Prevention and control of COVID-19 by primary health care facilities in China: a field-survey-based qualitative study in three typical cities.

BACKGROUND: During the coronavirus disease 2019 (COVID-19) containment, primary health care (PHC) facilities inChina played an important role in providing both healthcare and public care services to community populations. The tasks of COVID-19 containment facilitated by PHC facilities were different among different regions and during different periods of COVID-19 pandemic. We sought to investigate the gaps on task participation, explore existing problems and provide corresponding solutions. METHODS: Semi-structured face-to-face interviews with COVID-19 prevention and control management teams of PHC facilities were conducted. Purposive stratified sampling was used and 32 team members of 22 PHC facilities were selected from Wuhan (as high-risk city), Shanghai (as medium-risk city) and Zunyi (as low-risk city). Framework analysis was employed to analyze the transcribed recordings. RESULTS: The main tasks of PHC facilities during the early period of the pandemic included assisting in contact tracing and epidemiological investigation, screening of populations at high-risk at travel centers/internals, house-by-house, or pre-examination/triage within PHC facilities; at-home/ centralized quarantine management; the work of fever sentinel clinics. Further analyses revealed the existing problems and suggestions for improvement or resolutions. Regular medical supply reserves were recommended because of the medical supply shortage during the pre-outbreak period. Temporarily converted quarantine wards and centralized quarantine centers could be used to deal with pressures on patients' treatment and management of the febrile patients. Only after strict evaluation of nucleic acid testing (NAT) results and housing conditions, decision on quarantine at-home or centralized quarantine centers could be made. Settings of fever sentinel clinics at PHC facilities allowed fever patients with no COVID-19 infection risks for treatment without being transferred to fever clinics of the designed secondary hospitals. Psychological intervention was sometimes in need and really helped in addressing individuals' mental pressures. CONCLUSIONS: During the COVID-19 containment, PHC facilities in China were responsible for different tasks and several problems were encountered in the working process. Accordingly, specific and feasible suggestions were put forward for different problems. Our findings are highly beneficial for healthcare teams and governments in handling similar situations.

Implementation of multi-mode nursing insulation program for patients receiving surgery for spine tumor: a propensity score-matched analysis.

BACKGROUND: Spinal tumor surgery usually involved long operation time, large area of soft tissue resection and long wound, and was prone to hypothermia during the operation. Therefore, actively promoting insulation and optimizing the intraoperative insulation program have great potential in reducing the incidence of hypothermia and reducing the incidence of postoperative complications. In this study, we compared patients who did not implement multi-mode nursing insulation program (MNIP) with those who implemented MNIP, observing and comparing clinical outcomes, and complications in both groups, with the aim of developing an optimal management plan for the preoperative, intraoperative, and postoperative periods, respectively. METHODS: We selected 2 periods of 1 year, before (n = 120 patients) and after MINP implementation (n = 120 patients). Data were collected on patient demographics, operative, perioperative details, temperature changes, anesthesia recovery effect, incidence of postoperative wound infection, length of hospital stay and complications. PS analyses were used for dealing with confounding bias in this retrospective observational study. RESULTS: After PS matching, the outcomes of 120 well-balanced pairs of patients were compared (No-MNIP vs MNIP). There was no significant difference concerning the satisfaction survey. The results indicated that the MNIP had better insulation effect at 90 min, 120 min, 150 min after anesthesia induction and after surgery. There were 16 cases of complications in the No-MNIP group and 5 cases in the MNIP group postoperative, which have significant statistical difference. CONCLUSION: In this study, the incidence of intraoperative hypothermia was effectively reduced by adopting the multi-mode insulation scheme, thus reducing the incidence of incision infection and shortening the length of hospital stay of patients.

Factorization machines and deep views-based co-training for improving answer quality prediction in online health expert question-answering services.

