The Experts below are selected from a list of 213 Experts worldwide ranked by ideXlab platform

Lawrence Carin - One of the best experts on this subject based on the ideXlab platform.

  • On Classification with Incomplete Data
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007
    Co-Authors: David P. Williams, Xuejun Liao, Lawrence Carin, Balaji Krishnapuram
    Abstract:

    We address the incomplete-data problem in which feature vectors to be classified are missing data (features). A (supervised) logistic regression algorithm for the classification of incomplete data is developed. Single or multiple imputation for the missing data is avoided by performing analytic integration with an estimated Conditional Density Function (conditioned on the observed data). Conditional Density Functions are estimated using a Gaussian mixture model (GMM), with parameter estimation performed using both expectation-maximization (EM) and variational Bayesian EM (VB-EM). The proposed supervised algorithm is then extended to the semisupervised case by incorporating graph-based regularization. The semisupervised algorithm utilizes all available data-both incomplete and complete, as well as labeled and unlabeled. Experimental results of the proposed classification algorithms are shown

  • Classification of unexploded ordnance using incomplete multisensor multiresolution data
    IEEE Transactions on Geoscience and Remote Sensing, 2007
    Co-Authors: David Williams, Chunping Wang, Xuejun Liao, Lawrence Carin
    Abstract:

    We address the problem of unexploded ordnance (UXO) detection in which data to be classified are available from multiple sensor modalities and multiple resolutions. Specifically, features are extracted from measured magnetometer and electro- magnetic induction data; multiple-resolution data are manifested when the sensors are separated from the buried targets of interest by different distances (e.g., different sensor-platform heights). The proposed classification algorithm explicitly emphasizes features extracted from fine-resolution imagery over those extracted from less reliable coarse-resolution data. When fine-resolution features are unavailable (due to undeployed sensors), the algorithm an- alytically integrates out the missing features via an estimated Conditional Density Function, which is conditioned on the observed features (from deployed sensors). This Density Function exploits the statistical relationships that exist among features at different resolutions, as well as those among features fromdifferent sensors (in the multisensor case). Experimental classification results are shown for real UXO data, on which the proposed algorithm con- sistently achieves better classification performance than common alternative approaches.

  • ICML - Incomplete-data classification using logistic regression
    Proceedings of the 22nd international conference on Machine learning - ICML '05, 2005
    Co-Authors: David P. Williams, Xuejun Liao, Lawrence Carin
    Abstract:

    A logistic regression classification algorithm is developed for problems in which the feature vectors may be missing data (features). Single or multiple imputation for the missing data is avoided by performing analytic integration with an estimated Conditional Density Function (conditioned on the non-missing data). Conditional Density Functions are estimated using a Gaussian mixture model (GMM), with parameter estimation performed using both expectation maximization (EM) and Variational Bayesian EM (VB-EM). Using widely available real data, we demonstrate the general advantage of the VB-EM GMM estimation for handling incomplete data, vis-a-vis the EM algorithm. Moreover, it is demonstrated that the approach proposed here is generally superior to standard imputation procedures.

Laurent Donzé - One of the best experts on this subject based on the ideXlab platform.

  • Impact of the Density Support on the Matching Bias: A Matched-Pair Analysis Based on Business Survey Data
    2020
    Co-Authors: Laurent Donzé
    Abstract:

    Matching methods have been extensively used to evaluate economic policy. However, they are not without fault and, indeed, a selection bias may appear. In 1998, Heckman and al. (1998) have precisely characterize this bias by a decomposition in three parts which can be non parametrically estimated. These estimations depend among others on the support of the Conditional Density Function of the covariates X used for the matching. In a study of the impact of the policy of supporting the adoption of advanced manufacturing technologies by Swiss firms, we have measured the different components of the bias. We put into evidence the extreme sensibility of the results to the measure of support Density.

  • Impact of the Density Support on the Matching Bias: A Matched-Pair Analysis Based on Business Survey Data; KOF Working Papers
    2020
    Co-Authors: Laurent Donzé
    Abstract:

    Matching methods have been extensively used to evaluate economic policy. However, they are not without fault and, indeed, a selection bias may appear. In 1998, Heckman and al. (1998) have precisely characterize this bias by a decomposition in three parts which can be non parametrically estimated. These estimations depend among others on the support of the Conditional Density Function of the covariates X used for the matching. In a study of the impact of the policy of supporting the adoption of advanced manufacturing technologies by Swiss firms, we have measured the different components of the bias. We put into evidence the extreme sensibility of the results to the measure of support Density.

