Search

Abstract With the continuous development and combined application of cloud computing and artificial intelligence, some new methods have emerged to reduce task execution time for training neural network models in a cloud-edge collaborative environment. The most attractive method is neural network model segmentation. However, many factors affect the segmentation point, such as resource allocation, system energy consumption, load balancing, and network Bandwidth allocation. Some segmentation methods consider the shortest task execution time, which ignores the utilization of resources at the edge and can result in resource waste. Additionally, these factors are difficult to measure, which presents a challenge in calculating the best segmentation point to achieve the goal of maximum resource utilization and minimum task execution time. To solve this problem, this paper proposes a cloud-edge collaborative task scheduling method based on model segmentation (CECMS). This method first analyzes the factors affecting the segmentation point of the model and then obtains accurate factors that affect the segmentation point calculation through the pre-execution method. Furthermore, a multi-objective solution algorithm is improved to calculate the optimal model segmentation point. And tasks are separately offloaded to the edge and cloud based on the optimal model segmentation point. Finally, the experiments are conducted to verify the effectiveness of this method. Finally, the effectiveness of the CECMS method was verified through simulation experiments. Compared with the Dynamic Adaptive DNN Surgery (DADS) method and an adaptive DNN inference acceleration framework algorithm with end–edge–cloud collaborative computing algorithm (ADC), CECMS achieves the same effectiveness as DADS and ADC in optimizing task execution time by comprehensively considering the utilization of edge resources and minimizing task execution time, while also effectively ensuring resource utilization.

The emergence of advanced biotechnologies has intensified in recent years. The rapid development of these biotechnologies has had a wide and profound impact globally, with the majority on the frontier of biosecurity technologies. The global biosecurity situation is currently highly challenging, with the characteristics of internationalization, extremely serious harm, and complex development. The misuse and abuse of biotechnologies are common, which thereby endanger biosecurity. The international community and governments have attached great importance to cutting-edge biotechnologies and are implementing laws and regulations to control and prevent biosecurity-related influences. By tracking the frontier progress of the new technologies generated from gene editing, gene drives, synthetic biology, and related cross-disciplines as applied in the biosecurity field, we analyzed the development trends in biotechnologies and their potential influence. On the one hand, this paper proposes that there is an urgent need for global cooperation to formulate management, control strategies, and measures for cutting-edge technologies in biosecurity to jointly promote their sound development. On the other hand, the international community should jointly shoulder the responsibility for safeguarding biosecurity and restrict the biotechnologies to rational applications within legal and ethical frameworks. This paper aims to provide a reference for the prevention of misuse and abuse of biotechnologies and support global cooperation in biosecurity governance.

Foundation models (e.g., ChatGPT, DALL-E, PengCheng Mind, PanGu-$\Sigma$) have demonstrated extraordinary performance in key technological areas, such as natural language processing and visual recognition, and have become the mainstream trend of artificial general intelligence. This has led more and more major technology giants to dedicate significant human and financial resources to actively develop their foundation model systems, which drives continuous growth of these models' parameters. As a result, the training and serving of these models have posed significant challenges, including substantial computing power, memory consumption, bandwidth demands, etc. Therefore, employing efficient training and serving strategies becomes particularly crucial. Many researchers have actively explored and proposed effective methods. So, a comprehensive survey of them is essential for system developers and researchers. This paper extensively explores the methods employed in training and serving foundation models from various perspectives. It provides a detailed categorization of these state-of-the-art methods, including finer aspects such as network, computing, and storage. Additionally, the paper summarizes the challenges and presents a perspective on the future development direction of foundation model systems. Through comprehensive discussion and analysis, it hopes to provide a solid theoretical basis and practical guidance for future research and applications, promoting continuous innovation and development in foundation model systems.

Abstract Algorithms and frameworks for distributed machine learning have been widely used in numerous artificial intelligence engineering applications. A cloud platform provides a large number of resources at a lower cost and is a more convenient method for such applications. With the rapid development of containerization, native cloud combinations based on Docker and Kubernetes have provided effective resource support for distributed machine learning. However, native Kubernetes does not provide efficient priority or fair resource scheduling strategies for distributed machine learning in computationally intensive and time-consuming jobs, which easily leads to resource deadlock, resource waste, and low job execution efficiency. Therefore, to utilize the execution order between multiple jobs in distributed machine learning as well as the dependencies between multiple tasks for the same job, considering intra- and inter-group scheduling priorities, a combined priority scheduling method is proposed for distributed machine learning based on Kubernetes and Volcano. Considering the user priority, task priority, longest wait time, task parallelism, and affinity and non-affinity between the parameter server and worker nodes, a combined priority scheduling model of inter- and intra-job priority is proposed, which is mapped into a scheduling strategy of inter- and intra-group priorities of pods, enabling the efficient scheduling and training of distributed machine learning. The experiment results show that the proposed method achieves preferential resource allocation for urgent, high parallelism, and high-priority jobs with high-priority users and improves the job execution efficiency. The affinity and anti-affinity settings among pods reduce the time of information interaction between the parameter server and worker nodes to a certain extent, thereby improving the job completion efficiency. This group scheduling strategy alleviates the problems of resource deadlock and waste caused by insufficient resources in cloud computing.

