3-D Deconvolutional Networks for the Unsupervised Representation Learning of Human Motions.

Data representation learning is one of the most important problems in machine learning. Unsupervised representation learning becomes meritorious as it has no necessity of label information with observed data. Due to the highly time-consuming learning of deep-learning models, there are many machine-learning models directly adapting well-trained deep models that are obtained in a supervised and end-to-end manner as feature abstractors to distinct problems. However, it is obvious that different machine-learning tasks require disparate representation of original input data. Taking human action recognition as an example, it is well known that human actions in a video sequence are 3-D signals containing both visual appearance and motion dynamics of humans and objects. Therefore, the data representation approaches with the capabilities to capture both spatial and temporal correlations in videos are meaningful. Most of the existing human motion recognition models build classifiers based on deep-learning structures such as deep convolutional networks. These models require a large quantity of training videos with annotations. Meanwhile, these supervised models cannot recognize samples from the distinct dataset without retraining. In this article, we propose a new 3-D deconvolutional network (3DDN) for representation learning of high-dimensional video data, in which the high-level features are obtained through the optimization approach. The proposed 3DDN decomposes the video frames into spatiotemporal features under a sparse constraint in an unsupervised way. In addition, it also can be regarded as a building block to develop deep architectures by stacking. The high-level representation of input sequential data can be used in multiple downstream machine-learning tasks, we evaluate the proposed 3DDN and its deep models in human action recognition. The experimental results from three datasets: 1) KTH data; 2) HMDB-51; and 3) UCF-101, demonstrate that the proposed 3DDN is an alternative approach to feedforward convolutional neural networks (CNNs), that attains comparable results.

Correlation of the leucocyte count with traditional and non-traditional components of metabolic syndrome.

OBJECTIVES: To investigate correlation of the white blood cell (WBC) and its subtype count with the traditional and non-traditional components of the metabolic syndrome. METHODS: Between January 2012 and December 2013, 18,222 people were enrolled in this study. The height, weight, body mass index (BMI) and blood pressure were measured, and blood samples were tested for all subjects after an overnight fast. The count of WBC and its subtypes, total cholesterol, triglyceride, high density lipoprotein (HDL), low-density lipoprotein, aminotransferases, fibrinogen, uric acid, and fasting blood glucose were all assessed. RESULTS: Metabolic syndrome was found in 2502 of 18,222 healthy Chinese people (16.41%). The prevalence of metabolic syndrome was 22.61% for men significantly (P < 0.05) greater than for women (6.83%). The prevalence of obesity, hypertension, hyperglycemia and hyperlipidemia was significantly (P < 0.001) higher in people with than without metabolic syndrome. With increase of the WBC count, BMI, systolic and diastolic pressure, fasting blood glucose, triglyceride, glutamic-oxaloacetic transaminase, glutamic-pyruvic transaminase, glutamyltranspetidase, blood urea nitrogen fibrinogen and uric acid all went up significantly (P < 0.001) while HDL decreased significantly (P < 0.05). The creatinine remained relatively sTable After adjustment of age, sex, alcoholic drinking and education, the metabolic components of obesity, hypertension, diabetes and hyperlipidemia rose significantly (P < 0.05) positively with increased counts of the total WBC, neutrophil and lymphocyte, and the WBC and its subtypes were an independent risk factor for metabolic syndrome. CONCLUSION: Aminotransferases, fibrinogen and uric acid all significantly increase with increased WBC count in a dose-dependent manner. Increased counts of the total WBC and its subtypes are positively associated with presence of metabolic syndrome.

Management of antibiotic-associated pseudomembranous colitis in Non-hospitalized and hospitalized patients.

Proper management of antibiotic-associated pseudo membranous colitis is not clear. This article is to investigate proper treatment of antibiotic-associated pseudo membranous colitis. Data of 67 patients (aged 18-69 years, with 31 males and 46 females) with antibiotic-associated pseudo membranous colitis were retrospectively analyzed including the demography, antibiotics to induce and for treatment of the pseudo membranous colitis, and other supportive measures. All 67 patients had a positive cytotoxin test, which confirmed the pseudo membranous colitis. Antibiotics which induced the pseudo membranous colitis included clindamycin, ofloxacin, piperacillin, cefatriaxone, penbritin and ceftazidime. Once the correct diagnosis was made, the culprit antibiotics were discontinued immediately, and narrow-spectrum antibiotics like metronidazole and vancomycin were administered in combination with correction of fluid and electrolyte abnormalities, use of vitamins C and B complex to repair the intestinal mucosa, and avoidance of antispasmodic and antidiarrheal agents. After appropriate treatment for 2-20 days, all patients recovered with no sequela. Sixty-two patients were clinically cured while five (7.5%) had diarrhea recurrence within two months of the end of therapy. Retreatment with tapering and extended period of metronidazole and/or vancomycin led to complete recovery of the patients. Multiple antibiotic agents are associated with pseudo membranous colitis, and correction of fluid and electrolyte abnormalities and use of vitamins to repair the intestinal mucosa should be performed to speed up the cure process.