arXiv subjects

Jun Zhang

Publications and source records attributed to Jun Zhang.

At least 1,099 records · Page 61Linked to original sources

arXiv2017

Deep LSTM for Large Vocabulary Continuous Speech Recognition

Recurrent neural networks (RNNs), especially long short-term memory (LSTM) RNNs, are effective network for sequential task like speech recognition. Deeper LSTM models perform well on large vocabulary continuous speech recognition, because of their impressive learning ability. However, it is more difficult to train a deeper network. We introduce a training framework with layer-wise training and exponential moving average methods for deeper LSTM models. It is a competitive framework that LSTM models of more than 7 layers are successfully trained on Shenma voice search data in Mandarin and they outperform the deep LSTM models trained by conventional approach. Moreover, in order for online streaming speech recognition applications, the shallow model with low real time factor is distilled from the very deep model. The recognition accuracy have little loss in the distillation process. Therefore, the model trained with the proposed training framework reduces relative 14\% character error rate, compared to original model which has the similar real-time capability. Furthermore, the novel transfer learning strategy with segmental Minimum Bayes-Risk is also introduced in the framework. The strategy makes it possible that training with only a small part of dataset could outperform full dataset training from the beginning.

cs.CL↗

arXiv2017

A scalable approach for tree segmentation within small-footprint airborne LiDAR data

This paper presents a distributed approach that scales up to segment tree crowns within a LiDAR point cloud representing an arbitrarily large forested area. The approach uses a single-processor tree segmentation algorithm as a building block in order to process the data delivered in the shape of tiles in parallel. The distributed processing is performed in a master-slave manner, in which the master maintains the global map of the tiles and coordinates the slaves that segment tree crowns within and across the boundaries of the tiles. A minimal bias was introduced to the number of detected trees because of trees lying across the tile boundaries, which was quantified and adjusted for. Theoretical and experimental analyses of the runtime of the approach revealed a near linear speedup. The estimated number of trees categorized by crown class and the associated error margins as well as the height distribution of the detected trees aligned well with field estimations, verifying that the distributed approach works correctly. The approach enables providing information of individual tree locations and point cloud segments for a forest-level area in a timely manner, which can be used to create detailed remotely sensed forest inventories. Although the approach was presented for tree segmentation within LiDAR point clouds, the idea can also be generalized to scale up processing other big spatial datasets. Highlights: - A scalable distributed approach for tree segmentation was developed and theoretically analyzed. - ~2 million trees in a 7440 ha forest was segmented in 2.5 hours using 192 cores. - 2% false positive trees were identified as a result of the distributed run. - The approach can be used to scale up processing other big spatial data

cs.DC↗

arXiv2017

The classical correlation limits the ability of the measurement-induced average coherence

Coherence is the most fundamental quantum feature in quantum mechanics. For a bipartite quantum state, if a measurement is performed on one party, the other party, based on the measurement outcomes, will collapse to a corresponding state with some probability and hence gain the average coherence. It is shown that the average coherence is not less than the coherence of its reduced density matrix. In particular, it is very surprising that the extra average coherence (and the maximal extra average coherence with all the possible measurements taken into account) is upper bounded by the classical correlation of the bipartite state instead of the quantum correlation. We also find the sufficient and necessary condition for the null maximal extra average coherence. Some examples demonstrate the relation and, moreover, show that quantum correlation is neither sufficient nor necessary for the nonzero extra average coherence within a given measurement. In addition, the similar conclusions are drawn for both the basis-dependent and the basis-free coherence measure.

quant-ph↗

arXiv2017

Mobility-Aware Caching in D2D Networks

Caching at mobile devices can facilitate device-to-device (D2D) communications, which may significantly improve spectrum efficiency and alleviate the heavy burden on backhaul links. However, most previous works ignored user mobility, thus having limited practical applications. In this paper, we take advantage of the user mobility pattern by the inter-contact times between different users, and propose a mobility-aware caching placement strategy to maximize the data offloading ratio, which is defined as the percentage of the requested data that can be delivered via D2D links rather than through base stations (BSs). Given the NP-hard caching placement problem, we first propose an optimal dynamic programming (DP) algorithm to obtain a performance benchmark, but with much lower complexity than exhaustive search. We will then prove that the problem falls in the category of monotone submodular maximization over a matroid constraint, and propose a time-efficient greedy algorithm, which achieves an approximation ratio as 1/2. Simulation results with real-life data-sets will validate the effectiveness of our proposed mobility-aware caching placement strategy. We observe that users moving at either a very low or very high speed should cache the most popular files, while users moving at a medium speed should cache less popular files to avoid duplication.