Biosafety issues have become a major threat to the health of humans, animals, and ecosystems worldwide. As problems with food production and the food supply chain have become of greater concern to consumers, issues involving biosafety and food safety from a developmental perspective need to be urgently addressed. The term, food biosafety, is the combination of the core concepts of biosafety and food safety. It refers to the effective prevention of biological threats to food production and the food supply chain by controlling foodborne diseases arising from the consumption of edible plants and animal products, by preventing the establishment of invasive species, by strictly controlling the use of antibiotics, agricultural chemicals and veterinary drugs in the food supply chain, and by initiating food defense and anti-terrorism measures to protect the health of humans, animals, and ecosystems, thereby maintaining sustainable development in China. This article provides theoretical support for the extension of food biosafety to propose an innovative plan for the international co-governance of food safety.

An unmanned aerial vehicle (UAV) swarm is a fast-moving system where self-adaption is necessary when conducting a mission. The major causative factors of mission failures are inevitable disruptive events and uncertain threats. Given the unexpected disturbances of events and threats, it is important to study how a UAV swarm responds and enable the swarm to enhance resilience and alleviate negative influences. Cooperative adaptation must be established between the swarm’s structure and dynamics, such as communication links and UAV states. Thus, based on previous structural adaptation and dynamic adaptation models, we provide a co-adaptation model for UAV swarms that combines a swarm’s structural characteristics with its dynamic characteristics. The improved model can deal with malicious events and contribute to a rebound in the swarm’s performance. Based on the proposed co-adaptation model, an improved resilience metric revealing the discrepancy between the minimum performance and the standard performance is proposed. The results from our simulation experiments show that the surveillance performance of a UAV swarm bounces back to its initial state after disruptions happen in co-adaptation cases. This metric demonstrates that our model can contribute towards the swarm’s overall systemic resiliency by withstanding and resisting unpredictable threats and disruptions. The model and metric proposed in this article can help identify best practices in improving swarm resilience.

Highly robust networks can resist attacks, as when some UAVs fail, the remaining UAVs can still transmit data to each other. In order to improve the robustness of a multi-UAV network, most methods construct the network by adjusting the positions of the UAVs and adding a large number of links. However, having a large number of links greatly consumes communication resources and increases serious signal interference. Therefore, it is necessary to study a method that can improve robustness and reduce the number of links. In this paper, we propose a method that consists of combining formation control and link selection, which can work in a distributed manner. For formation control, our method keeps the UAVs compact in the obstacle environment through an improved artificial potential field. The compact formation enables UAVs to have a large number of neighbors. For link selection, reinforcement learning is used to improve the robustness of the network and reduce the number of network edges. In the simulation of the 3D urban environment, three failure modes are used to verify the robustness of the network. The experimental results show that even if the number of links is reduced by 20%, the networks designed by our method still have strong robustness.

Renal fibrosis is considered as the final pathway of all types of kidney diseases, which can lead to the progressive loss of kidney functions and eventually renal failure. The mechanisms behind are diversified, in which the mammalian target of rapamycin (mTOR) pathway is one of the most important regulatory pathways that accounts for the disease. Several processes that are regulated by the mTOR pathway, such as autophagy, epithelial-mesenchymal transition (EMT), and endoplasmic reticulum (ER) stress, are tightly associated with renal fibrosis. In this study, we have reported that the expression of tripartite motif-containing (TRIM) protein 6, a member of TRIM family protein, was highly expressed in renal fibrosis patients and positively correlated with the severity of renal fibrosis. In our established in vitro and in vivo renal fibrosis models, its expression was upregulated by the Angiotensin II-induced nuclear translocation of nuclear factor-κB (NF-κB) p50 and p65. In HK2 cells, the expression of TRIM6 promoted the ubiquitination of tuberous sclerosis proteins (TSC) 1 and 2, two negative regulators of the mTORC1 pathway. Moreover, the knockdown of TRIM6 was found efficient for alleviating renal fibrosis and inhibiting the downstream processes of EMT and ER in both HK2 cells and 5/6-nephrectomized rats. Clinically, the level of TRIM6, TSC1/2, and NF-κB p50 was found closely related to renal fibrosis. As a result, we have presented the first study on the role of TRIM6 in the mTORC1 pathway in renal fibrosis models and our findings suggested that TRIM6 may be a potential target for the treatment of renal fibrosis.

