查看更多>>摘要：Clustering is fundamental in many fields with big data. In this paper, a novel method based on Topological graph partition (TGP) is proposed to group objects. A topological graph is created for a data set with many objects, in which an object is connected to k nearest neighbors. By computing the weight of each object, a decision graph under probability comes into being. A cut threshold is conveniently selected where the probability of weight anomalously becomes large. With the threshold, the topological graph is cut apart into several sub-graphs after the noise edges are cut off, in which a connected sub-graph is treated as a cluster. The compared experiments demonstrate that the proposed method is more robust to cluster the data sets with high dimensions, complex distribution, and hidden noises. It is not sensitive to input parameter, we need not more priori knowledge.
查看更多>>摘要：Despite the large work made in interactive mesh deformation, manipulating a geometrically complex mesh and producing realistic deformation result is still a challenging work. Example-driven deformation methods distinctly simplify the modeling process and produce realistic deformation result by incorporating knowledge learned from shape space. We introduce a rotation invariant feature representation and a reconstruction framework to accurately reconstruct the vertex positions. Our feature representation allows both interpolation as well as extrapolation and can effectively blend multiple shapes. Based on this, we achieve an example-driven approach to mesh deformations. By using a collection of models as examples, our method produces natural deformation results guided by them even with large movement of handles. We will apply our representation and reconstruction method to semantic deformation transfer. The experimental results have demonstrated the effectiveness of the proposed methods.
查看更多>>摘要：The problem of approximation of ho-mogeneous random field from asymmetric local average sampling is considered in this paper. As a general sampling result, a sufficient condition is obtained to ensure the homogeneous random field be approximated from local averages with probability 1, which extended the result for weak sense stochastic process from local averages to homogeneous random field.
查看更多>>摘要：False alarms and misdetections caused by the carrier leakage problem attract increasing attention for resource-limited platforms. To tackle the problem, we propose a target protected carrier leakage cancellation algorithm to rebuild the leakage signal from reference cells with those cells randomly updated based on each detection result to exclude the targets. The proposed algorithm eliminates the carrier leakage effectively without introducing false target due to the non-involvement of the target information in the reference cells based on a random update strategy. The noise level of our proposed algorithm is less than that of the commonly used algorithms such as the Cell average constant false alarm rate (CA-CFAR) detection method. As verified in the simulations, the proposed algorithm performs better than the CA-CFAR method.
查看更多>>摘要：Long short-term memory RNNs (LSTM-RNNs) have shown great success in the Automatic speech recognition (ASR) field and have become the state-of-the-art acoustic model for time-sequence modeling tasks. However, it is still difficult to train deep LSTM-RNNs while keeping the parameter number small. We use the highway connections between memory cells in adjacent layers to train a small-footprint highway LSTM-RNNs (HLSTM-RNNs), which are deeper and thinner compared to conventional LSTM-RNNs. The experiments on the Switchboard (SWBD) indicate that we can train thinner and deeper HLSTM-RNNs with a smaller parameter number than the conventional 3-layer LSTM-RNNs and a lower Word error rate (WER) than the conventional one. Compared with the counterparts of small-footprint LSTM-RNNs, the small-footprint HLSTM-RNNs show greater reduction in WER.
查看更多>>摘要：This study attempted to answer complicated free-description questions in Chinese Gaokao Reading comprehension (RC) tasks. We found that quite a few questions can be answered by extracting sentences from the document and combining them, so we used a pipeline approach with two components:Answer sentence extraction (ASE) and Answer sentence fusion (ASF). Semantic vector similarity and topical distribution similarity were explored for ASE. Integer linear programming strategy was used for ASF, which combined dependencies with the language model, based on word importance. As a first step towards the new challenge, we obtained some encouraging results on actual exam questions in Chinese subject's RC tasks of Beijing Gaokao, which helped us obtain insights into techniques needed to solve real-word complex questions.
查看更多>>摘要：The major challenge that text sentiment classification modeling faces is how to capture the intrinsic semantic, emotional dependence information and the key part of the emotional expression of text. To solve this problem, we proposed a Coordinated CNN-LSTM-Attention(CCLA) model. We learned the vector representations of sentence with CCLA unit. Semantic and emotional information of sentences and their relations are adaptively encoded to vector representations of document. We used softmax regression classifier to identify the sentiment tendencies in the text. Compared with other methods, the CCLA model can well capture the local and long distance semantic and emotional information. Experimental results demonstrated the effectiveness of CCLA model. It shows superior performances over several state-of-the-art baseline methods.
查看更多>>摘要：In order to overcome some problems caused by improper parameters selection when applying Least mean square (LMS), Normalized LMS (NLMS) or Recursive least square (RLS) algorithms to estimate co-efficients of second-order Volterra filter, a novel Davidon-Fletcher-Powell-based Second-order Volterra filter (DFP-SOVF) is proposed. Analysis of computational complexity and stability are presented. Simulation results of system parameter identification show that the DFP algorithm has fast convergence and excellent robustness than LMS and RLS algorithm. Prediction results of applying DFP-SOVF model to single step predictions for Lorenz chaotic time series illustrate stability and convergence and there have not divergence problems. For the measured multi-frame speech signals, prediction accuracy using DFP-SOVF model is better than that of Linear prediction (LP). The DFP-SOVF model can better predict chaotic time series and the real measured speech signal series.
查看更多>>摘要：Granular computing is a very hot research field in computer science in recent years. This paper introduces a new granular computing model based on algebraic structure, in which the granule structure is assumed as a binary operator and the granulation is based on a congruence relation. Following the homomorphic consistency principle, the methods of granulation (granularity coarsening) and granularity combination (granularity refinement) are introduced, and the corresponding numerical examples show that these methods are efficient and applicable. These works have enriched the granular computing models from structure and provided theoretical basis for the combination of granular computing theory and algebraic theory.
查看更多>>摘要：By combing Spatial modulation (SM) with the Amplify-and-forward (AF) cooperative communica-tion, a cooperative SM scheme with AF (AF-SM) protocol is presented, and the corresponding Bit error rate (BER) performance is investigated. Based on the performance analysis, the error probability of antenna index estimation (Pa) and the error probability of symbol estimation (Pd), which constitutes the overall average BER, are derived, respectively. As a result, tightly approximate closed-form expressions of Pa and Pd are attained, respectively. Using these expressions, the closed-form overall average BER is achieved. During the analysis, the computational complexity of AF-SM detector is also provided in terms of the complex addition and multiplication numbers. Besides, the asymptotic BER at high SNR and the corresponding diversity gain are further derived, and the resultant diversity gain of Nr + 1 is obtained for the system with Nr receive antennas. Simulation results show that our theoretical analysis is valid, and provides good performance evaluation method for cooperative SM system. Moreover, the BER performance can be effectively improved as Nr increases due to higher diversity gain.
NETL
NSTL
万方数据