In online health expert question-answering (HQA) services, it is significant to automatically determine the quality of the answers. There are two prominent challenges in this task. First, the answers are usually written in short text, which makes it difficult to absorb the text semantic information. Second, it usually lacks sufficient labeled data but contains a huge amount of unlabeled data. To tackle these challenges, we propose a novel deep co-training framework based on factorization machines (FM) and deep textual views to intelligently and automatically identify the quality of HQA systems. More specifically, we exploit additional domain-specific semantic information from domain-specific word embeddings to expand the semantic space of short text and apply FM to excavate the non-independent interaction relationships among diverse features within individual views for improving the performance of the base classifier via co-training. Our learned deep textual views, the convolutional neural networks (CNN) view which focuses on extracting local features using convolution filters to locally model short text and the dependency-sensitive convolutional neural networks (DSCNN) view which focuses on capturing long-distance dependency information within the text to globally model short text, can then overcome the challenge of feature sparseness in the short text answers from the doctors. The developed co-training framework can effectively mine the highly non-linear semantic information embedded in the unlabeled data and expose the highly non-linear relationships between different views, which minimizes the labeling effort. Finally, we conduct extensive empirical evaluations and demonstrate that our proposed method can significantly improve the predictive performance of the answer quality in the context of HQA services.

A deep learning approach for predicting the quality of online health expert question-answering services.

Recently, online health expert question-answering (HQA) services (systems) have attracted more and more health consumers to ask health-related questions everywhere at any time due to the convenience and effectiveness. However, the quality of answers in existing HQA systems varies in different situations. It is significant to provide effective tools to automatically determine the quality of the answers. Two main characteristics in HQA systems raise the difficulties of classification: (1) physicians' answers in an HQA system are usually written in short text, which yields the data sparsity issue; (2) HQA systems apply the quality control mechanism, which refrains the wisdom of crowd. The important information, such as the best answer and the number of users' votes, is missing. To tackle these issues, we prepare the first HQA research data set labeled by three medical experts in 90days and formulate the problem of predicting the quality of answers in the system as a classification task. We not only incorporate the standard textual feature of answers, but also introduce a set of unique non-textual features, i.e., the popular used surface linguistic features and the novel social features, from other modalities. A multimodal deep belief network (DBN)-based learning framework is then proposed to learn the high-level hidden semantic representations of answers from both textual features and non-textual features while the learned joint representation is fed into popular classifiers to determine the quality of answers. Finally, we conduct extensive experiments to demonstrate the effectiveness of including the non-textual features and the proposed multimodal deep learning framework.

Enhanced RAGE Expression in the Dorsal Root Ganglion May Contribute to Neuropathic Pain Induced by Spinal Nerve Ligation in Rats.

OBJECTIVE: There is some evidence implicating receptor for advanced glycation end products (RAGE) signaling in the pathogenesis of neuropathic pain (NP). The objective was to investigate whether RAGE signaling in the dorsal root ganglion (DRG) might contribute to NP following peripheral nerve injury. DESIGN: Experimental study before and after spinal nerve ligation (SNL) surgery. SETTING: Caged in a controlled environment. SUBJECTS: Male Sprague-Dawley rats. METHODS: A SNL rat model of NP was used. Mechanical hyperalgesia was measured by the paw withdrawal threshold (PWT) to mechanical stimuli (1.4-15 g). Protein expressions of RAGE (immunofluorescence and western blotting), glial fibrillary acidic protein (GFAP; satellite glial cell [SGC] activation marker), IL-1ß (ELISA), TNF-α (ELISA), and NF-κB (western blotting) in the DRG were determined. RAGE signaling was inhibited by intrathecal injection of anti-RAGE antibody. RESULTS: After 7 days, SNL surgery reduced the PWT and upregulated the protein expression of RAGE, GFAP, NF-κB, TNF-α, and IL-1ß. Intrathecal injection of RAGE-neutralizing antibody attenuated the SNL-induced mechanical hyperalgesia, activation of SGCs, and upregulation of NF-κB, TNF-α, and IL-1ß in the DRG. CONCLUSION: RAGE signaling may contribute to the pain hypersensitivity observed in the rat SNL model of NP. Although the precise mechanism remains to be established, NF-κB, TNF-α, and IL-1ß likely play a role, together with the activation of SGCs.

