This paper proposes a novel smart parking scheme for the parking lot. Automatic car detection is the core technology of the proposed scheme. However, new challenges arise in car detection in aerial ...views, such as a large number of tiny objects and complex backgrounds. In order to solve these issues, this paper proposes a car detection method based on multi-task cost-sensitive-convolutional neural network (MTCS-CNN). In the proposed network framework, multi-task partition layer which is composed of some sub-task selection units is first developed. The sub-task selection unit is constructed by introducing local mask and non-zero pooling, which can divide the complex detection task into many simple sub-tasks. To tackle the obtained sub-tasks, cost-sensitive sub-network is proposed based on faster R-CNN framework with the introduction of cost-sensitive loss. In the proposed Multi-task partition layer, the sub-task selection unit is used to capture the local map of the original aerial view image. In each local map, the scale and the number of objects are enlarged and decreased, respectively. Therefore, multi-task partition layer can divide a complex tiny objects detection task into many simple enlarged objects detection sub-tasks, which is helpful for performance improvement. In addition, the proposed cost-sensitive loss can effectively discount the effect of easy examples and focus attention on the hard examples, which may improve the detection performance on hard examples. Therefore, the model with incorporation of proposed cost-sensitive loss is robust to the complex background, further improving the performance. On our dataset, the proposed method obtained an mAP of 85.3%, outperformed state-of-the-art method.
Recommendation system (RS) is designed to provide personalized services based on the users’ historical data. It has been applied in various fields and is expected to recommend the suitable services ...for the different kinds of users. Considering the importance of individual privacy, current users gradually tend not to expose personal information. This means RS may face the highly sparse datasets in the fields of cloud security. In general, the accuracy of recommendation will be improved with the growth of individual data, but the cold start problem is exactly in this contradictory phenomenon: this question evolves to produce sufficiently accurate recommendation result under the data scarcity problem. RS has to recommend services for the rarely historical data users and the latent users might drain along with the production of counter effects. To alleviate data scarcity problem in cloud security environment, this work is to introduce similar domain knowledge based on the transfer learning. Besides, the content and location based methods have been proved that these ideas work under this situation. So, this work also employs latent dirichlet allocation (LDA) to analysis the service descriptions and explore the relationship between the content and location information. In this framework, the suitable combination of LDA and word2vec models will balance the accuracy and speed which benefit service recommendation particularly. The related experiments demonstrate the effectiveness on the real word dataset. It can be found that the transfer learning based word2vec model shows the potentiality to explore the relationship between topic words, and improve the LDA algorithm from the content relationship. This proves that in both cold start environment and warm start environment, the proposed algorithm is more robust than other model-based state-of-art methods.
With the development of Internet of things (IoT), billions of sensors, actuators and other intelligent terminal devices (ITDs) are connected to the Internet. In this situation, traditional cloud ...computing models are not fully suitable because these ITDs generate too much data which may cause severe network congestion. As a result, edge computing is proposed to solve this problem, which can provide computation and storage services for ITDs with a distributed model on the edge networks. While, this novel edge paradigm brings emerged problems of restricted computation, limited storage and unstable network, as the storage is distributed typically as a key role. So, in this paper, it proposes a distributed multi-level storage (DMLS) model with a multiple-factors least frequently used (mLFU) algorithm to solve the problem. In this model, storage levels are composed of ITDs on the edge, so when the storage space of a node is not enough, the mLFU is used to remove a part of data from the current nodes and upload the date to the upper storage levels. To reduce the impact of data loss caused by unstable edge networks, a factor of importance is introduced in the mLFU, which means that data with high importance is uploaded to the upper levels first. Experiments show that the hit rate of the mLFU is stabilized at 74% that is almost equal to the typical LFU, while the important data loss rate is about 35% lower than the LFU and the random replacement algorithm.
Autonomous driving has received widespread attention in recent years, while the limited battery life and computing capability of autonomous vehicles cannot support some necessary ...computation-intensive and urgent tasks with strict response time requirements. The results of the tasks would be useless and may cause serious safety hazards if the given time constraints are exceeded. On the other side, mobile edge computing (MEC) offers the possibility of autonomous vehicles to complete these time-constraint tasks due to its proximity and strong computing capabilities, with the faster 5G wireless networks to enable vehicles and MEC servers to exchange data in milliseconds. Then, it is a key issue to make the MEC servers to execute and complete these time-constraint autonomous-driving tasks as many as possible. So, we propose a task scheduling algorithm that can consider characteristics of autonomous-driving tasks and select more suitable MEC servers with task migration, based on an improved earliest deadline first algorithm through the replacement and recombination of tasks. From the experimental results, it can be concluded that the algorithm can schedule more tasks and benefit the urgent tasks effectively with the increase of the task amounts.
