Results 11  20
of
103
Using Imputation Techniques to Help Learn Accurate Classifiers
"... It is difficult to learn good classifiers when training data is missing attribute values. Conventional techniques for dealing with such omissions, such as mean imputation, generally do not significantly improve the performance of the resulting classifier. We proposed imputationhelped classifiers, w ..."
Abstract

Cited by 5 (3 self)
 Add to MetaCart
(Show Context)
It is difficult to learn good classifiers when training data is missing attribute values. Conventional techniques for dealing with such omissions, such as mean imputation, generally do not significantly improve the performance of the resulting classifier. We proposed imputationhelped classifiers, which use accurate imputation techniques, such as Bayesian multiple imputation (BMI), predictive mean matching (PMM), and Expectation Maximization (EM), as preprocessors for conventional machine learning algorithms. Our empirical results show that EMhelped and BMIhelped classifiers work effectively when the data is “missing completely at random”, generally improving predictive performance over most of the original machine learned classifiers we investigated. 1.
An integrated machine learning approach to stroke prediction
 In KDD
, 2010
"... Stroke is the third leading cause of death and the principal cause of serious longterm disability in the United States. Accurate prediction of stroke is highly valuable for early intervention and treatment. In this study, we compare the Cox proportional hazards model with a machine learning approac ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
(Show Context)
Stroke is the third leading cause of death and the principal cause of serious longterm disability in the United States. Accurate prediction of stroke is highly valuable for early intervention and treatment. In this study, we compare the Cox proportional hazards model with a machine learning approach for stroke prediction on the Cardiovascular Health Study (CHS) dataset. Specifically, we consider the common problems of data imputation, feature selection, and prediction in medical datasets. We propose a novel automatic feature selection algorithm that selects robust features based on our proposed heuristic: conservative mean. Combined with Support Vector Machines (SVMs), our proposed feature selection algorithm achieves a greater area under the ROC curve (AUC) as compared to the Cox proportional hazards
Fuzzy kmeans clustering with missing values
 Proc AMIA Symp
, 2001
"... Fuzzy Kmeans clustering algorithm is a popular approach for exploring the structure of a set of patterns, especially when the clusters are overlapping or fuzzy. However, the fuzzy Kmeans clustering algorithm cannot be applied when the data contain missing values. In many cases, the number of patte ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
(Show Context)
Fuzzy Kmeans clustering algorithm is a popular approach for exploring the structure of a set of patterns, especially when the clusters are overlapping or fuzzy. However, the fuzzy Kmeans clustering algorithm cannot be applied when the data contain missing values. In many cases, the number of patterns with missing values is so large that if these patterns are removed, then the number of patterns to characterize the data set is insufficient. This paper proposes a technique to exploit the information provided by the patterns with the missing values so that the clustering results are enhanced. There are various preprocessing methods to substitute the missing values before clustering the data. However, instead of repairing the data set at the beginning, the repairing can be carried out incrementally in each iteration based on the context. It is thus more likely that less uncertainty is added while incorporating the repair work. Finetuning the missing values using the information from other attributes further consolidates this scheme. Applications of the proposed method in medical domain have produced good performance.
2013: A treeringbased reconstruction of Delaware River basin streamflow using hierarchical Bayesian regression
 J. Climate
"... A hierarchical Bayesian regression model is presented for reconstructing the average summer streamflow at five gauges in the Delaware River basin using eight regional treering chronologies. The model provides estimates of the posterior probability distribution of each reconstructed streamflow serie ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
A hierarchical Bayesian regression model is presented for reconstructing the average summer streamflow at five gauges in the Delaware River basin using eight regional treering chronologies. The model provides estimates of the posterior probability distribution of each reconstructed streamflow series considering parameter uncertainty. The vectors of regression coefficients aremodeled as draws from a commonmultivariate normal distribution with unknown parameters estimated as part of the analysis. This leads to a multilevel structure. The covariance structure of the streamflow residuals across sites is explicitlymodeled. The resulting partial pooling of information across multiple stations leads to a reduction in parameter uncertainty. The effect of no pooling and full pooling of station information, as end points of the method, is explored. The nopoolingmodel considers independent estimation of the regression coefficients for each streamflow gauge with respect to each treering chronology. The fullpooling model considers that the same regression coefficients apply across all streamflow sites for a particular treering chronology. The crosssite correlation of residuals is modeled in all cases. Performance on metrics typically used by treering reconstruction experts, such as reduction of error, coefficient of efficiency, and coverage rates under credible intervals is comparable to, or better, for the partialpooling model relative to the nopooling model, and streamflow estimation uncertainty is reduced. Long record simulations from reconstructions are used to develop estimates of the probability of duration and severity of droughts in the region. Analysis of monotonic trends in the reconstructed drought events do not reject the null hypothesis of no trend at the 90 % significance over 1754–2000.
