Iterative Multiview Subspace Learning for Unpaired Multiview Clustering.

In real applications, several unpredictable or uncertain factors could result in unpaired multiview data, i.e., the observed samples between views cannot be matched. Since joint clustering among views is more effective than individual clustering in each view, we investigate unpaired multiview clustering (UMC), which is a valuable but insufficiently studied problem. Due to lack of matched samples between views, we could fail to build the connection between views. Therefore, we aim to learn the latent subspace shared by views. However, existing multiview subspace learning methods usually rely on the matched samples between views. To address this issue, we propose an iterative multiview subspace learning strategy iterative unpaired multiview clustering (IUMC), aiming to learn a complete and consistent subspace representation among views for UMC. Moreover, based on IUMC, we design two effective UMC methods: 1) Iterative unpaired multiview clustering via covariance matrix alignment (IUMC-CA) that further aligns the covariance matrix of subspace representations and then performs clustering on the subspace and 2) iterative unpaired multiview clustering via one-stage clustering assignments (IUMC-CY) that performs one-stage multiview clustering (MVC) by replacing the subspace representations with clustering assignments. Extensive experiments show the excellent performance of our methods for UMC, compared with the state-of-the-art methods. Also, the clustering performance of observed samples in each view can be considerably improved by those observed samples from the other views. In addition, our methods have good applicability in incomplete MVC.

Learning Latent Low-Rank and Sparse Embedding for Robust Image Feature Extraction.

To defy the curse of dimensionality, the inputs are always projected from the original high-dimensional space into the target low-dimension space for feature extraction. However, due to the existence of noise and outliers, the feature extraction task for corrupted data is still a challenging problem. Recently, a robust method called low rank embedding (LRE) was proposed. Despite the success of LRE in experimental studies, it also has many disadvantages: 1) The learned projection cannot quantitatively interpret the importance of features. 2) LRE does not perform data reconstruction so that the features may not be capable of holding the main energy of the original "clean" data. 3) LRE explicitly transforms error into the target space. 4) LRE is an unsupervised method, which is only suitable for unsupervised scenarios. To address these problems, in this paper, we propose a novel method to exploit the latent discriminative features. In particular, we first utilize an orthogonal matrix to hold the main energy of the original data. Next, we introduce an l2,1 -norm term to encourage the features to be more compact, discriminative and interpretable. Then, we enforce a columnwise l2,1 -norm constraint on an error component to resist noise. Finally, we integrate a classification loss term into the objective function to fit supervised scenarios. Our method performs better than several state-of-the-art methods in terms of effectiveness and robustness, as demonstrated on six publicly available datasets.

Multi-output parameter-insensitive kernel twin SVR model.

Multi-output regression aims at mapping a multivariate input feature space to a multivariate output space. Currently, it is effective to extend the traditional support vector regression (SVR) mechanism to solve the multi-output case. However, some methods adopting a combination of single-output SVR models exhibit the severe drawback of not considering the possible correlations between outputs, and other multi-output SVRs show high computational complexity and are typically sensitive to parameters due to the influence of noise. To handle these problems, in this study, we determine the multi-output regression function through a pair of nonparallel up- and down-bound functions solved by two smaller-sized quadratic programming problems, which results in a fast learning speed. This method is named multi-output twin support vector regression (M-TSVR). Moreover, when the noise is heteroscedastic, based on our M-TSVR, we introduce a pair of multi-input/output nonparallel parameter insensitive up- and down-bound functions to evaluate a regression model named multi-output parameter-insensitive twin support vector regression (M-PITSVR). To handle the nonlinear case, we derive the kernelized extensions of M-TSVR and M-PITSVR. Finally, a series of comparative experiments with several other multi-output-based methods are performed on twelve multi-output datasets. The experimental results indicate that the proposed multi-output regressors yield fast learning speed as well as a better and more stable prediction performance.

Relationship between adiposity parameters and cognition: the "fat and jolly" hypothesis in middle-aged and elderly people in China.

