Face detection and recognition is a hot research subject in the fields of image processing, pattern recognition and computer vision, which emphasize on the enhancement of the precision of detection in as short a time ...
详细信息
As tokenization is usually ambiguous for many natural languages such as Chinese and Korean, tokenization errors might potentially introduce translation mistakes for translation systems that rely on 1-best tokenization...
详细信息
As tokenization is usually ambiguous for many natural languages such as Chinese and Korean, tokenization errors might potentially introduce translation mistakes for translation systems that rely on 1-best tokenizations. While using lattices to offer more alternatives to translation systems have elegantly alleviated this problem, we take a further step to tokenize and translate jointly. Taking a sequence of atomic units that can be combined to form words in different ways as input, our joint decoder produces a tokenization on the source side and a translation on the target side simultaneously. By integrating tokenization and translation features in a discriminative framework, our joint decoder outperforms the baseline translation systems using 1-best tokenizations and lattices significantly on both Chinese- English and Korean-Chinese tasks. Interestingly, as a tokenizer, our joint decoder achieves significant improvements over monolingual Chinese tokenizers.
Parsing plays an important role in semantic role lab.ling (SRL) because most SRL systems infer semantic relations from I-best parses. Therefore, parsing errors inevitably lead to lab.ling mistakes. To alleviate this p...
详细信息
A new incident source with different angles was constructed for dealing with wide-angle scattering problems. Considering the impendence matrix in method of moments (MOM) is independent from incident angles, the equiva...
详细信息
Ana\ysis of electromagnetic scattering characteristics from wire antennas using method of moments has been performed over the past decades. In this paper, an efficient technique is adopted to evaluate the potential an...
详细信息
Ana\ysis of electromagnetic scattering characteristics from wire antennas using method of moments has been performed over the past decades. In this paper, an efficient technique is adopted to evaluate the potential and reaction integrals arising in radar cross section (RCS) computation of cylindrical antenna by method of moments. The singularity of potential and reaction integrals is extracted through the semi-analytical method. The induced current on the surface and RCS of cylindrical antenna are calculated. Numerical examples are presented to illustrate the computational accuracy and efficiency of the proposed technique.
Traditional 1-best translation pipelines suffer a major drawback: the errors of 1- best outputs, inevitably introduced by each module, will propagate and accumulate along the pipeline. In order to alleviate this probl...
详细信息
Traditional 1-best translation pipelines suffer a major drawback: the errors of 1- best outputs, inevitably introduced by each module, will propagate and accumulate along the pipeline. In order to alleviate this problem, we use compact structures, lattice and forest, in each module instead of 1-best results. We integrate both lattice and forest into a single tree-to-string system, and explore the algorithms of lattice parsing, lattice-forest-based rule extraction and decoding. More importantly, our model takes into account all the probabilities of different steps, such as segmentation, parsing, and translation. The main advantage of our model is that we can make global decision to search for the best segmentation, parse-tree and translation in one step. Medium-scale experiments show an improvement of +0.9 BLEU points over a state-of-the-art forest-based baseline.
Hand segmentation is the basis of many vision-based hand gesture applications in human computer interaction (HCI). This paper proposes a novel method of skin color weighted disparity competition to incorporate the ski...
详细信息
One fundamental problem in services computing is how to bridge the gap between business requirements and various heterogeneous IT services. This involves eliciting business requirements and building a solution accordi...
详细信息
Active contours have been one of the most successful methods for image segmentation during the last two decades, but one of the shortcomings of being unable to converge to concavity is a handicap to its effectiveness....
详细信息
Accurate endpoint detection is a necessary capability for speech recognition. A new energy measure method based on the empirical mode decomposition (EMD) algorithm and Teager energy operator (TEO) is proposed to locat...
详细信息
Accurate endpoint detection is a necessary capability for speech recognition. A new energy measure method based on the empirical mode decomposition (EMD) algorithm and Teager energy operator (TEO) is proposed to locate endpoint intervals of a speech signal embedded in noise. With the EMD, the noise signals can be decomposed into different numbers of sub-signals called intrinsic mode functions (IMFs), which is a zero-mean AM-FM component. Then TEO can be used to extract the desired feature of the modulation energy for IMF components. In order to show the effectiveness of the proposed method, examples are presented to show that the new measure is more effective than traditional measures. The present experimental results show that the measure can be used to improve the performance of endpoint detection algorithms and the accuracy of this algorithm is quite satisfactory and acceptable.
暂无评论