Virtualization and cloud computing can help an organization achieve significant datacenter savings in hardware costs, operational expenditures, and energy demands while achieving improvements in ...quality of service and business agility. The combination of a hardware based root of trust such as
trust platform module
(TPM) on
virtual machine
(VM) based system have being widely adopted. In this paper, combined with the trusted computing and cloud computing security, we establish a trust system with a
certificate authority
(CA) and
trusted platform module
(TPM). It takes the CA as the root of trust cloud computing platform. The servers use the TPM through the operations of acquisition, registration, certification for the certification and the operations of new construction, launch, running, transfer and maintenance for the
virtual machine
(VM). To implement the trusted ensure of the security, it designs the trusted module which take the TPM as the core, and develops the VM as a complete trust system with a measurement algorithm. It can be used into the VM authentication mechanism and the access user authentication mechanism of the VM.
Cyber-Physical System (CPS) that integrates computational and physical capabilities has emerged as a promising topic. It interacts with physical world and humans through ad hoc communications. In ...contemporary, with the fast development of CPS theory and applications, CPS generates a large volume of data, which may lead its control into a chaotic status. Hence, there is a pressing demand to solve the chaotic status CPS control. To this end, the chaotic time series prediction algorithm is employed to resolve the chaotic status featured by a fuzzy feedback linearization model. Modeling the CPS under big data without taking chaotic features into account may lead to unexpected results. This is because chaotic CPS is dramatically sensitive to small disturbances or minor changes of initiators. This paper developed a CPS model in light of fuzzy feedback linearization. Further, the chaotic time prediction algorithm is applied to solve the chaotic control problem in CPS. The developed algorithm takes both tracking control problem and synchronization control problem into account. The numerical results suggest that the developed method is feasible and efficient in tracking control and synchronization of two chaotic CPS.
It has become a basic precursor and facilitator to analyze the emergence of big data with the rise of cloud computing and cloud storage by means of the novel standardized technologies. Then, binary ...relevance method is carried out as one of the widely known classifier chain methods for multi-label classification. It achieves a higher predictive performance, but it still retains a complex process and takes much computation time. So, in this paper, we present a enhanced classifier chain algorithm with
K
-means cluster method to confirm the order of the binary classifiers. It has a different strategy that several times of
K
-means algorithms are employed to get the correlations between labels and to confirm the order of binary classifiers. The algorithm ensures the precise correlations to be transmitted persistently to improve the earlier predictions accuracy. The experiments on a sample data sets of Reuters-21578 show that the approach is effective and appealing in the common cases, it is accurate for a preliminary classification to provide a basis for the further refined classifications.
WD repeat domain 5 (WDR5) is a prominent target for pharmacological inhibition in cancer through its scaffolding role with various oncogenic partners such as MLL and MYC. WDR5-related drug discovery ...efforts center on blocking these binding interfaces or degradation have been devoted to developing small-molecule inhibitors or degraders of WDR5 for cancer treatment. Nevertheless, the precise role of WDR5 in these cancer cells has not been well elucidated genetically. Here, by using an MLL-AF9 murine leukemia model, we found that genetically deletion of Wdr5 impairs cell growth and colony forming ability of MLL-AF9 leukemia cells in vitro or ex vivo and attenuates the leukemogenesis in vivo as well, which acts through direct regulation of ribosomal genes. Pharmacological inhibition of Wdr5 recapitulates genetic study results in the same model. In conclusion, our current study demonstrated the first genetic evidence for the indispensable role of Wdr5 in MLL-r leukemogenesis in vivo, which supports therapeutically targeting WDR5 in MLL-rearranged leukemia by strengthening its disease linkage genetically and deepening insights into its mechanism of action.
Display omitted
•Wdr5 conditional knockout mice were generated utilizing CRISPR/Cas9 KI approach.•Genetically loss of Wdr5 impairs MLL-r leukemogenesis in vitro and in vivo.•Wdr5 regulates MLL-r leukemia via cell-cycle, apoptosis, and cell differentiation.•Wdr5 functions through direct regulation of ribosomal genes.•Wdr5 inhibitor recapitulates the effect by loss of Wdr5 in MLL-r leukemia.