Emerging technologies and applications including Internet of Things (IoT), social networking, and crowd-sourcing generate large amounts of data at the network edge. Machine learning models are often built from the collected data, to enable the detection, classification, and prediction of future events. Due to bandwidth, storage, and privacy concerns, it is often impractical to send all the data to a centralized location. In this paper, we consider the problem of learning model parameters from data distributed across multiple edge nodes, without sending raw data to a centralized place. Our focus is on a generic class of machine learning models that are trained using gradientdescent based approaches. We analyze the convergence bound of distributed gradient descent from a theoretical point of view, based on which we propose a control algorithm that determines the best trade-off between local update and global parameter aggregation to minimize the loss function under a given resource budget. The performance of the proposed algorithm is evaluated via extensive experiments with real datasets, both on a networked prototype system and in a larger-scale simulated environment. The experimentation results show that our proposed approach performs near to the optimum with various machine learning models and different data distributions.
Emerging technologies and applications including Internet of Things (IoT), social networking, and crowd-sourcing generate large amounts of data at the network edge. Machine learning models are often built from the collected data, to enable the detection, classification, and prediction of future events. Due to bandwidth, storage, and privacy concerns, it is often impractical to send all the data to a centralized location. In this paper, we consider the problem of learning model parameters from data distributed across multiple edge nodes, without sending raw data to a centralized place. Our focus is on a generic class of machine learning models that are trained using gradientdescent based approaches. We analyze the convergence rate of distributed gradient descent from a theoretical point of view, based on which we propose a control algorithm that determines the best trade-off between local update and global parameter aggregation to minimize the loss function under a given resource budget. The performance of the proposed algorithm is evaluated via extensive experiments with real datasets, both on a networked prototype system and in a larger-scale simulated environment. The experimentation results show that our proposed approach performs near to the optimum with various machine learning models and different data distributions.
Abstract-p38 Mitogen-activated protein kinase (MAPK) is one of the most ancient signaling molecules and is involved in multiple cellular processes, including cell proliferation, cell growth, and cell death. In the heart, enhanced activation of p38 MAPK is associated with ischemia/reperfusion injury and the onset of heart failure. In the present study, we investigated the function of p38 MAPK in regulating cardiac contractility and its underlying mechanisms. In cultured adult rat cardiomyocytes, activation of p38 MAPK by adenoviral gene transfer of an activated mutant of its upstream kinase, MKK3bE, led to a significant reduction in baseline contractility, compared with uninfected cells or those infected with a control adenoviral vector (Adv--galactosidase). The inhibitory effect of MKK3bE on contractility was largely prevented by coexpressing a dominant-negative mutant of p38 MAPK or treating cells with a p38 MAPK inhibitor, SB203580. Conversely, inhibition of endogenous p38 MAPK activity by SB203580 rapidly and reversibly enhanced cell contractility in a dose-dependent manner, without altering L-type Ca 2ϩ currents or Ca 2ϩ i transients. MKK3bE-induced p38 activation had no significant effect on pH i , whereas SB203580 had a minor effect to elevate pH i . Furthermore, activation of p38 MAPK was unable to increase troponin I phosphorylation. Thus, we conclude that the negative inotropic effect of p38 MAPK is mediated by decreasing myofilament response to Ca 2ϩ , rather than by altering Ca Key Words: p38 mitogen-activated protein kinase Ⅲ cardiac contractility Ⅲ excitation-contraction coupling Ⅲ troponin I Ⅲ intracellular pH M itogen-activated protein kinase (MAPK) superfamily is one of the most important signal transduction systems conserved in all eukaryotes. 1-4 There are three major subgroups identified, including the extracellular signal-regulated kinase (ERK1/2), p38 MAPK, and c-jun-NH 2 -terminal kinase (JNK). p38 MAPK is a subfamily of stress-activated MAPKs (SAPKs) known to be involved in a multitude of cellular events, such as inflammation, cell injury, cell proliferation or differentiation, and cell growth or death. [5][6][7] In the heart, activation of p38 MAPK has been observed in pressure overload or ischemia/infarction-induced cardiac hypertrophy and heart failure in humans 8,12,13 and animal models. 14 -19 In cultured cardiac myocytes, activation of p38 MAPK induces myocyte hypertrophy and apoptosis 18,20 and is also implicated in the preconditioning process and ischemia/reperfusion injury. [21][22][23][24][25] Our recent studies have shown that -adrenergic stimulation is able to activate p38 MAPK via a protein kinase A (PKA)-dependent mechanism and that activation of p38 MAPK provides a negative feedback to PKA-mediated positive contractile response in intact cardiac myocytes. 26 Furthermore, evidence from our recent in vivo studies in transgenic mice has demonstrated that cardiacspecific activation of p38 MAPK markedly attenuates cardiac contractility. 27 However, the mechanism underlying the...
Mobile edge clouds (MECs) bring the benefits of the cloud closer to the user, by installing small cloud infrastructures at the network edge. This enables a new breed of real-time applications, such as instantaneous object recognition and safety assistance in intelligent transportation systems, that require very low latency. One key issue that comes with proximity is how to ensure that users always receive good performance as they move across different locations. Migrating services between MECs is seen as the means to achieve this. This article presents a layered framework for migrating active service applications that are encapsulated either in virtual machines (VMs) or containers. This layering approach allows a substantial reduction in service downtime. The framework is easy to implement using readily available technologies, and one of its key advantages is that it supports containers, which is a promising emerging technology that offers tangible benefits over VMs. The migration performance of various real applications is evaluated by experiments under the presented framework. Insights drawn from the experimentation results are discussed.
