Video person re-identification is receiving academic interest. However, the practical application of the algorithm is hardly supported because of prohibitive annotated data. Hence, the study for unlabeled data will le...
Video person re-identification is receiving academic interest. However, the practical application of the algorithm is hardly supported because of prohibitive annotated data. Hence, the study for unlabeled data will lead to an attractive alternative. This work explores an innovative strategy, namely, learning to cluster unlabeled person in the video through graph convolutional networks. In this paper, we find that the possibility of inter-frame linkage can be inferred from context. Therefore, a pose-guided topology linkage clustering framework is proposed. Our framework consists of three modules: (i) a pose-guided representation module; (ii) a pose-guided embedding module; (iii) a link prediction module. Firstly, the representation coding alone is performed at the level of relational induction bias, embedding the implicit pose structure information in image features. Then, based on the consideration of the topology relationship between adjacent and cross-frame, graph convolutional network is introduced to infer the likelihood of linkage between frame nodes. Experiments show that the method demonstrates excellent scalability in addition to being an effective response to person clustering in case of changes, and does not need the number of clusters as a prior.
Detecting community structure based on node similarity cost lower time complexity, but they ignore the indirect relationships of nodes. We proposed an improved algorithm for detecting community structure based on node...
详细信息
Synchronization decay of resting state EEG has shown that cognitive dysfunction in mild cognitive impairment (MCI) was relevant to a loss of functional connectivity in intermediate frequency bands. A new estimator ana...
详细信息
Synchronization decay of resting state EEG has shown that cognitive dysfunction in mild cognitive impairment (MCI) was relevant to a loss of functional connectivity in intermediate frequency bands. A new estimator analysis method called new S estimator (NSE) proposed recently by us quantifies synchronization between neuronal signals at multiple *** paper meant to explore the behavior of synchronization of multichannel EEG in MCI and healthy normal controls at rest, and preliminarily make clear the clinical significance of the NSE in MCI *** (EEG) were recorded from 10 MCI patients and 12 age-matched healthy normal controls (NC). NSE values were computed both across the all frequency band and separately in the delta, theta, alpha, beta (including beta1, beta2 and beta3), and gamma bands. The Montreal Cognitive Assessment (MoCA) was used to assess the symptom severity of MCI patients and *** values in the alpha and beta1 bands were significantly lower in MCI patients than in NC. NSE values in the alpha and beta1 bands were positively correlated with the MoCA scores in all participants (MCI and NC). In MCI patients, NSE values in the alpha and beta1 bands were also positively correlated with MoCA *** results suggest that NSE values are a useful correlate of EEG synchronization in MCI patients.
Most of algorithms based on bit vector for mining frequent pattern always produce candidate itemsets and scan the same data repeatedly;this increases the running time and the space consumption. In this paper, an algor...
详细信息
Most of algorithms based on bit vector for mining frequent pattern always produce candidate itemsets and scan the same data repeatedly;this increases the running time and the space consumption. In this paper, an algorithm BVDAHL based on bit vector decomposition and hash linklist for mining frequent patterns on data streams which solves the above issues, is proposed. The itemsets (transactions) whose corresponding number of 1 is k, denoted as k-one itemsets (k-one transactions). The arrival transactions are converted into bit vectors, and permutations and combinations are used to decompose converted results, then the decomposed itemsets are stored in the hash linklist. In hash function, according to the position of 1 of bit vector to compute the address, if the transaction has same address and same content that already exists in data domain, then the value adds one in count domain, otherwise the pointer in pointer domain points to the next node directly, it is no need to deal with conflict. Afterwards the property of anti-monotonic will be used to prune the k+1-one transactions in bit vector table if there exists infrequent k-one itemsets that decomposed by such k+1-one transactions. Then the algorithm repeats the above processes until the transaction in bit vector table can not be decomposed. Finally we will receive the frequent itemsets by comparing the obtained value of count domain in hash linklist with minSup. Experiment results show that BVDAHL is very efficiency and scalable.
It is significant for measuring the importance of nodes accurately to improve software stability and robustness in software network. A software execution directed network takes function as a node and relationship of f...
详细信息
A novel method based on multilevel meshes is proposed to simulate the cloth draping process rapidly on a complex model. Considering of the difficulty of computing the realtime collision between cloth and complex model...
详细信息
With the widespread use of touch-screen devices, it is more and more convenient for people to draw sketches on screen. This results in the demand for automatically understanding the sketches. Thus, the sketch recognit...
详细信息
With the popularity of the service-oriented development paradigm in the cloud era, numerous cloud APIs have been developed. The continuous proliferation of cloud APIs has submerged developers in a sea of options, brin...
详细信息
ISBN:
(数字)9798350349184
ISBN:
(纸本)9798350349191
With the popularity of the service-oriented development paradigm in the cloud era, numerous cloud APIs have been developed. The continuous proliferation of cloud APIs has submerged developers in a sea of options, bringing the need for cloud API recommendation techniques. Most existing cloud API recommendation methods focus on theoretical studies and neglect practical applications. As a result, these methods face several challenges, such as covering recommendation scenarios, generating real-time responses, and handling concurrent requests. In this paper, we propose and implement a practical cloud API complementary recommendation service for mashup creation. To address the above challenges, we first propose a cloud API complementary recommendation framework named FICR, which exploits feature interactions to mine implicit complementarity and models complementarity from the mashup side and the target cloud API side for scenario coverage. Then, we conduct performance optimization via precoding and batch size tuning to optimize FICR for real-time responses. Finally, we design a minimum-connection load balancing algorithm with resource early warning and build a distributed system for concurrent requests. Experiments conducted on real-world datasets show that FICR outperforms state-of-the-art baseline models in complementary recommendation scenarios, and demonstrate that the implemented service meets the real-time and concurrency requirements. The service will serve as a practical tool to help developers during mashup creation.
In order to solve the overload problem of root ONS in the EPC network, a load balancing algorithm based on multi-root ONS is proposed. Based on the proposed load balancing ONS (LB ONS) architecture, the ONS Root is de...
详细信息
The existing model simplification algorithm in simplified speed and quality can't reach a better compromise, so we present an improved quadric error metrics edge collapse mesh simplification algorithm. This algori...
详细信息
暂无评论