Search

Current unsupervised domain adaptation methods for intelligent fault diagnosis focus on learning domain-invariant representations under covariate shift assumption. However, the covariate shift assumption is usually unsatisfied when each fault class in different domains consists of multiple modes with skewed proportions, which is common in industrial scenarios. Imbalanced data from multiple modes cause the presence of within-domain class imbalance and between-domain label distributional shift. This paper introduces a novel subpopulation shift that further considers the domain shift from a subpopulation perspective, i.e., that covariate shift and label distributional shift across domains are caused by shifts in the multiple modes. To address this issue, a label propagation-based unsupervised domain adaptation is proposed based on a realistic expansion assumption. We apply the theoretical analysis of the proposed method with a bi-level optimization strategy adapted from meta-learning. Using joint optimization of a teacher model and a student model, the label propagation-based model-agnostic meta-learning (LPMAML) not only propagates supervision information from the source to the target but also adjusts the teacher's strategy throughout the student's learning process. To alleviate the noise caused by label distributional shift, we integrate a sampling-based alignment method that aligns the empirical label distributions across the two domains into LPMAML. Experimental results on three bearing datasets show that the proposed method has impressive generalization ability under covariate, label distributional, and subpopulation shifts. The proposed method offers consistent improvements to unsupervised domain adaptation (UDA) methods. Compared with the vanilla UDA methods, the average diagnosis accuracies of the proposed method on the label distributional shift benchmark and subpopulation shift benchmark are improved by 8.21% and 7.63%, respectively. [ABSTRACT FROM AUTHOR]

Autonomous driving has been widely applied in commercial and industrial applications, along with the upgrade of environmental awareness systems. Tasks such as path planning, trajectory tracking, and obstacle avoidance are strongly dependent on the ability to perform real-time object detection and position regression. Among the most commonly used sensors, camera provides dense semantic information but lacks accurate distance information to the target, while LiDAR provides accurate depth information but with sparse resolution. In this paper, a LiDAR-camera-based fusion algorithm is proposed to improve the above-mentioned trade-off problems by constructing a Siamese network for object detection. Raw point clouds are converted to camera planes to obtain a 2D depth image. By designing a cross feature fusion block to connect the depth and RGB processing branches, the feature-layer fusion strategy is applied to integrate multi-modality data. The proposed fusion algorithm is evaluated on the KITTI dataset. Experimental results demonstrate that our algorithm has superior performance and real-time efficiency. Remarkably, it outperforms other state-of-the-art algorithms at the most important moderate level and achieves excellent performance at the easy and hard levels. [ABSTRACT FROM AUTHOR]

Due to the existence of domain shifts, the distributions of data acquired from different machines show significant discrepancies in industrial applications, which leads to performance degradation of traditional machine learning methods. In this paper, we propose a novel method that combines supervised domain adaptation with prototype learning for fault diagnosis. The proposed method consists of two modules, i.e., feature learning and condition recognition. The module of feature learning applies the Siamese architecture based on one-dimensional convolutional neural networks to learn a domain invariant subspace, which reduces the inter-domain discrepancy of distributions. The module of condition recognition applies a prototypical layer to learn the prototypes of each class. Then the classification task is simplified to find the nearest class prototype. Compared with existing intelligent fault diagnosis methods, this proposed method can be extended to address the problem when the classes from the source and target domains are partially overlapped. The model must generalize to unknown classes in the source domain, given only a few samples of each new target class. The effectiveness of the proposed method is verified using two bearing datasets. The model quickly converges with high classification accuracy using a few labeled target samples in training, even one per class can be effective. [ABSTRACT FROM AUTHOR]

Silky chicken is a breed of chickens with black skin and slow growth rate used in Chinese traditional medicine, whereas Arbor Acres broiler is a well-known commercial breed in the poultry industry, it is featured by a large size, rapid-growth rate, high feed-conversion rate and strong adaptability. The difference in their rate of growth may be primarily related to different mechanism for glucose metabolism. Here we compared the insulin sensitivity of the two breeds; we investigated the temporal changes (at 0 min, 120 min and 240 min) of serum insulin and other biochemical parameters and determined the spatio-temporal changes of gene mRNA abundance in response to exogenous insulin (80 μg/kg body weight). The results indicated that: (1) Silky chickens showed stronger blood glucose recovery than broilers in the insulin resistance test. (2) The serum urea level in Silky chickens was twice of broilers; exogenous insulin significantly up-regulated serum uric acid level in Silky fowls in a time-dependent manner and increased serum cholesterol content at 120 min. (3) Two breeds showed distinctly different temporal changed in serum insulin in response to exogenous insulin stimulation. The fasting serum insulin concentration of broilers was three-fold of Silky chickens at the basal state; it decreased significantly after insulin injection and the levels at 120 min and 240 min of broilers were only 23% (P < 0.01) and 14% (P < 0.01) of the basal state, respectively. Whereas the serum insulin content in Silky chickens showed stronger recovery, and the 240 min level was close to the 0 min level. (4) GLUT2, GLUT12, neuropeptide Y and insulin receptor (IR) were predominantly expressed in the liver, pectoralis major, olfactory bulb and pancreas, respectively, where these genes presented stronger insulin sensitivity. In addition, the IR mRNA level was strongly positively with the GLUT12 level. In conclusion, our findings suggested that Silky chickens have a stronger ability to regulate glucose homeostasis than broilers, owing to their higher IR levels in the basal state, stronger serum insulin homeostasis and candidate genes functioning primarily in their predominantly expressed tissue in response to exogenous insulin. [ABSTRACT FROM AUTHOR]