A tenet of beta1-adrenergic receptor (beta1AR) signaling is that stimulation of the receptor activates the adenylate cyclase-cAMP-protein kinase A (PKA) pathway, resulting in positive inotropic and relaxant effects in the heart. However, recent studies have suggested the involvement of Ca2+/calmodulin-dependent protein kinase II (CaMKII) in beta1AR-stimulated cardiac apoptosis. In this study, we determined roles of CaMKII and PKA in sustained versus short-term beta1AR modulation of excitation-contraction (E-C) coupling in cardiac myocytes. Short-term (10-minute) and sustained (24-hour) beta1AR stimulation with norepinephrine similarly enhanced cell contraction and Ca2+ transients, in contrast to anticipated receptor desensitization. More importantly, the sustained responses were largely PKA-independent, and were sensitive to specific CaMKII inhibitors or adenoviral expression of a dominant-negative CaMKII mutant. Biochemical assays revealed that a progressive and persistent CaMKII activation was associated with a rapid desensitization of the cAMP/PKA signaling. Concomitantly, phosphorylation of phospholamban, an SR Ca2+ cycling regulatory protein, was shifted from its PKA site (16Ser) to CaMKII site (17Thr). Thus, beta1AR stimulation activates dual signaling pathways mediated by cAMP/PKA and CaMKII, the former undergoing desensitization and the latter exhibiting sensitization. This finding may bear important etiological and therapeutical ramifications in understanding beta1AR signaling in chronic heart failure.
FAM3A belongs to a novel cytokine-like gene family, and its physiological role remains largely unknown. In our study, we found a marked reduction of FAM3A expression in the livers of db/db and high-fat diet (HFD)-induced diabetic mice. Hepatic overexpression of FAM3A markedly attenuated hyperglycemia, insulin resistance, and fatty liver with increased Akt (pAkt) signaling and repressed gluconeogenesis and lipogenesis in the livers of those mice. In contrast, small interfering RNA (siRNA)-mediated knockdown of hepatic FAM3A resulted in hyperglycemia with reduced pAkt levels and increased gluconeogenesis and lipogenesis in the livers of C57BL/6 mice. In vitro study revealed that FAM3A was mainly localized in the mitochondria, where it increases adenosine triphosphate (ATP) production and secretion in cultured hepatocytes. FAM3A activated Akt through the p110a catalytic subunit of PI3K in an insulin-independent manner. Blockade of P2 ATP receptors or downstream phospholipase C (PLC) and IP3R and removal of medium calcium all significantly reduced FAM3A-induced increase in cytosolic free Ca 21 levels and attenuated FAM3A-mediated PI3K/Akt activation. Moreover, FAM3A-induced Akt activation was completely abolished by the inhibition of calmodulin (CaM). Conclusion: FAM3A plays crucial roles in the regulation of glucose and lipid metabolism in the liver, where it activates the PI3K-Akt signaling pathway by way of a Ca 21 /CaM-dependent mechanism. Up-regulating hepatic FAM3A expression may represent an attractive means for the treatment of insulin resistance, type 2 diabetes, and nonalcoholic fatty liver disease (NAFLD). (HEPATOLOGY 2014;59:1779-1790 T ype 2 diabetes has become one of the most prevalent and debilitating chronic diseases, with a global prevalence 6.4%, affecting about 285 million adults in the year 2010.1 Hepatic insulin resistance and fatty liver play a crucial role in the development and progression of type 2 diabetes. Liver is the key tissue regulating release of glucose into circulation during the fasting state, and hepatic insulin resistance is a decisive factor causing fasting hyperglycemia and type 2 diabetes. The liver is also one of the major organs regulating triglyceride (TG) and cholesterol (CHO) metabolism.2 Hepatic insulin resistance is mainly described as the failure of insulin to repress the expression of gluconeogenic genes through the PI3K/ Akt signaling pathway and is closely associated with the dysregulation of glucose and lipid metabolism in the liver.2 Although the underlying mechanisms remain largely unknown, increasing evidence points to
We study the dynamic service migration problem in mobile edge-clouds that host cloud-based services at the network edge. This offers the benefits of reduction in network overhead and latency but requires service migrations as user locations change over time. It is challenging to make these decisions in an optimal manner because of the uncertainty in node mobility as well as possible non-linearity of the migration and transmission costs. In this paper, we formulate a sequential decision making problem for service migration using the framework of Markov Decision Process (MDP). Our formulation captures general cost models and provides a mathematical framework to design optimal service migration policies. In order to overcome the complexity associated with computing the optimal policy, we approximate the underlying state space by the distance between the user and service locations. We show that the resulting MDP is exact for uniform one-dimensional mobility while it provides a close approximation for uniform two-dimensional mobility with a constant additive error term. We also propose a new algorithm and a numerical technique for computing the optimal solution which is significantly faster in computation than traditional methods based on value or policy iteration. We illustrate the effectiveness of our approach by simulation using real-world mobility traces of taxis in San Francisco.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
334 Leonard St
Brooklyn, NY 11211
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.