No abstract

Deep learning performs remarkably well on many time series analysis tasks recently. The superior performance of deep neural networks relies heavily on a large number of training data to avoid overfitting. However, the labeled data of many real-world time series applications may be limited such as classification in medical time series and anomaly detection in AIOps. As an effective way to enhance the size and quality of the training data, data augmentation is crucial to the successful application of deep learning models on time series data. In this paper, we systematically review different data augmentation methods for time series. We propose a taxonomy for the reviewed methods, and then provide a structured review for these methods by highlighting their strengths and limitations. We also empirically compare different data augmentation methods for different tasks including time series classification, anomaly detection, and forecasting. Finally, we discuss and highlight five future directions to provide useful research guidance.

The promising n-Si-based solar cell is constructed for the purpose of realizing hole- and electron-selective passivating contact, using a textured front indium tin oxide/MoO structure and a planar rear a-SiO/poly(Si(n)) structure severally. The simple MoO /n-Si heterojunction device obtains an efficiency of 16.7%. It is found that the accompanying ternary hybrid SiO(Mo) interlayer (3.5-4.0 nm) is formed at the MoO /n-Si boundary zone without preoxidation and is of amorphous structure, which is determined by a high-resolution transmission electron microscope with energy-dispersive X-ray spectroscopy mapping. The creation of lower-oxidation states in MoO film indicates that the gradient distribution of SiO with Mo element occurs within the interlayer, acting as a passivation of silicon substrate, which is revealed by X-ray photoelectron spectroscopy with depth etching. Specifically, calculations by density functional theory manifest that there are two half-filled levels (localized states) and three unoccupied levels (extended states) relating to Mo component in the ternary hybrid a-SiO(Mo) interlayer, which play the roles of defect-assisted tunneling and direct tunneling for photogenerated holes, respectively. The transport process of photogenerated holes in the MoO /n-Si heterojunction device is well-described by the tunnel-recombination model. Meanwhile, the a-SiO/poly(Si(n)) has been assembled on the rear of the device for direct tunneling of photoinduced electrons and blocking photoinduced holes.

Decomposing complex time series into trend, seasonality, and remainder components is an important task to facilitate time series anomaly detection and forecasting. Although numerous methods have been proposed, there are still many time series characteristics exhibiting in real-world data which are not addressed properly, including 1) ability to handle seasonality fluctuation and shift, and abrupt change in trend and reminder; 2) robustness on data with anomalies; 3) applicability on time series with long seasonality period. In the paper, we propose a novel and generic time series decomposition algorithm to address these challenges. Specifically, we extract the trend component robustly by solving a regression problem using the least absolute deviations loss with sparse regularization. Based on the extracted trend, we apply the the non-local seasonal filtering to extract the seasonality component. This process is repeated until accurate decomposition is obtained. Experiments on different synthetic and real-world time series datasets demonstrate that our method outperforms existing solutions.

Klebsiella pneumoniae is an important cause of healthcare-associated infections worldwide. Selective pressure, the extensive use of antibiotics, and the conjugational transmission of antibiotic resistance genes across bacterial species and genera facilitate the emergence of multidrug-resistant (MDR) K. pneumoniae. Here, we examined the occurrence, phenotypes and genetic features of MDR K. pneumoniae isolated from patients in intensive care units (ICUs) at the First Affiliated Hospital of Xiamen University in Xiamen, China, from January to December 2011. Thirty-eight MDR K. pneumoniae strains were collected. These MDR K. pneumoniae isolates possessed at least seven antibiotic resistance determinants, which contribute to the high-level resistance of these bacteria to aminoglycosides, macrolides, quinolones and β-lactams. Among these isolates, 24 strains were extended-spectrum β-lactamase (ESBL) producers, 2 strains were AmpC producers, and 12 strains were both ESBL and AmpC producers. The 38 MDR isolates also contained class I (28/38) and class II integrons (10/38). All 28 class I-positive isolates contained aacC1, aacC4, orfX, orfX’ and aadA1 genes. β-lactam resistance was conferred through bla SHV (22/38), bla TEM (10/38), and bla CTX-M (7/38). The highly conserved bla KPC-2 (37/38) and bla OXA-23(1/38) alleles were responsible for carbapenem resistance, and a gyrAsite mutation (27/38) and the plasmid-mediated qnrB gene (13/38) were responsible for quinolone resistance. Repetitive-sequence-based PCR (REP-PCR) fingerprinting of these MDR strains revealed the presence of five groups and sixteen patterns. The MDR strains from unrelated groups showed different drug resistance patterns; however, some homologous strains also showed different drug resistance profiles. Therefore, REP-PCR-based analyses can provide information to evaluate the epidemic status of nosocomial infection caused by MDR K. pneumoniae; however, this test lacks the power to discriminate some isolates. Thus, we propose that both genotyping and REP-PCR typing should be used to distinguish genetic groups beyond the species level.

This study aims to examine the properties of composites that different carbon materials with different measurements can reinforce. Using a melt compounding method, this study combines polypropylene (PP) and graphene nano-sheets (GNs) or carbon fiber (CF) to make PP/GNs and PP/CF conductive composites, respectively. The DSC results and optical microscopic observation show that both GNs and CF enable PP to crystalize at a high temperature. The tensile modulus of PP/GNs and PP/CF conductive composites remarkably increases as a result of the increasing content of conductive fillers. The tensile strength of the PP/GNs conductive composites is inversely proportional to the loading level of GNs. Containing 20 wt% of GNs, the PP/GNs conductive composites have an optimal S/m when composed of no less than 3 wt% of CF, and an optimal EMI SE of 25 dB when composed of 20 wt% of CF.

No abstract

Extracellular matrix metalloproteinase inducer, also knowns as cluster of differentiation 147 (CD147) or basigin, is a widely distributed cell surface glycoprotein that is involved in numerous physiological and pathological functions, especially in tumor invasion and metastasis. Monocarboxylate transporters (MCTs) catalyze the proton-linked transport of monocarboxylates such as L-lactate across the plasma membrane to preserve the intracellular pH and maintain cell homeostasis. As a chaperone to some MCT isoforms, CD147 overexpression significantly contributes to the metabolic transformation of tumor. This overexpression is characterized by accelerated aerobic glycolysis and lactate efflux, and it eventually provides the tumor cells with a metabolic advantage and an invasive phenotype in the acidic tumor microenvironment. This review highlights the roles of CD147 and MCTs in tumor cell metabolism and the associated molecular mechanisms. The regulation of CD147 and MCTs may prove to be with a therapeutic potential for tumors through the metabolic modification of the tumor microenvironment.