Mitochondrial DNA (mtDNA) copy number reflects the abundance of mitochondria in cells and is dependent on the energy requirements of tissues. We hypothesized that the mtDNA copy number in poultry may change with age and tissue, and feed restriction may affect the growth and health of poultry by changing mtDNA content in a tissue-specific pattern. TaqMan real-time PCR was used to quantify mtDNA copy number using three different segments of the mitochondrial genome (D-loop, ATP6, and ND6) relative to the nuclear single-copy preproglucagon gene (GCG). The effect of sex, age, and dietary restriction (quantitative, energy, and protein restriction) on mtDNA copy number variation in the tissues of broilers was investigated. We found that mtDNA copy number varied among tissues (P < 0.01) and presented a distinct change in spatiotemporal pattern. After hatching, the number of mtDNA copies significantly decreased with age in the liver and increased in muscle tissues, including heart, pectoralis, and leg muscles. Newborn broilers (unfed) and embryos (E 11 and E 17) had similar mtDNA contents in muscle tissues. Among 42 d broilers, females had a higher mtDNA copy number than males in the tissues examined. Feed restriction (8–21 d) significantly reduced the body weight but did not significantly change the mtDNA copy number of 21 d broilers. After three weeks of compensatory growth (22–42 d), only the body weight of broilers with a quantitatively restricted diet remained significantly lower than that of broilers in the control group (P < 0.05), while any type of early feed restriction significantly reduced the mtDNA copy number in muscle tissues of 42 d broilers. In summary, the mtDNA copy number of broilers was regulated in a tissue- and age-specific manner. A similar pattern of spatiotemporal change in response to early feed restriction was found in the mtDNA content of muscle tissues, including cardiac and skeletal muscle, whereas liver mtDNA content changed differently with age and dietary restriction. It seems that early restrictions in feed could effectively lower the mtDNA content in muscle cells to reduce the tissue overload in broilers at 42 d to some degree. [ABSTRACT FROM AUTHOR]

The stratum corneum (SC), the outermost layer of skin, represents the primary barrier to molecules penetrating the skin. Menthol is widely used in clinical medicine as a penetration enhancer due to its high efficiency and relative safety. In this study, molecular dynamics simulations have been performed to investigate the effect of menthol molecules on the structural and permeability of both single component and ternary mixed bilayers. The lipid matrix is modeled as pure ceramide (CER2) or as a 2:2:1 mixture of CER2, cholesterol (CHOL), and free fatty acid (FFA). The effect of menthol on the SC bilayer was investigated at various concentrations of menthol. For both models, the area per lipid decreases and the membrane thickness increases with increased menthol concentration, which may be due to the fact that menthol molecules penetrate into the bilayer and aggregate at the bilayer center. As for ternary mixed bilayer at high concentration, the lipids rearranged, and one more layer formed inside the former two leaflets. Our simulation results are consistent with the experimental evidence that high concentrations of menthol fluidize the SC lipids and enhance permeability. The penetration enhancement of menthol may take place through direct interactions with lipids rather than by forming water pores. [ABSTRACT FROM AUTHOR]

Both the cycloaddition mechanism of 5-methylcytosine with adenine and the deamination mechanism of the cycloaddition product have been studied using density functional theory method. The results suggest that the cycloaddition reaction could occur more easily through photochemical reaction pathway than through thermal reaction pathway. The obtained four-member ring structure could be easily transformed to an eight-member ring structure through bond cleavage of C-C (the energy barrier is <2 kcal/mol). Then hydrolytic deamination reaction takes place with water assistance. The hydroxyl group of one water molecule attacks the C atom and the hydrogen atom of another water molecule attacks N atom to form a tetrahedral intermediate. Subsequently, the hydrogen atom of hydroxyl group transfers to N to produce ammonia, and the amino group of the former 5-methylcytosine changes to carboxyl oxygen. Our calculations explain the phenomena that 5-methylcytosine and adenine could obtain the same photoproduct as thymine and adenine from theoretical aspects. [ABSTRACT FROM AUTHOR]