Search

Convolutional Neural Networks (CNNs) have been widely applied in various edge computing devices based on intelligent sensors. However, due to the high computational demands of CNN tasks, the limited computing resources of edge intelligent terminal devices, and significant architectural differences among these devices, it is challenging for edge devices to independently execute inference tasks locally. Collaborative inference among edge terminal devices can effectively utilize idle computing and storage resources and optimize latency characteristics, thus significantly addressing the challenges posed by the computational intensity of CNNs. This paper targets efficient collaborative execution of CNN inference tasks among heterogeneous and resource-constrained edge terminal devices. We propose a pre-partitioning deployment method for CNNs based on critical operator layers, and optimize the system bottleneck latency during pipeline parallelism using data compression, queuing, and "micro-shifting" techniques. Experimental results demonstrate that our method achieves significant acceleration in CNN inference within heterogeneous environments, improving performance by 71.6% compared to existing popular frameworks. [ABSTRACT FROM AUTHOR]

Remote sensing image scene classification is a fundamental task in intelligent interpretation of remote sensing images. Although Transformers possess a powerful attention mechanism, they require lengthy training procedures to achieve good performance levels. To address this issue, this paper proposes a novel deep learning network model by combining CNN and Swin Transformer named SwinHCST. Firstly, the model uses Weighted Normalized CNN to quickly extract low-level features of the image. Secondly, the Receptive Field Block module facilitates multi-scale information fusion, Thirdly, the Information Fusion Transformer further excavates the deep-level features of the image. Furthermore, this paper has designed a plug-and-play Cross Spatial Information Fusion Block, which is used to encodes dimensional information and extracts global information to enhance information exchange. The scene classification experiments show that the proposed model outperforms other methods on the three selected datasets and can achieve excellent performance without requiring large amounts of data and training. Specifically, the classification accuracy of the proposed method on the three datasets is 93.76%, 93.60%, and 98.10%, which is 1.7% to 3.71% higher than ResNet50 and 3.7% to 5.7% higher than Swin Transformer. [ABSTRACT FROM AUTHOR]

The technique of pig cloning holds great promise for the livestock industry, life science, and biomedicine. However, the prenatal death rate of cloned pig embryos is extremely high, resulting in a very low cloning efficiency. This limits the development and application of pig cloning. In this study, we utilized embryo biopsy combined with microproteomics to identify potential factors causing the developmental arrest in cloned pig embryos. We verified the roles of two potential regulators, PDCD6 and PLK1, in cloned pig embryo development. We found that siRNA-mediated knockdown of PDCD6 reduced mRNA and protein expression levels of the pro-apoptotic gene, CASP3, in cloned pig embryos. PDCD6 knockdown also increased the cleavage rate and blastocyst rate of cloned porcine embryos. Overexpression of PLK1 via mRNA microinjection also improved the cleavage rate of cloned pig embryos. This study provided a new strategy to identify key factors responsible for the developmental defects in cloned pig embryos. It also helped establish new methods to improve pig cloning efficiency, specifically by correcting the expression pattern of PDCD6 and PLK1 in cloned pig embryos. [ABSTRACT FROM AUTHOR]

The technique of cloning has wide applications in animal husbandry and human biomedicine. However, the very low developmental efficiency of cloned embryos limits the application of cloning. Ectopic XIST-expression-induced abnormal X chromosome inactivation (XCI) is a primary cause of the low developmental competence of cloned mouse and pig embryos. Knockout or knockdown of XIST improves cloning efficiency in both pigs and mice. The transcription factor Yin yang 1(YY1) plays a critical role in XCI by triggering the transcription of X-inactive specific transcript (XIST) and facilitating the localization of XIST RNA on the X chromosome. This study aimed to investigate whether RNA interference to suppress the expression of YY1 can inhibit erroneous XIST expression, rescue abnormal XCI, and improve the developmental ability of cloned pig embryos. The results showed that YY1 binds to the 5′ regulatory region of the porcine XIST gene in pig cells. The microinjection of YY1 siRNA into cloned pig embryos reduced the transcript abundance of XIST and upregulated the mRNA level of X-linked genes at the 4-cell and blastocyst stages. The siRNA-mediated knockdown of YY1 altered the transcriptome and enhanced the in vitro and in vivo developmental efficiency of cloned porcine embryos. These results suggested that YY1 participates in regulating XIST expression and XCI in cloned pig embryos and that the suppression of YY1 expression can increase the developmental rate of cloned pig embryos. The present study established a new method for improving the efficiency of pig cloning. [ABSTRACT FROM AUTHOR]