An innovative near-infrared fluorescent probe with FRET effect for the continuous detection of Zn2+ and PPi with high sensitivity and selectivity, and its application in bioimaging.

As the second most abundant transition metal element in the human body, zinc ions play an important role in the normal growth and development of the human body. We have successfully synthesized a near-infrared fluorescent probe with FRET effect for the detection of Zn2+. Probe DR6G has good selectivity and anti-interference ability for Zn2+. When Zn2+ is added to the probe DR6G solution, it responds completely within seconds, releasing red fluorescence with a detection limit of 2.02 × 10-8 M. As the main product of ATP hydrolysis, PPi is indispensable in various metabolic activities in cells and the human body. Due to the strong binding ability of Zn2+ and PPi, it is easy to form ZnPPi precipitation, so we added PPi to the solution to complete the Zn2+ detection, and realized the continuous detection of PPi, and the detection limit was 2.06 × 10-8 M. Since Zn2+ and PPi play an important role in vivo, it is of great practical significance to design and synthesize a fluorescent probe that can continuously detect Zn2+ and PPi. Biological experiments have shown that the probe DR6G has low cytotoxicity and can complete the detection of exogenous Zn2+ and PPi in cells and living mice in vitro. Bacterial experiments have shown that the DR6G probe also has certain research value in the field of environmental monitoring and microbiology. Due to the constant variation of the fluorescence signals of Zn2+ and PPi during detection, we designed the logic gate program. In practical applications, the probe DR6G can quantitatively detect Zn2+ in zinc-containing oral liquids and qualitatively detect PPi in toothpaste.

Dual-modality probe nanodrug delivery systems with ROS-sensitivity for atherosclerosis diagnosis and therapy.

Most acute cardiovascular and cerebrovascular diseases are caused by atherosclerotic plaque rupture leading to blocked arteries. Targeted nanodelivery systems deliver imaging agents or drugs to target sites for diagnostic imaging or the treatment of various diseases, providing new insights for the detection and treatment of atherosclerosis. Based on the pathological characteristics of atherosclerosis, a hydrogen peroxide-sensitive bimodal probe PPIS@FC with integrated diagnosis and treatment function was designed. Bimodal probes Fe3O4@SiO2-CDs (FC) were prepared by coupling superparamagnetic iron oxide and carbon quantum dots synthesized with citric acid, and self-assembled with hydrogen peroxide stimulus-responsive amphiphilic block polymer PGMA-PEG modified with simvastatin (Sim) and target molecule ISO-1 to obtain drug-loaded micelles PGMA-PEG-ISO-1-Sim@FC (PPIS@FC). PPIS@FC could release Sim and FC in an H2O2-triggered manner, achieving the goal of releasing drugs using the special microenvironment at the plaque. At the same time, in vivo magnetic resonance and fluorescence imaging results proved that PPIS@FC possessed targeting ability, magnetic resonance imaging and fluorescence imaging effects. The results of the FeCl3 and ApoE-/- model showed that PPIS@FC had an excellent therapeutic effect and in vivo safety. Therefore, dual-modality imaging drug delivery systems with ROS response will become a promising strategy for the diagnosis and treatment of atherosclerosis.

Ni-CoSe2 heterojunction coated by N-doped carbon for modified separators of high-performance Li-sulfur batteries.

Functional separators modified by transition metal compounds have been proven to be effective in suppressing the shuttle effect of polysulfides and accelerating sluggish electrode dynamics in lithium-sulfur batteries (LSBs). However, the behaviors of heterojunctions composed of transition metals and their compounds in LSBs are still rarely studied. Herein, we report a novel Ni-CoSe2 heterostructure coated with nitrogen-doped carbon. Compared to homogeneous cobalt diselenide, it exhibits much stronger adsorption and catalytic conversion abilities towards polysulfides. With the modified separators, the lithium-sulfur batteries exhibit significantly improved capacity retention and reduced polarization during cycling.