Sparsity, redundancy and optimal image support towards knowledgebased segmentation
 In CVPR
, 2008
"... In this paper, we propose a novel approach to model shape variations. It encodes sparsity, exploits geometric redundancy, and accounts for the different degrees of local variation and image support. In this context we consider a controlpoint based shape representation. Their sparse distribution is ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
(Show Context)
In this paper, we propose a novel approach to model shape variations. It encodes sparsity, exploits geometric redundancy, and accounts for the different degrees of local variation and image support. In this context we consider a controlpoint based shape representation. Their sparse distribution is derived based on a shape model metric learned from the training data, and the ambiguity of local appearance with regard to segmentation changes. The resulting sparse model of the object improves reconstruction and search behavior, in particular for data that exhibit a heterogeneous distribution of image information and shape complexity. Furthermore, it goes beyond conventional imagebased segmentation approaches since it is able to identify reliable image structures which are then encoded within the model and used to determine the optimal segmentation map. We report promising experimental results comparing our approach with standard models on MRI data of calf muscles an application where traditional imagebased methods fail and CT data of the left heart ventricle. 1.
MultiSource Learning for Joint Analysis of Incomplete MultiModality Neuroimaging Data
"... Incomplete data present serious problems when integrating largescale brain imaging data sets from different imaging modalities. In the Alzheimer’s Disease Neuroimaging Initiative (ADNI), for example, over half of the subjects lack cerebrospinal fluid (CSF) measurements; an independent half of the su ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
(Show Context)
Incomplete data present serious problems when integrating largescale brain imaging data sets from different imaging modalities. In the Alzheimer’s Disease Neuroimaging Initiative (ADNI), for example, over half of the subjects lack cerebrospinal fluid (CSF) measurements; an independent half of the subjects do not have fluorodeoxyglucose positron emission tomography (FDGPET) scans; many lack proteomics measurements. Traditionally, subjects with missing measures are discarded, resulting in a severe loss of available information. We address this problem by proposing two novel learning methods where all the samples (with at least one available data source) can be used. In the first method, we divide our samples according to the availability of data sources, and we learn shared sets of features with stateoftheart sparse learning methods. Our second method learns a base classifier for each data source independently, based on which we represent each source using a single column of prediction scores; we then estimate the missing prediction scores, which, combined with the existing prediction scores, are used to build a multisource fusion model. To illustrate the proposed approaches, we classify patients from the ADNI study into groups with Alzheimer’s disease (AD), mild cognitive impairment (MCI) and normal controls, based on the multimodality data. At baseline, ADNI’s 780 participants (172 AD, 397 MCI, 211 Normal), have at least one of four data types: magnetic resonance imaging (MRI), FDGPET, CSF and proteomics. These data are used to test our algorithms. Comprehensive experiments show that our proposed methods yield stable and promising results.
2005: Coupled patterns of spatiotemporal variability in Northern Hemisphere sea level pressure and conterminous
"... [1] We apply the multitaper frequency domain–singular value decomposition signal detection method to the investigation of coherent patterns of variation in seasonal Northern Hemisphere sea level pressure and conterminous U.S. summer drought during the period 1895–1995. The analysis identifies statis ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
(Show Context)
[1] We apply the multitaper frequency domain–singular value decomposition signal detection method to the investigation of coherent patterns of variation in seasonal Northern Hemisphere sea level pressure and conterminous U.S. summer drought during the period 1895–1995. The analysis identifies statistically significant patterns of spatiotemporal variability on interannual and bidecadal timescales indicative of both coldseason and warmseason atmospheric influences on North American drought patterns. The most robust signal found appears to be associated with the influences of the El Niño–Southern Oscillation (ENSO) on North American summer drought. Evidence is also found to support the existence of a roughly bidecadal drought signal tied to warmseason atmospheric circulation changes. The ‘‘Dust Bowl’ ’ conditions of the 1930s appear to result from a combination of these bidecadal influences on drought conditions that coincide with a decrease in the amplitude of interannual ENSOrelated variability during the 1930s. Citation: Zhang, Z., and M. E. Mann (2005), Coupled patterns of spatiotemporal variability in Northern Hemisphere sea level pressure and conterminous U.S. drought, J. Geophys. Res., 110, D03108, doi:10.1029/2004JD004896. 1.