In the simulation process for multi-body systems, the generated redundant constraints will result in ill-conditioned dynamic equations, which are not good for stable simulations when the system motion proceeds near a singular configuration. In order to overcome the singularity problems, the paper presents a regularization method with an explicit expression based on Gauss principle, which does not need to eliminate the constraint violation after each iteration step compared with the traditional methods. Then the effectiveness and stability are demonstrated through two numerical examples, a slider-crank mechanism and a planar four-bar linkage. Simulation results obtained with the proposed method are analyzed and compared with augmented Lagrangian formulation and the null space formulation in terms of constraints violation, drift mechanical energy and computational efficiency, which shows that the proposed method is suitable to perform efficient and stable dynamic simulations for multi-body systems with singular configurations. [ABSTRACT FROM AUTHOR]

The discontinuous dynamical problem of multi-point contact and collision in multi-body system has always been a hot and difficult issue in this field. Based on the Gauss' principle of least constraint, a unified optimization model for multibody system dynamics with multi-point contact and collision is established. The paper presents the study of the numerical solution scheme, in which particle swarm optimization method is used to deal with the corresponding optimization model. The article also presents the comparison of the Gauss optimization method (GOM) and the hybrid linear complementarity method (i.e. combining differential algebraic equations (DAEs) and linear complementarity problems (LCP)), commonly used to solve the dynamic contact problem of multibody systems with bilateral constraints. The results illustrate that, the GOM has the same advantage of dynamical modelling with LCP and when the redundant constraint exists, the GOM always has a unique solution and so no additional processing is needed, whereas the corresponding DAE-LCP method may have singular cases with multiple solutions or no solutions. Using numerical examples, the GOM is verified to effectively solve the dynamics of multibody systems with redundant unilateral and bilateral constraints without additional redundancy processing. The GOM can also be applied to the optimal control of systems in the future and combined with the parameter optimization of systems to handle dynamic problems. The work given provides the dynamics and control of the complex system with a new train of thought and method. The paper establishes a unified optimization model and the corresponding numerical solution scheme for multibody system dynamics with multi-point contact and collision based on Gauss' principle of least constraint. Comparisons between Gauss optimization method (GOM) and the hybrid linear complementarity method (HLCM) are carried out through theoretical analysis and numerical examples, which show that GOM has the same advantage of dynamical modelling as HLCM and when redundant constraints exist, GOM has unique solutions without additional processes, whereas HLCM may have singular cases with multiple or no solutions. [ABSTRACT FROM AUTHOR]

Redundancy in the constrained mechanical systems often occurs in complex multibody mechanic systems in the existence of excessive constraints and singular positions due to system motion. In this work, Gauss principle of least constraint (GPLC) is applied to solve the dynamic motion of system with redundant constraints without changing the physics of system. Furthermore, the particle swarm optimization method is used to handle the minimization optimization problem. Eventually, the effectiveness of GPLC is validated through the dynamic modelling and simulation of two numerical examples (a planar four-bar mechanism and a spatial parallelogram mechanism). The simulation results are analyzed and compared with those obtained from Udwaia-Phohomsiri formulation and augmented Lagrangian formulation, in terms of constraint violation, computational efficiency and variation of the mechanical energy. From the viewpoint of computational efficiency and accuracy, GPLC can be regarded as a practical real-time simulation method for multibody systems with redundant constraints. [ABSTRACT FROM AUTHOR]

Background: Intracranial angiomatous meningioma (AM) is a rare subtype of meningioma. Here, we investigated the clinical and pathological features of AMs. Materials and Methods: We performed a retrospective study of 23 intracranial AMs verified by postoperative pathology at Huashan Hospital North between 2013 and 2018. Clinical data, radiological and pathological findings, and information on treatment and outcomes were collected and analyzed. Additionally, the literature on intracranial AMs was reviewed. Results: The sample comprised 13 men and 10 women with AMs. The mean age was 54.2 years, and the mean duration of symptoms was 14.9 months. Headache and epilepsy were the most common symptoms. The most common AMs locations were the cerebral convexity and parasagittal/falx region. The rates of vascular signs, homogeneous enhancement, and peritumoral brain edema (PTBE) on magnetic resonance images were high. Histologically, besides typical meningioma cells, AMs had an abundant vascular component and low Ki-67 index. The extent of PTBE was related to microvessel density (MVD) of tumors, but not to the expression of MMP9 or VEGF. Simpson grade I resection was achieved in 15 cases, and grade II resection was achieved in 7 cases. Twenty-one cases were followed up, and they all had favorable outcomes without recurrence. Conclusion: AM is a type of meningioma with a rich blood supply and distinct clinical and pathological features. It showed a slight male predominance and was common at the cerebral convexity or parasagittal/falx region. Histologically, it showed benign biological characteristics despite frequent and severe PTBE, and the extent of PTBE was related to MVD of tumors. Simpson I resection is the best treatment, and the prognosis is usually good after total tumor removal, while gamma knife is recommended for small residual tumor. [ABSTRACT FROM AUTHOR]