The association between adiposity parameters and cognition is complex. The purpose of this study was to assess the relationship between adiposity parameters and cognition in middle-aged and elderly people in China.Data were obtained from a cross-sectional study. Cognitive function was evaluated in 5 domains, and adiposity parameters were measured. The association between adiposity parameters and cognition was analyzed using multiple linear and binary logistic regression analyses.After controlling for confounders, men with overweight and obesity had better scores in TICS-10 ([1] total, overweight vs normal: P = .006, ß = 0.04; obesity vs normal: P = .005, ß = 0.04. [2] stratification by age, with age ≥ 59 years, overweight vs normal: P = .006, ß = 0.05; obesity vs normal: P = .014, ß = 0.05. [3] stratification by educational levels, with less than elementary education, overweight vs normal: P = .011, ß = 0.05; obesity vs normal: P = .005, ß = 0.05), immediate word recall ([1] total, overweight vs normal: P = .015, ß = 0.04. [2] stratification by age, with age 45-58 years, overweight vs normal: P = .036, ß = 0.05. [3] stratification by educational levels, with less than elementary education, overweight vs normal: P = .044, ß = 0.04; above high school, overweight vs normal: P = .041, ß = 0.09), self-rated memory ([1] stratification by age, with age ≥ 59 years, overweight vs normal: P = .022, ß = 0.05. [2] stratification by educational levels, with less than elementary education, overweight vs normal: P = .023, ß = 0.04), and drawing a picture ([1] total, overweight vs normal: OR = 1.269, 95% CI = 1.05-1.53. [2] stratification by educational levels, with less than elementary education, overweight vs normal: OR = 1.312, 95% CI = 1.06-1.63); obesity vs normal: OR = 1.601, 95% CI = 1.11-2.31 than the normal weight; women with overweight and obesity had better measure scores in the TICS-10 ([1] total, overweight vs normal: P < .0001, ß = 0.06; obesity vs normal: P < .0001, ß = 0.05. [2] stratification by age, with age 45-58 years, obesity vs normal: P = .007, ß = 0.05; with age ≥ 59 years: overweight vs normal: P < .0001, ß = 0.07, obesity vs normal: P = .002, ß = 0.06. [3] stratification by educational levels, with illiterate, overweight vs normal: P = .001, ß = 0.08; obesity vs normal: P = .004, ß = 0.06; with less than elementary education, overweight vs normal: P < .0001, ß = 0.07; obesity vs normal: P = .010, ß = 0.05), immediate word recall ([1] total, overweight vs normal: P = .011, ß = 0.04; obesity vs normal: P = .002, ß = 0.04. [2] stratification by age, with age 45-58 years, obesity vs normal: P = .021, ß = 0.05; with age ≥ 59 years: overweight vs normal: P = .003, ß = 0.06. [3] stratification by educational levels, with illiterate, obesity vs normal: P = .028, ß = 0.05; with less than elementary education, obesity vs normal: P = .016, ß = 0.05), delay word recall ([1] total, overweight vs normal: P = .015, ß = 0.03; obesity vs normal: P = .031, ß = 0.03. [2] stratification by age, with age ≥ 59 years: overweight vs normal: P = .004, ß = 0.06. [3] stratification by educational levels, with less than elementary education, obesity vs normal: P = .043, ß = 0.04), self-rated memory ([1] total, obesity vs normal: P = .026, ß = 0.03. [2] stratification by age, with age ≥ 59 years, overweight vs normal: P = .044, ß = 0.04; obesity vs normal: P = .018, ß = 0.05), and drawing a picture ([1] total, overweight vs normal: OR = 1.226, 95% CI = 1.06-1.42. [2] stratification by age, with age 45-58 years: overweight vs normal: OR = 1.246, 95% CI = 1.02-1.53) than the normal weight. Regarding the association between WC and cognitive function, the obesity demonstrated better mental capacity ([1] total, men: P < .0001, ß = 0.06; women: P < .0001, ß = 0.05. [2] stratification by age, men with age 45-58 years: P < .0001, ß = 0.08; men with ≥ 59 years: P = .006, ß = 0.05. women with age 45-58 years: P = .001, ß = 0.06; women with ≥ 59 years: P = .012, ß = 0.04. [3] stratification by educational levels, men with illiterate: P = .045, ß = 0.09; men with less than elementary education: P < .0001, ß = 0.08; women with illiterate: P < .0001, ß = 0.09), ability to recall immediately ([1] total, men: P = .030, ß = 0.03; women: P = .001, ß = 0.05. [2] stratification by age, women with age 45-58 years: P = .028, ß = 0.04; women with ≥ 59 years: P = .007, ß = 0.05. [3] stratification by educational levels, men with less than elementary education: P = .007, ß = 0.05; women with illiterate: P = .027, ß = 0.05; women with less than elementary education: P = .002, ß = 0.06), delay word recall ([1] total, women: P = .044, ß = 0.03. [2] stratification by educational levels, men with less than elementary education: P = .023, ß = 0.04), self-rated memory (stratification by educational levels, women with less than elementary education: P = .030, ß = 0.04), and draw a picture ([1] total, men: OR = 1.399, 95% CI = 1.17-1.67; women: OR = 1.273, 95% CI = 1.12-1.45. [2] stratification by age, men with age 45-58 years: OR = 1.527, 95% CI = 1.15-2.03; men with age ≥ 59 years: OR = 1.284, 95% CI = 1.02-1.61; women with age 45-58 years: OR = 1.320, 95% CI = 1.10-1.58; women with age ≥ 59 years: OR = 1.223, 95% CI = 1.01-1.49. [3] stratification by educational levels, men with less than elementary education: OR = 1.528, 95% CI = 1.25-1.87; women with illiterate: OR = 1.404, 95% CI = 1.14-1.73) than the participants with normal weight after the multivariate adjustment.Our study demonstrated a significant relationship between adiposity parameters and cognition that supports the "jolly fat" hypothesis.