-00034) and Experimentation and development of advanced artificial intelligence techniques for the computer simulation of social dynamics and historical evolution project (HAR2009-12258), funded by the Spanish Ministry of Science and Innovation. Joan Negre Pérez has worked on this research on a postdoctoral fellowship from the Argentinian Ministry of Science, Technology and Productive Innovation (PICT2012-2148) and partially funded by research grant 2008UAB503 from the Autonomous University of Barcelona (Spain). Facundo Muñoz has worked on this research on a postdoctoral fellowship from the French National Institute for Agricultural Research and partially funded by research grant MTM2013-42323-P from the Spanish Ministry of Economy and Competitiveness and ACOMP/2015/202 from Generalitat Valenciana (Spain). Carla Lancelotti is part of the CaSEs Research Group, a recognised excellence group of the Generalitat de Catalunya (SGR2014-1417). Her work in this paper is partially conducted within the framework of the MoMarq Project (Modelado y Simulación de Marcadores de Actividades Antrópicas: de lo etnográfico a lo arqueológico) funded by the Spanish Ministry of Economy and Competitiveness (HAR2014-55518-P).
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Risk analysis for unintentional slide deployment during airline operations

  • Ayra, Eduardo S.
  • Ríos Insua, David
  • Castellanos, Maria Eugenia
  • Larbi, Lydia
We present a risk analysis undertaken to mitigate problems in relation to the unintended deployment of slides under normal operations within a commercial airline. This type of incident entails relevant costs for the airline industry. After assessing the likelihood and severity of its consequences, we conclude that such risks need to be managed. We then evaluate the effectiveness of various countermeasures, describing and justifying the chosen ones. We also discuss several issues faced when implementing and communicating the proposed measures, thus fully illustrating the risk analysis process., The work of DRI is supported by the AXAICMAT Chair in Adversarial Risk Analysis, MINECO projects MTM2011-28983-C3-1 and MTM2014-56949-C3-1-R, and the ISCH COST Action IS1304 on Expert Judgement. The work of MEC is supported by MINECO project MTM2013-42323-P., Peer reviewed, Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Artículo científico (JournalArticle)

Geostatistical modelling of chemical residues on archaeological floors in the presence of barriers

  • Lancelotti, Carla
  • Muñoz, Facundo
  • Negre Pérez, Joan
Maps representing the distribution of chemical residues over anthropogenic floors are the main diagnostic tools used by archaeologists for addressing the identification of geochemical signatures of past actions. Geostatistics allows producing these maps from a sample of locations by modelling the spatial autocorrelation structure of these kind of phenomena. However, the homogeneity of the prediction regions is a strong assumption in the model. The presence of barriers, such as the inner walls of domestic units, introduces discontinuities in prediction areas. In this paper, we investigate how to incorporate information of a geographical nature into the process of geostatistical prediction. We propose the use of cost-based distances to quantify the correlation between locations, a solution which has proved to be a practical alternative approach for archaeological intrasite analysis. The cost-based approach produces more reliable results avoiding the unrealistic assumption of the homogeneity of the study area. As a working example, a case study of the distribution of two specific chemical signatures in domestic floors is presented within a controlled ethnographical context in Northern Gujarat (India). On a broad disciplinary scale, the benefits of using our approach include improved estimates in regions with complex geometry and lower uncertainty in the kriging predictions., This work was supported by SimulPast Consolider Ingenio project (CSD2010-00034) and Experimentation and development of advanced artificial intelligence techniques for the computer simulation of social dynamics and historical evolution project (HAR2009-12258), funded by the Spanish Ministry of Science and Innovation. Joan Negre Pérez has worked on this research on a postdoctoral fellowship from the Argentinian Ministry of Science, Technology and Productive Innovation (PICT2012-2148) and partially funded by research grant 2008UAB503 from the Autonomous University of Barcelona (Spain). Facundo Muñoz has worked on this research on a postdoctoral fellowship from the French National Institute for Agricultural Research and partially funded by research grant MTM2013-42323-P from the Spanish Ministry of Economy and Competitiveness and ACOMP/2015/202 from Generalitat Valenciana (Spain). Carla Lancelotti is part of the CaSEs Research Group, a recognised excellence group of the Generalitat de Catalunya (SGR2014-1417). Her work in this paper is partially conducted within the framework of the MoMarq Project (Modelado y Simulación de Marcadores de Actividades Antrópicas: de lo etnográfico a lo arqueológico) funded by the Spanish Ministry of Economy and Competitiveness (HAR2014-55518-P).