Discriminants of twentiethcentury changes in Earth surface temperatures
, 2000
"... We present an approach to identifying climate changes that does not hinge on simulations of natural climate variations or anthropogenic changes. We decompose observed interdecadal climate variations into several discriminants, mutually uncorrelated spatiotemporal components with maximal ratio of int ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
We present an approach to identifying climate changes that does not hinge on simulations of natural climate variations or anthropogenic changes. We decompose observed interdecadal climate variations into several discriminants, mutually uncorrelated spatiotemporal components with maximal ratio of interdecadal to intradecadal variance. The dominant discriminants of twentiethcentury variations in surface temperature exhibit largescale warming in which, particularly in the Northern Hemisphere summer months, localized cooling is embedded. The structure of the largescale warming is consistent with expected effects of increases in greenhouse gas concentrations. The localized cooling, with maxima on scales of 1,0002,000 km over East Asia, eastern Europe, and North America, is suggestive of radiative effects of anthropogenic sulphate aerosols.
Gama: Sparse gaussian graphical model estimation via alternating minimization. arXiv preprint arXiv:1405.3034
, 2014
"... Several methods have been recently proposed for estimating sparse Gaussian graphical models using `1 regularization on the inverse covariance matrix. Despite recent advances, contemporary applications require methods that are even faster in order to handle illconditioned high dimensional modern da ..."
Abstract

Cited by 2 (2 self)
 Add to MetaCart
(Show Context)
Several methods have been recently proposed for estimating sparse Gaussian graphical models using `1 regularization on the inverse covariance matrix. Despite recent advances, contemporary applications require methods that are even faster in order to handle illconditioned high dimensional modern day datasets. In this paper, we propose a new method, GAMA, to solve the sparse inverse covariance estimation problem using Alternating Minimization Algorithm (AMA), that effectively works as a proximal gradient algorithm on the dual problem. Our approach has several novel advantages over existing methods. First, we demonstrate that GAMA is faster than the previous best algorithms by many orders of magnitude and is thus an ideal approach for modern high throughput applications. Second, global linear convergence of GAMA is demonstrated rigorously, underscoring its good theoretical properties. Third, the dual algorithm operates on the covariance matrix, and thus easily facilitates incorporating additional constraints on pairwise/marginal relationships between feature pairs based on domain specific knowledge. Over and above estimating a sparse inverse covariance matrix, we also illustrate how to (1) incorporate constraints on the (bivariate) correlations and, (2) incorporate equality (equisparsity) or linear constraints between individual inverse covariance elements. Fourth, we also show that GAMA is better adept at handling extremely illconditioned problems, as is often the case with real data. The methodology is demonstrated on both simulated and real datasets to illustrate its superior performance over recently proposed methods. ar
Comment on "Spatiotemporal filling of missing points in . . .
 XIOGENIC AND NEUROENDOCRINE PROFILES IN ADULTHOOD
, 2007
"... ... method for imputing missing values in incomplete datasets that can exploit both spatial and temporal covariability to estimate missing values from available values. Temporal covariability has not been exploited as widely as spatial covariability in imputing missing values in geophysical datasets ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
... method for imputing missing values in incomplete datasets that can exploit both spatial and temporal covariability to estimate missing values from available values. Temporal covariability has not been exploited as widely as spatial covariability in imputing missing values in geophysical datasets, but, as KG show, doing so can improve estimates of missing values. However, there are several inaccuracies in KG’s paper. Since similar inaccuracies have surfaced in other recent papers, for example, in the literature on paleoclimate reconstructions, I would like to point them out here. (i) In estimating covariance matrices, KG treat an incomplete dataset with imputed values filled in as if it were a complete dataset. Possible variations of the missing values around the imputed values are ignored, leading to biased estimates of covariance matrices (Little and Rubin, 2002). The