Previous studies have shown that semantically meaningful representations of words and text can be acquired through neural embedding models. In particular, paragraph vector (PV) models have shown impressive performance...
详细信息
Single image dehazing is an important low-level vision task with many applications. Early researches have inves-tigated different kinds of visual priors to address this problem. However, they may fail when their assum...
详细信息
Phase information is ignored for almost all voice activity detection (VAD). To exploit full information in the original signal, this paper proposes a deep neural network (DNN) using magnitude and phase information (th...
详细信息
ISBN:
(纸本)9781509060689
Phase information is ignored for almost all voice activity detection (VAD). To exploit full information in the original signal, this paper proposes a deep neural network (DNN) using magnitude and phase information (that is, phase aware DNN) to achieve better VAD performance. Mel-frequency cepstral coefficient (MFCC), power-normalized cepstral coefficients (PNCC), instantaneous frequency derivative (IF), baseband phase difference (BPD) and modified group delay cepstral coefficient (MGDCC) are used as magnitude and phase information. The proposed methods were evaluated using CENSREC-1-C database under noise condition. The results show that the phase aware DNN significantly outperforms the DNN using only magnitude information. For DNN-based classifier, the equal error rate (EER) was reduced from 23.70% of MFCC, to 20.43% of joint dual magnitude and single phase features (augmenting PNCC, MGDCC and IF), to 19.92% of joint dual phase and single magnitude feature features (augmenting PNCC, MGDCC and BPD). By combining joint dual magnitude and single phase features with joint dual phase and single magnitude features, the EER was reduced to 19.44%.
The speed control of permanent magnet brushed (PMB) DC motors at low speeds is difficult due to the nonlinearity caused by various types of frictions. Under parameter uncertainty, the speed control becomes more diffic...
详细信息
ISBN:
(纸本)9781509061839
The speed control of permanent magnet brushed (PMB) DC motors at low speeds is difficult due to the nonlinearity caused by various types of frictions. Under parameter uncertainty, the speed control becomes more difficult. In this paper, to handle the parameter uncertainty, we propose a dynamic neural network to adaptively reconstruct or learn the dynamics of PMB DC motors. Then, based on the parameters of the neural dynamic model, a near-optimal dynamic neural controller is designed and proposed for the speed control of PMB DC motors with frictions considered under parameter uncertainty. Simulations substantiate the efficacy of the proposed dynamic neural model and adaptive near-optimal controller for PMB DC motors with fully unknown parameters.
Results are reported from a search for the lepton flavor violating decay τ → 3µ in proton-proton collisions at √s = 13 TeV. The data sample corresponds to an integrated luminosity of 33.2 fb−1 recorded by the ...
Neural Machine Translation (NMT) has become a popular technology in recent years, and beam search is its de facto decoding method due to the shrunk search space and reduced computational complexity. However, since it ...
ISBN:
(纸本)9781510860964
Neural Machine Translation (NMT) has become a popular technology in recent years, and beam search is its de facto decoding method due to the shrunk search space and reduced computational complexity. However, since it only searches for local optima at each time step through one-step forward looking, it usually cannot output the best target sentence. Inspired by the success and methodology of AlphaGo, in this paper we propose using a prediction network to improve beam search, which takes the source sentence x, the currently available decoding output y1, ···, yt-1 and a candidate word w at step t as inputs and predicts the long-term value (e.g., BLEU score) of the partial target sentence if it is completed by the NMT model. Following the practice in reinforcement learning, we call this prediction network value network. Specifically, we propose a recurrent structure for the value network, and train its parameters from bilingual data. During the test time, when choosing a word w for decoding, we consider both its conditional probability given by the NMT model and its long-term value predicted by the value network. Experiments show that such an approach can significantly improve the translation accuracy on several translation tasks.
The transverse momentum (pT) distributions of Λ, Ξ−, and Ω− baryons, their antiparticles, and KS0 mesons are measured in proton-proton (pp) and proton-lead (pPb) collisions at a nucleon-nucleon center-of-mass energ...
详细信息
The transverse momentum (pT) distributions of Λ, Ξ−, and Ω− baryons, their antiparticles, and KS0 mesons are measured in proton-proton (pp) and proton-lead (pPb) collisions at a nucleon-nucleon center-of-mass energy of 5.02 TeV over a broad rapidity range. The data, corresponding to integrated luminosities of 40.2 nb−1 and 15.6 μb−1 for pp and pPb collisions, respectively, were collected by the CMS experiment. The nuclear modification factor RpPb, which is defined as the ratio of the particle yield in pPb collisions and a scaled pp reference, is measured for each particle. A strong dependence on particle species is observed in the pT range from 2 to 7 GeV, where RpPb for KS0 is consistent with unity, while an enhancement ordered by strangeness content and/or particle mass is observed for the three baryons. In pPb collisions, the strange hadron production is asymmetric about the nucleon-nucleon center-of-mass rapidity. Enhancements, which depend on the particle type, are observed in the direction of the Pb beam. The results are compared with predictions from epos lhc, which includes parametrized radial flow. The model is in qualitative agreement with the RpPb data, but fails to describe the dependence on particle species in the yield asymmetries measured away from midrapidity in pPb collisions.
The differential cross section and charge asymmetry for inclusive W boson production at √s = 13 TeV is measured for the two transverse polarization states as a function of the W boson absolute rapidity. The measureme...
详细信息
By moving computations from computing nodes to storage nodes, active storage technology provides an efficient for data-intensive high-performance computing applications. The existing studies have neglected the heterog...
详细信息
By moving computations from computing nodes to storage nodes, active storage technology provides an efficient for data-intensive high-performance computing applications. The existing studies have neglected the heterogeneity of storage nodes on the performance of active storage systems. We introduce CADP, a capability-aware data placement scheme for heterogeneous active storage systems to obtain high-performance data processing. The basic idea of CADP is to place data on storage nodes based on their computing capability and storage capability, so that the load-imbalance among heterogeneous servers can be avoided. We have implemented CADP under a parallel I/O system. The experimental results show that the proposed capability-aware data placement scheme can improve the active storage system performance significantly.
暂无评论