Incidence of Acute Kidney Injury and Risk Factors of Prognosis in Patients with Acute Stanford Type A Aortic Dissection.

PURPOSE: We aimed to investigate the prognosis and impact of postoperative acute kidney injury (AKI) in acute Stanford type A aortic dissection (ATAAD) patients, and to analyze the predictors of short- and medium-term survival. METHODS: A total of 192 patients who underwent ATAAD surgery were included between May 2014 and May 2019. Perioperative data of these patients were analyzed. All of the discharged patients were followed up for 2 years. RESULTS: Postoperative AKI was identified in 43 of 192 patients (22.4%). The two-year survival rate of patients with AKI after discharge was 88.2% and that without AKI was 97.2%.The difference was statistically significant (χ2 = 5.355, log-rank P = 0.021). Cox hazards regression showed that age (hazard ratio [HR], 1.070; P = 0.002), cardiopulmonary bypass (CPB) time (HR, 1.026; P = 0.026), postoperative AKI (HR, 3.681; P = 0.003), and red blood cell transfusion (HR, 1.548; P = 0.001) were independent risk factors for the short- and medium-term total mortality of ATAAD patients. CONCLUSION: The incidence of postoperative AKI is high in ATAAD, and the mortality of patients with AKI increases significantly within 2 years. Age, CPB time, and red blood cell transfusion were also independent risk factors for short-and medium-term prognoses.

Shear Stress and ROS Dual-Responsive RBC-Hitchhiking Nanoparticles for Atherosclerosis Therapy.

Atherosclerosis (AS), a leading cause of death worldwide, is a chronic inflammatory disease rich in lipids and reactive oxygen species (ROS) within plaques. Therefore, lowering lipid and ROS levels is effective in treating AS and reducing AS-induced mortality. In this study, an intelligent biomimetic drug delivery system that specifically responded to both shear stress and ROS microenvironment was developed, consisting of red blood cells (RBCs) and cross-linked polyethyleneimine nanoparticles (SA PEI) loaded with a lipid-lowering drug simvastatin acid (SA), and RBCs were self-assembled with SA PEI to obtain biresponsive SA PEI@RBCs for the treatment of AS. SA PEI could achieve sustained release of SA in response to ROS and reduce ROS and lipid levels to achieve the purpose of treating AS. Shear stress model experiments showed that SA PEI@RBCs could respond to the high shear stress level (100 dynes/cm2) at plaques, realizing the desorption and enrichment of SA PEI and improving the therapeutic efficiency of SA PEI@RBCs. In vitro and in vivo experiments have confirmed that SA PEI@RBCs exhibits better in vivo safety and therapeutic efficacy than SA PEI and free SA. Therefore, shaping SA PEI@RBCs into a biomimetic drug delivery system with dual sensitivity to ROS and shear stress is an effective strategy and treatment to facilitate their delivery into plaques.

Stromal cells in the tumor microenvironment: accomplices of tumor progression?