Glomerular hypertrophy is an early morphological alteration in diabetic nephropathy. Cyclin-Dependent Kinases have been shown to be required for high glucose (HG)-induced hypertrophy; however, the upstream regulators of CDKN1B in glomerular hypertrophy remain unclear. Herein we describe a novel pathway in which Long noncoding RNA (lncRNA) NEAT1 regulates the progression of mesangial cell hypertrophy via a competing endogenous RNA (ceRNA) mechanism. Real-time PCR was performed to detect the relative NEAT1 and miR-222-3p expressions and further confirmed the relationship between NEAT1 and miR-222-3p. Cell cycle was evaluated by flow cytometry. The related mechanisms were explored by Western blot, RNA immunoprecipitation and chromatin immunoprecipitation assay. We show that NEAT1 forms double stranded RNA (dsRNA) with miR-222-3p, thus limiting miR-222-3p’s binding with CDKN1B. This release of CDKN1B mRNA leads to elevated CDKN1B protein expression, resulting in hypertrophy. In addition, we demonstrated that STAT3 which is activated by HG induces the transcription of NEAT1 by binding to its promoter. Our findings underscore an unexpected role of lncRNAs on gene regulation and introduce a new mode of proliferation regulation in mesangial cells.

The acute inflammatory lung injury is an important cause of death due to influenza A virus (IAV) infection. Insulin-like growth factor 1 (IGF1) played an important role in the regulation of inflammation in the immune system. To investigate the role of IGF1 in IAV-mediated acute inflammatory lung injury, the expression of IGF1 and inflammatory cytokines was tested after IAV A/Puerto Rico/8/1934 (H1N1; abbreviated as PR8) infection in A549 cells. Then, a BALB/c mouse model of PR8 infection was established. On days 3, 5, 7, and 9 post-infection, the mice lung tissue was collected to detect the expression changes in IGF1 mRNA and protein. The mice were divided into four groups: (1) PBS (abbreviation of phosphate buffered saline); (2) PR8 + PBS; (3) PR8 + IGF1; and (4) PR8 + PPP (abbreviation of picropodophyllin, the IGF1 receptor inhibitor). The body weight and survival rate of the mice were monitored daily, and the clinical symptoms of the mice were recorded. On day 5 post-infection, the mice were sacrificed to obtain the serum and lung tissues. The expression of inflammatory cytokines in the serum was detected by enzyme linked immunosorbent assay; lung injury was observed by hematoxylin-eosin staining; the viral proliferation in the lung was detected by real-time quantitative PCR; and the protein expression of the main molecules in the phosphatidylinositol-3-kinases/protein kinase B (PI3K/AKT) and mitogen-activated protein kinase (MAPK) signaling pathways was detected by Western blot. It was found that IGF1 expression is upregulated in A549 cells and BALB/c mice infected with PR8, whereas IGF1 regulated the expression of inflammatory cytokines induced by PR8 infection. Overexpression of IGF1 aggravated the IAV-mediated inflammatory response, whereas the inhibition of IGF1 receptor reduced such inflammatory response. The phosphorylation of IGF1 receptor triggered the PI3K/AKT and MAPK signaling pathways to induce an inflammatory response after IAV infection. Therefore, IGF1 plays an important immune function in IAV-mediated acute inflammatory lung injury. IGF1 may provide a therapeutic target for humans in response to an influenza outbreak, and inhibition of IGF1 or IGF1 receptor may represent a novel approach to influenza treatment.

Objectives: The prevalence and dissemination of diverse NDM-producing bacteria in China was investigated. Methods: We collected 1,162 isolates from 8 cities during December 2013 ∼ May 2015 in China. The NDM-positive strains as well as the NDM genotypes in these sample were detected via Vitek 2 compact system (bioMérieux, France), 16S rRNA gene sequencing, PCR and an S1- pulsed-field gel electrophoresis assay and Southern blot hybridization. The horizontal-transfer capability of the blaNDM gene was assessed by filter mating by using a standard E.coli J53 azide-resistant strain as the recipient. Results: Three genotypes (NDM-1, NDM-3 and NDM-5) of NDM-producing bacteria were identified, among which the NDM-1-positive isolates were the most frequent one. For the first time, we found NDM-5-produing S.typhimurium and NDM-3-produing E.coli in China. We also found that the NDM-positive (especially NDM-3 and NDM-5) strains were completely resistant to nearly all of the antimicrobial drugs utilized and blaNDM was mostly located on diverse plasmids with sizes ranging from 30 to 670 kb. Conclusion: Various species of bacteria especially the enteric pathogens with diverse NDM genotypes had spread in China. Hence, an ongoing surveillance of their dissemination is essential to prevent and control the spread of these organisms.