Salah Khardani - One of the best experts on this subject based on the ideXlab platform.

  • Strong consistency of local linear estimation of a Conditional Density Function under random censorship
    Arabian Journal of Mathematics, 2020
    Co-Authors: Abdelkader Benkhaled, Fethi Madani, Salah Khardani
    Abstract:

    In this paper, we study nonparametric local linear estimation of the Conditional Density of a randomly censored scalar response variable given a Functional random covariate. We establish under general conditions the pointwise almost sure convergence with rates of this estimator under $$\alpha $$ α -mixing dependence. Finally, to show interests of our results, on the practical point of view, we have conducted a computational study, first on a simulated data and, then on some real data concerning Kidney transplant data.

  • On the Central Limit Theorem for a Conditional Mode Estimator of a Randomly Censored Time Series
    Journal of statistical theory and practice, 2014
    Co-Authors: Salah Khardani, Mohamed Lemdani, E. Ould Saïd
    Abstract:

    In this article, we consider a kernel estimator of the Conditional Density Function from which we derive an estimator of the Conditional mode. We address the case of a randomly right-censored model when the data exhibit some kind of dependency. The Conditional mode estimator is defined as the random variable that maximizes the Conditional Density estimator. Under classical conditions we establish a central-limit theorem for this estimator. We carry out a simulation study to illustrate our results.

Xuejun Liao - One of the best experts on this subject based on the ideXlab platform.

  • On Classification with Incomplete Data
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007
    Co-Authors: David P. Williams, Xuejun Liao, Lawrence Carin, Balaji Krishnapuram
    Abstract:

    We address the incomplete-data problem in which feature vectors to be classified are missing data (features). A (supervised) logistic regression algorithm for the classification of incomplete data is developed. Single or multiple imputation for the missing data is avoided by performing analytic integration with an estimated Conditional Density Function (conditioned on the observed data). Conditional Density Functions are estimated using a Gaussian mixture model (GMM), with parameter estimation performed using both expectation-maximization (EM) and variational Bayesian EM (VB-EM). The proposed supervised algorithm is then extended to the semisupervised case by incorporating graph-based regularization. The semisupervised algorithm utilizes all available data-both incomplete and complete, as well as labeled and unlabeled. Experimental results of the proposed classification algorithms are shown

  • Classification of unexploded ordnance using incomplete multisensor multiresolution data
    IEEE Transactions on Geoscience and Remote Sensing, 2007
    Co-Authors: David Williams, Chunping Wang, Xuejun Liao, Lawrence Carin
    Abstract:

    We address the problem of unexploded ordnance (UXO) detection in which data to be classified are available from multiple sensor modalities and multiple resolutions. Specifically, features are extracted from measured magnetometer and electro- magnetic induction data; multiple-resolution data are manifested when the sensors are separated from the buried targets of interest by different distances (e.g., different sensor-platform heights). The proposed classification algorithm explicitly emphasizes features extracted from fine-resolution imagery over those extracted from less reliable coarse-resolution data. When fine-resolution features are unavailable (due to undeployed sensors), the algorithm an- alytically integrates out the missing features via an estimated Conditional Density Function, which is conditioned on the observed features (from deployed sensors). This Density Function exploits the statistical relationships that exist among features at different resolutions, as well as those among features fromdifferent sensors (in the multisensor case). Experimental classification results are shown for real UXO data, on which the proposed algorithm con- sistently achieves better classification performance than common alternative approaches.

  • ICML - Incomplete-data classification using logistic regression
    Proceedings of the 22nd international conference on Machine learning - ICML '05, 2005
    Co-Authors: David P. Williams, Xuejun Liao, Lawrence Carin
    Abstract:

    A logistic regression classification algorithm is developed for problems in which the feature vectors may be missing data (features). Single or multiple imputation for the missing data is avoided by performing analytic integration with an estimated Conditional Density Function (conditioned on the non-missing data). Conditional Density Functions are estimated using a Gaussian mixture model (GMM), with parameter estimation performed using both expectation maximization (EM) and Variational Bayesian EM (VB-EM). Using widely available real data, we demonstrate the general advantage of the VB-EM GMM estimation for handling incomplete data, vis-a-vis the EM algorithm. Moreover, it is demonstrated that the approach proposed here is generally superior to standard imputation procedures.

Han-ying Liang - One of the best experts on this subject based on the ideXlab platform.