The tumor microenvironment (TME) is made up of cells and extracellular matrix (non-cellular component), and cellular components include cancer cells and non-malignant cells such as immune cells and stromal cells. These three types of cells establish complex signals in the body and further influence tumor genesis, development, metastasis and participate in resistance to anti-tumor therapy. It has attracted scholars to study immune cells in TME due to the significant efficacy of immune checkpoint inhibitors (ICI) and chimeric antigen receptor T (CAR-T) in solid tumors and hematologic tumors. After more than 10 years of efforts, the role of immune cells in TME and the strategy of treating tumors based on immune cells have developed rapidly. Moreover, ICI have been recommended by guidelines as first- or second-line treatment strategies in a variety of tumors. At the same time, stromal cells is another major class of cellular components in TME, which also play a very important role in tumor metabolism, growth, metastasis, immune evasion and treatment resistance. Stromal cells can be recruited from neighboring non-cancerous host stromal cells and can also be formed by transdifferentiation from stromal cells to stromal cells or from tumor cells to stromal cells. Moreover, they participate in tumor genesis, development and drug resistance by secreting various factors and exosomes, participating in tumor angiogenesis and tumor metabolism, regulating the immune response in TME and extracellular matrix. However, with the deepening understanding of stromal cells, people found that stromal cells not only have the effect of promoting tumor but also can inhibit tumor in some cases. In this review, we will introduce the origin of stromal cells in TME as well as the role and specific mechanism of stromal cells in tumorigenesis and tumor development and strategies for treatment of tumors based on stromal cells. We will focus on tumor-associated fibroblasts (CAFs), mesenchymal stem cells (MSCs), tumor-associated adipocytes (CAAs), tumor endothelial cells (TECs) and pericytes (PCs) in stromal cells.

Anti-apoptosis effect of heme oxygenase-1 on lung injury after cardiopulmonary bypass.

BACKGROUND: This study aimed to investigate the anti-apoptosis effects of heme oxygenase-1 (HO-1) on lung injury (LI) after cardiopulmonary bypass (CPB) and its probable mechanisms. METHODS: One hundred and forty-four male Wistar rats were divided into 3 groups randomly: group A (control group), group B (cobalt protoporphyrin, CoPP), and group C [CoPP plus zinc protoporphyrin (ZnPP)]. Lung tissues were harvested at different time: before CPB (T0), 0 min after CPB (T1), 2 h after CPB (T2), 6 h (T3), 12 h (T4), and 24 h (T5). RESULTS: The HO-1 protein expressions in lung tissue in group B were higher than those in group A and group C in any given time, and the same as HO-1 activity (P<0.05). The expressions of Bcl-2 protein in group B at all time point after bypass were higher than those in group A and group C, and the difference was statistically significant (P<0.05). Apoptosis index (AI) in group B at any time point after bypass were lower than those in group A and group C (P<0.05). CONCLUSIONS: CoPP can significantly increase the expression of HO-1 protein in lung tissue. HO-1 is still highly expressed after CPB, so as to play an important part in anti-apoptosis, and reduce LI.

Block-term tensor neural networks.

Deep neural networks (DNNs) have achieved outstanding performance in a wide range of applications, e.g., image classification, natural language processing, etc. Despite the good performance, the huge number of parameters in DNNs brings challenges to efficient training of DNNs and also their deployment in low-end devices with limited computing resources. In this paper, we explore the correlations in the weight matrices, and approximate the weight matrices with the low-rank block-term tensors. We name the new corresponding structure as block-term tensor layers (BT-layers), which can be easily adapted to neural network models, such as CNNs and RNNs. In particular, the inputs and the outputs in BT-layers are reshaped into low-dimensional high-order tensors with a similar or improved representation power. Sufficient experiments have demonstrated that BT-layers in CNNs and RNNs can achieve a very large compression ratio on the number of parameters while preserving or improving the representation power of the original DNNs.

Effective Data-Aware Covariance Estimator From Compressed Data.

Estimating covariance matrix from massive high-dimensional and distributed data is significant for various real-world applications. In this paper, we propose a data-aware weighted sampling-based covariance matrix estimator, namely DACE, which can provide an unbiased covariance matrix estimation and attain more accurate estimation under the same compression ratio. Moreover, we extend our proposed DACE to tackle multiclass classification problems with theoretical justification and conduct extensive experiments on both synthetic and real-world data sets to demonstrate the superior performance of our DACE.

Online Nonlinear AUC Maximization for Imbalanced Data Sets.