Multidrug resistant microbes present in the environment are a potential public health risk. In this study, we investigate the presence of New Delhi metallo-β-lactamase 1 (NDM-1) producing bacteria in the 99 water samples in Beijing City, including river water, treated drinking water, raw water samples from the pools and sewage from 4 comprehensive hospitals. For the bla NDM-1 positive isolate, antimicrobial susceptibility testing was further analyzed, and Pulsed Field Gel Electrophoresis (PFGE) was performed to determine the genetic relationship among the NDM-1 producing isolates from sewage and human, as well as the clinical strains without NDM-1. The results indicate that there was a higher isolation of NDM-1 producing Acinetobacter baumannii from the sewage of the hospitals, while no NDM-1 producing isolates were recovered from samples obtained from the river, drinking, or fishpond water. Surprisingly, these isolates were markedly different from the clinical isolates in drug resistance and pulsed field gel electrophoresis profiles, suggesting different evolutionary relationships. Our results showed that the hospital sewage may be one of the diffusion reservoirs of NDM-1 producing bacteria.

MicroRNAs have been implicated in the regulation of gene expression of various biological processes in a post-transcriptional manner under physiological and pathological conditions including host responses to viral infections. The 2009 pandemic H1N1 influenza virus is an emerging reassortant strain of swine, human and bird influenza virus that can cause mild to severe illness and even death. To further understand the molecular pathogenesis of the 2009 pandemic H1N1 influenza virus, we profiled cellular microRNAs of lungs from BALB/c mice infected with wild-type 2009 pandemic influenza virus A/Beijing/501/2009 (H1N1) (hereafter referred to as BJ501) and mouse-adapted influenza virus A/Puerto Rico/8/1934 (H1N1) (hereafter referred to as PR8) for comparison. Microarray analysis showed both the influenza virus BJ501 and PR8 infection induced strain- and temporal-specific microRNA expression patterns and that their infection caused a group of common and distinct differentially expressed microRNAs. Characteristically, more differentially expressed microRNAs were aroused on day 5 post infection than on day 2 and more up-regulated differentially expressed microRNAs were provoked than the down-regulated for both strains of influenza virus. Finally, 47 differentially expressed microRNAs were obtained for the infection of both strains of H1N1 influenza virus with 29 for influenza virus BJ501 and 43 for PR8. Among them, 15 microRNAs had no reported function, while 32 including miR-155 and miR-233 are known to play important roles in cancer, immunity and antiviral activity. Pathway enrichment analyses of the predicted targets revealed that the transforming growth factor-β (TGF-β) signaling pathway was the key cellular pathway associated with the differentially expressed miRNAs during influenza virus PR8 or BJ501 infection. To our knowledge, this is the first report of microRNA expression profiles of the 2009 pandemic H1N1 influenza virus in a mouse model, and our findings might offer novel therapy targets for influenza virus infection.

Unmanned Aerial Vehicle (UAV) swarms are utilized in various missions and operated within an open environment that is prone to disruptions. The resilience of UAV swarms, an important requirement, mainly revolves around ensuring stable and uninterrupted operations. Malicious attacks can implement the adverse impacts of potential threats through swarm communication links. In this context, the SIS (Susceptible → Infected → Susceptible) method is suitable for describing the information transmission within UAV swarms. An enhanced resilience model of the UAV swarm is proposed in this study, which incorporates the factors of self-dynamics, dynamics of topology, dynamics of information transmission, and SIS into the complex network model. Self-dynamics refer to the internal dynamics of the drones. In this paper, dynamics of topology consist of three factors: the varying distance between drones, the incoming degrees of each drone, and the number of communication types between drones. Lastly, dynamics of information transmission are characterized by SIS. The model proposed in this paper has the capability to effectively capture changes in the network topology as well as the dynamics of the system, which are significant contributors to the loss of resilience. And then, the average number of susceptible drones is utilized as the metric to evaluate the resilience of the swarm. Furthermore, an experiment is conducted where a UAV swarm successfully carries out a surveillance mission to demonstrate the advantages of our proposed method. The proposed model not only enables the support of mission planning but also facilitates the design enhancements of UAV swarms. [ABSTRACT FROM AUTHOR]