As the number of modules in system-on-Chip increases,the topology is more likely to suffer from excessive end-to-end hopcounts,causing an increase of power consumption and area *** this,a novel hierarchical topology c...
详细信息
As the number of modules in system-on-Chip increases,the topology is more likely to suffer from excessive end-to-end hopcounts,causing an increase of power consumption and area *** this,a novel hierarchical topology called CHMesh is proposed,which is based on mesh and ***,this article elaborates on a shortest deadlock-free CHXY routing algorithm which has a low *** results demonstrate that,compared with traditional mesh andref-mesh,the CHMesh can increase the average throughput by about 60% and 10% respectively under the nonuniform traffic patterns,presenting more advantages on large-scale NoC.
The process of feature point detection is the first stage of many computer vision applications including object detection, recognition andreconstruction. Aimed at the drawbacks(e.g. inefficiency and poorrotation inv...
详细信息
The process of feature point detection is the first stage of many computer vision applications including object detection, recognition andreconstruction. Aimed at the drawbacks(e.g. inefficiency and poorrotation invariance) of some classical detectors for volumetric data, we present a new scale-invariant center surround feature detector. Combining a 3dcenter surround feature point detector and the use of an integral image will accelerate the building of the scale space. Meantime, we also use Harris criterion to perform line suppression to enhance the stability of feature point detection. The experimental results indicate that our new detectors make the balance between computational efficiency andrepetition rate.
The role division on social networks is an important technology,for its potential values in analyzing social communities,exploring information dissemination rules and improving directional *** paper is focused on the ...
详细信息
ISBN:
(纸本)9781479932801
The role division on social networks is an important technology,for its potential values in analyzing social communities,exploring information dissemination rules and improving directional *** paper is focused on the research of role division from the perspective of information *** propose to measure users' secondary ability of information dissemination by weighting their neighborhoods' capability of information dissemination based on existing user characterization *** initial clustercenters in k-means are optimized by considering the characteristics of information sharing anddissemination on social ***,users are classified into three categories: ordinary users,bridge users and core *** results on Sina micro-blog dataset show that the proposed algorithm can betterdifferentiate users according to information dissemination.
An appropriate traffic model is the fundamental of many research topics on Internet. But the flourish of Internet applications results in more and more traffic generating mechanisms and the complexity of the traffic t...
详细信息
An appropriate traffic model is the fundamental of many research topics on Internet. But the flourish of Internet applications results in more and more traffic generating mechanisms and the complexity of the traffic transmitting makes the problem worse. Thus it is a key issue to keep the traffic model simple and easy to be adjusted. This paper proposes a new traffic model based on decomposition. In this model, the traffic is decomposed into two components and each component is modeled with an L-system. Analysis shows that the model can reserve the inherent structure anddynamics introduced in traffic generating and transmitting progress and is easy to be tuned. Simulations on real-world traces demonstrate the performance.
recurrent neural network language models have solved the problems of data sparseness anddimensionality disaster which exist in traditional N-gram *** have recently demonstrated state-of-the-art performance in speech ...
详细信息
recurrent neural network language models have solved the problems of data sparseness anddimensionality disaster which exist in traditional N-gram *** have recently demonstrated state-of-the-art performance in speech recognition,machine translation and other *** this paper,we improve the model performance by providing contextual word vectors in association with *** method can reinforce the ability of learning long-distance information using vectors training from Skip-gram *** experimental results show that the proposed method can improve the perplexity performance significantly on Penn Treebank *** we further apply the models to speech recognition task on the Wall Street Journal corpora,where we achieve obvious improvements in word-error-rate.
Hash tables are widely used in networking. Traditional hash bucket has to read out all the items in the bucket to confirm a false positive, which is inefficient. In this paper, an indexed hash bucket structure is prop...
详细信息
Hash tables are widely used in networking. Traditional hash bucket has to read out all the items in the bucket to confirm a false positive, which is inefficient. In this paper, an indexed hash bucket structure is proposed. In the structure, all items in a bucket would be indexed with unique identifiers using a Perfect Hash Function. When accessing a bucket to check a possible match, the identifiers were read out instead of the entire items. Then, according to the identifiers, we can calculate where the only possible match item locates in the bucket, which is the only item we need to read out then, no matter it’s a match or not, no matter how many items are in the bucket. Thus, at most two memory accesses are needed for every search. Analysis and simulations show that the proposed method can enhance the detecting efficiency and greatly improve the performance in the worst case.
The past decade has witnessed the birth of the golden age of precision cosmology. In particular, cosmic microwave back- ground （CMB） measurements have provided a powerful observational window to probe the earliest e...
详细信息
The past decade has witnessed the birth of the golden age of precision cosmology. In particular, cosmic microwave back- ground （CMB） measurements have provided a powerful observational window to probe the earliest epochs of our Universe. recently （March, 2014）, the Background Imaging of Cosmic Extragalactic Polarization （BICEP2） telescope at the South Pole announced the first detection of primordial B-mode polarization of the CMB that, if confirmed,
As the number of modules in system-on-Chip increases, the topology is more likely to suffer from excessive end-to-end hop-counts, causing an increase of power consumption and area overhead. Concerning this, a novel hi...
详细信息
ISBN:
(纸本)9781479932801
As the number of modules in system-on-Chip increases, the topology is more likely to suffer from excessive end-to-end hop-counts, causing an increase of power consumption and area overhead. Concerning this, a novel hierarchical topology called CHMesh is proposed, which is based on mesh and Cmesh. Correspondingly, this article elaborates on a shortest deadlock-free CHXY routing algorithm which has a low complexity. Experimental results demonstrate that, compared with traditional mesh andref-mesh, the CHMesh can increase the average throughput by about 60% and 10% respectively under the nonuniform traffic patterns, presenting more advantages on large-scale NoC.
The calculation is large to estimate the TdOA and FdOA with cross ambiguity function. Existing algorithms which are based on the ergodic theory have poorreal-time performance. To solve this problem, the genetic algor...
详细信息
The calculation is large to estimate the TdOA and FdOA with cross ambiguity function. Existing algorithms which are based on the ergodic theory have poorreal-time performance. To solve this problem, the genetic algorithm is proposed with improvements based on the characteristics of cross ambiguity function. With the self-adapting mutation probability by following the convergence extent of the population and multiple population initializations, the diversity of the population is effectively improved to prevent the algorithm into a local optimum. The simulation results show that the computational efficiency of the improved algorithm, compared with the existing algorithms, is greatly improved, and the TdOA/FdOA estimation results can quickly be obtained.
As an efficient pre-processing technology, superpixel has been widely used in image segmentation, which improves the final results andreduces the complexity of the subsequent processing tasks. This paper presents an ...
详细信息
ISBN:
(纸本)9781479958368
As an efficient pre-processing technology, superpixel has been widely used in image segmentation, which improves the final results andreduces the complexity of the subsequent processing tasks. This paper presents an interactive image segmentation method based on the hierarchical superpixels initialization andregion merging. It firstly proposes a novel hierarchical superpixels initialization framework including the coarse segmentation and the precise re-segmentation, which significantly reduces the number of superpixels, while keeping the boundaries adherence of superpixels satisfactory. Then the color histogram feature is applicable to the similarity measurement between two neighboring superpixels, and the Maximum A Posteriori Probability (MAP) estimation based on the color histogram is incorporated to estimate the probability that each non-marker superpixel belongs to the background. This paper establishes two different similarity measurements for the different stages in the MSrM merging procedure. Experiments verify the effectiveness of the proposed hierarchical framework to generate desired superpixels. The new interactive segmentation method can rapidly extract the object from the background.
暂无评论