Classifying binary imbalanced streaming data is a significant task in both machine learning and data mining. Previously, online area under the receiver operating characteristic (ROC) curve (AUC) maximization has been proposed to seek a linear classifier. However, it is not well suited for handling nonlinearity and heterogeneity of the data. In this paper, we propose the kernelized online imbalanced learning (KOIL) algorithm, which produces a nonlinear classifier for the data by maximizing the AUC score while minimizing a functional regularizer. We address four major challenges that arise from our approach. First, to control the number of support vectors without sacrificing the model performance, we introduce two buffers with fixed budgets to capture the global information on the decision boundary by storing the corresponding learned support vectors. Second, to restrict the fluctuation of the learned decision function and achieve smooth updating, we confine the influence on a new support vector to its -nearest opposite support vectors. Third, to avoid information loss, we propose an effective compensation scheme after the replacement is conducted when either buffer is full. With such a compensation scheme, the performance of the learned model is comparable to the one learned with infinite budgets. Fourth, to determine good kernels for data similarity representation, we exploit the multiple kernel learning framework to automatically learn a set of kernels. Extensive experiments on both synthetic and real-world benchmark data sets demonstrate the efficacy of our proposed approach.

Maximizing sensitivity in medical diagnosis using biased minimax probability machine.

The challenging task of medical diagnosis based on machine learning techniques requires an inherent bias, i.e., the diagnosis should favor the "ill" class over the "healthy" class, since misdiagnosing a patient as a healthy person may delay the therapy and aggravate the illness. Therefore, the objective in this task is not to improve the overall accuracy of the classification, but to focus on improving the sensitivity (the accuracy of the "ill" class) while maintaining an acceptable specificity (the accuracy of the "healthy" class). Some current methods adopt roundabout ways to impose a certain bias toward the important class, i.e., they try to utilize some intermediate factors to influence the classification. However, it remains uncertain whether these methods can improve the classification performance systematically. In this paper, by engaging a novel learning tool, the biased minimax probability machine (BMPM), we deal with the issue in a more elegant way and directly achieve the objective of appropriate medical diagnosis. More specifically, the BMPM directly controls the worst case accuracies to incorporate a bias toward the "ill" class. Moreover, in a distribution-free way, the BMPM derives the decision rule in such a way as to maximize the worst case sensitivity while maintaining an acceptable worst case specificity. By directly controlling the accuracies, the BMPM provides a more rigorous way to handle medical diagnosis; by deriving a distribution-free decision rule, the BMPM distinguishes itself from a large family of classifiers, namely, the generative classifiers, where an assumption on the data distribution is necessary. We evaluate the performance of the model and compare it with three traditional classifiers: the k-nearest neighbor, the naive Bayesian, and the C4.5. The test results on two medical datasets, the breast-cancer dataset and the heart disease dataset, show that the BMPM outperforms the other three models.

Imbalanced learning with a biased minimax probability machine.

Imbalanced learning is a challenged task in machine learning. In this context, the data associated with one class are far fewer than those associated with the other class. Traditional machine learning methods seeking classification accuracy over a full range of instances are not suitable to deal with this problem, since they tend to classify all the data into a majority class, usually the less important class. In this correspondence, the authors describe a new approach named the biased minimax probability machine (BMPM) to deal with the problem of imbalanced learning. This BMPM model is demonstrated to provide an elegant and systematic way for imbalanced learning. More specifically, by controlling the accuracy of the majority class under all possible choices of class-conditional densities with a given mean and covariance matrix, this model can quantitatively and systematically incorporate a bias for the minority class. By establishing an explicit connection between the classification accuracy and the bias, this approach distinguishes itself from the many current imbalanced-learning methods; these methods often impose a certain bias on the minority data by adapting intermediate factors via the trial-and-error procedure. The authors detail the theoretical foundation, prove its solvability, propose an efficient optimization algorithm, and perform a series of experiments to evaluate the novel model. The comparison with other competitive methods demonstrates the effectiveness of this new model.

Is tuberculosis health education reaching the public in China? A cross-sectional survey in Guizhou Province.