The real-time control of the non-holonomic wheel robot has put forward higher requirements for the accuracy and speed of dynamical simulation, so it is necessary to study the new dynamic modeling and calculation methods adapting to modern information processing technology. Different from the traditional method solving differential–algebraic equation, the objective is to establish optimization model and effective calculating scheme for dynamics of non-holonomic system based on basic dynamical principle. The optimization model cannot be obtained directly from the traditional Gauss' principle. By using Gauss' principle of variational form, this paper deduces the minimum principle in the form of generalized coordinates and quasi-coordinates, respectively, thus allowing dynamical problems of non-holonomic systems to be incorporated into the framework of solving constrained or unconstrained optimization problems. Furthermore, we study a numerical calculation scheme that uses an optimization algorithm for the second form of the above optimization models. As an example, the dynamical problem of a differential-driven wheeled mobile-robot system is discussed. The optimization dynamic model of a non-holonomic robot system and the calculation model of the optimization algorithm are established. Comparing the results of the optimization calculation with the differential–algebraic equations commonly used in dynamical problem for non-holonomic system reveals that the method in this paper is superior in terms of calculation speed and can more effectively handle constraint violations without extra constraint revision needed. [ABSTRACT FROM AUTHOR]

In smart cities and factories, robotic applications require high accuracy and security, which depends on precise inverse dynamics modeling. However, the physical modeling methods cannot include the nondeterministic factors of the manipulator, such as flexibility, joint clearance, and friction. In this paper, the Semiparametric Deep Learning (SDL) method is proposed to model robot inverse dynamics. SDL is a type of deep learning framework, designed for optimal inference, combining the Rigid Body Dynamics (RBD) model and Nonparametric Deep Learning (NDL) model. The SDL model takes advantage of the global characteristics of classic RBD and the powerful fitting capabilities of the deep learning approach. Moreover, the parametric and nonparametric parts of the SDL model can be optimized at the same time instead of being optimized separately. The proposed method is validated using experiments, performed on a UR5 robotic platform. The results show that the performance of SDL model is better than that of RBD model and NDL model. SDL can always provide relatively accurate joint torque prediction, even when the RBD or NDL model is not accurate. [ABSTRACT FROM AUTHOR]

Pig cloning technique is valuable in agriculture, biomedicine, and life sciences. However, the full-term developmental efficiency of cloned pig embryos is only about 1%, which limits pig cloning application. The quality of recipient oocytes greatly affects the developmental competence of cloned pig embryos. Thus, this study investigated the effects of a recipient oocyte source (in vivo matured [IVVM] oocytes vs. slaughter house-derived in vitro matured [IVTM] oocytes), and follicular liquid treatment (slaughter house-derived immature follicle-derived fluid [IFF] vs. in vivo-matured follicle-derived fluid [MFF]) during the in vitro maturation (IVM) of oocytes on the development of the cloned pig embryos. Our results showed that using IVVM oocytes to replace IVTM oocytes as recipient oocytes, and using 10% MFF IVM medium to replace 10% IFF IVM medium could enhance the development of the cloned pig embryos. IFF and MFF contained different levels of oocyte quality-related proteins, resulting in different oocyte quality-related gene expression levels and reactive oxygen species levels between the 10% MFF medium-cultured oocytes and 10% IFF medium-cultured oocytes. This study provided useful information for enhancing the pig cloning efficiency by improving the quality of recipient oocytes. [ABSTRACT FROM AUTHOR]

Summary Although the widely used anticoagulant drug heparin has been shown to have many other biological functions independent of its anticoagulant role, its effects on energy homeostasis are unknown. Here, we demonstrate that heparin level is negatively associated with nutritional states and that heparin treatment increases food intake and body weight gain. By using electrophysiological, pharmacological, molecular biological, and chemogenetic approaches, we provide evidence that heparin increases food intake by stimulating AgRP neurons and increasing AgRP release. Our results support a model whereby heparin competes with insulin for insulin receptor binding on AgRP neurons, and by doing so it inhibits FoxO1 activity to promote AgRP release and feeding. Heparin may be a potential drug target for food intake regulation and body weight control. [ABSTRACT FROM AUTHOR]