BACKGROUND: Knowledge about tuberculosis (TB) is important for TB control, and China's national TB control guidelines emphasise TB health promotion. A 2010 national TB epidemiology survey showed that the general public had limited knowledge and awareness of TB. OBJECTIVE: To assess the level of TB knowledge after 5 years of TB health promotion in Guizhou Province, one of the regions with the highest TB burden in China. DESIGN AND SETTING: A community-based, cross-sectional survey of 10 237 residents of Guizhou Province from June to August 2015. Multiple logistic regression models were used to examine factors associated with core TB knowledge and TB health education among respondents. RESULTS: Overall, residents of Guizhou Province had inadequate knowledge of TB. The overall awareness of TB was 41.5%. Less than 30% of respondents were familiar with China's policy of free treatment for TB or knew that the disease could be cured. Factors associated with core TB knowledge included gender, age, ethnicity, education, occupation, region, and having received TB health education. Women, older adults, people employed in non-government institutions, and those living in counties with low TB burdens had little access to TB health education, whereas people with higher education levels had greater access. Respondents' sources of TB knowledge did not necessarily match their preferred channels for delivery of TB health education. CONCLUSIONS: Our findings indicate that TB health education should be further strengthened in China and other countries with a high TB burden. TB health education programmes require further formative and implementation research in order to improve programme effectiveness.

Maximum margin semi-supervised learning with irrelevant data.

Semi-supervised learning (SSL) is a typical learning paradigms training a model from both labeled and unlabeled data. The traditional SSL models usually assume unlabeled data are relevant to the labeled data, i.e., following the same distributions of the targeted labeled data. In this paper, we address a different, yet formidable scenario in semi-supervised classification, where the unlabeled data may contain irrelevant data to the labeled data. To tackle this problem, we develop a maximum margin model, named tri-class support vector machine (3C-SVM), to utilize the available training data, while seeking a hyperplane for separating the targeted data well. Our 3C-SVM exhibits several characteristics and advantages. First, it does not need any prior knowledge and explicit assumption on the data relatedness. On the contrary, it can relieve the effect of irrelevant unlabeled data based on the logistic principle and maximum entropy principle. That is, 3C-SVM approaches an ideal classifier. This classifier relies heavily on labeled data and is confident on the relevant data lying far away from the decision hyperplane, while maximally ignoring the irrelevant data, which are hardly distinguished. Second, theoretical analysis is provided to prove that in what condition, the irrelevant data can help to seek the hyperplane. Third, 3C-SVM is a generalized model that unifies several popular maximum margin models, including standard SVMs, Semi-supervised SVMs (S(3)VMs), and SVMs learned from the universum (U-SVMs) as its special cases. More importantly, we deploy a concave-convex produce to solve the proposed 3C-SVM, transforming the original mixed integer programming, to a semi-definite programming relaxation, and finally to a sequence of quadratic programming subproblems, which yields the same worst case time complexity as that of S(3)VMs. Finally, we demonstrate the effectiveness and efficiency of our proposed 3C-SVM through systematical experimental comparisons.

Budget constrained non-monotonic feature selection.

Feature selection is an important problem in machine learning and data mining. We consider the problem of selecting features under the budget constraint on the feature subset size. Traditional feature selection methods suffer from the "monotonic" property. That is, if a feature is selected when the number of specified features is set, it will always be chosen when the number of specified feature is larger than the previous setting. This sacrifices the effectiveness of the non-monotonic feature selection methods. Hence, in this paper, we develop an algorithm for non-monotonic feature selection that approximates the related combinatorial optimization problem by a Multiple Kernel Learning (MKL) problem. We justify the performance guarantee for the derived solution when compared to the global optimal solution for the related combinatorial optimization problem. Finally, we conduct a series of empirical evaluation on both synthetic and real-world benchmark datasets for the classification and regression tasks to demonstrate the promising performance of the proposed framework compared with the baseline feature selection approaches.