cs.IT↗

arXiv2017

Exponential Moving Average Model in Parallel Speech Recognition Training

As training data rapid growth, large-scale parallel training with multi-GPUs cluster is widely applied in the neural network model learning currently.We present a new approach that applies exponential moving average method in large-scale parallel training of neural network model. It is a non-interference strategy that the exponential moving average model is not broadcasted to distributed workers to update their local models after model synchronization in the training process, and it is implemented as the final model of the training system. Fully-connected feed-forward neural networks (DNNs) and deep unidirectional Long short-term memory (LSTM) recurrent neural networks (RNNs) are successfully trained with proposed method for large vocabulary continuous speech recognition on Shenma voice search data in Mandarin. The character error rate (CER) of Mandarin speech recognition further degrades than state-of-the-art approaches of parallel training.

cs.CL↗

arXiv2017

Kriging interpolating cosmic velocity field. II. Taking anistropies and multistreaming into account

Measuring the volume-weighted peculiar velocity statistics from inhomogeneously and sparsely distributed galaxies/halos, by existing velocity assignment methods, suffers from a significant sampling artifact. As an alternative, the Kriging interpolation based on Gaussian processes was introduced and evaluated [Y. Yu, J. Zhang, Y. Jing, and P. Zhang, Phys. Rev. D 92, 083527 (2015)]. Unfortunately, the most straightforward application of Kriging does not perform better than the existing methods in the literature. In this work, we investigate two physically motivated extensions. The first takes into account of the anisotropic velocity correlations. The second introduces the nugget effect, on account of multi-streaming of the velocity field. We find that the performance is indeed improved. For sparsely sampled data [$n_P\lesssim 6\times 10^{-3}(h^{-1} {\rm Mpc})^{-3}$] where the sampling artifact is the most severe, the improvement is significant and is two-fold: 1) The scale of reliable measurement of the velocity power spectrum is extended by a factor $\sim 1.6$, and 2) the dependence on the velocity correlation prior is weakened by a factor of $\sim 2$. We conclude that such extensions are desirable for accurate velocity assignment by Kriging.

astro-ph.CO↗

arXiv2017

Mobility Increases the Data Offloading Ratio in D2D Caching Networks

Caching at mobile devices, accompanied by device-to-device (D2D) communications, is one promising technique to accommodate the exponentially increasing mobile data traffic. While most previous works ignored user mobility, there are some recent works taking it into account. However, the duration of user contact times has been ignored, making it difficult to explicitly characterize the effect of mobility. In this paper, we adopt the alternating renewal process to model the duration of both the contact and inter-contact times, and investigate how the caching performance is affected by mobility. The data offloading ratio, i.e., the proportion of requested data that can be delivered via D2D links, is taken as the performance metric. We first approximate the distribution of the communication time for a given user by beta distribution through moment matching. With this approximation, an accurate expression of the data offloading ratio is derived. For the homogeneous case where the average contact and inter-contact times of different user pairs are identical, we prove that the data offloading ratio increases with the user moving speed, assuming that the transmission rate remains the same. Simulation results are provided to show the accuracy of the approximate result, and also validate the effect of user mobility.

cs.IT↗

arXiv2017

Stochastic Joint Radio and Computational Resource Management for Multi-User Mobile-Edge Computing Systems

Mobile-edge computing (MEC) has recently emerged as a prominent technology to liberate mobile devices from computationally intensive workloads, by offloading them to the proximate MEC server. To make offloading effective, the radio and computational resources need to be dynamically managed, to cope with the time-varying computation demands and wireless fading channels. In this paper, we develop an online joint radio and computational resource management algorithm for multi-user MEC systems, with the objective as minimizing the long-term average weighted sum power consumption of the mobile devices and the MEC server, subject to a task buffer stability constraint. Specifically, at each time slot, the optimal CPU-cycle frequencies of the mobile devices are obtained in closed forms, and the optimal transmit power and bandwidth allocation for computation offloading are determined with the Gauss-Seidel method; while for the MEC server, both the optimal frequencies of the CPU cores and the optimal MEC server scheduling decision are derived in closed forms. Besides, a delay-improved mechanism is proposed to reduce the execution delay. Rigorous performance analysis is conducted for the proposed algorithm and its delay-improved version, indicating that the weighted sum power consumption and execution delay obey an $\left[O\left(1\slash V\right),O\left(V\right)\right]$ tradeoff with $V$ as a control parameter. Simulation results are provided to validate the theoretical analysis and demonstrate the impacts of various parameters.

cs.IT↗

arXiv2017

Joint Task Offloading Scheduling and Transmit Power Allocation for Mobile-Edge Computing Systems

Mobile-edge computing (MEC) has emerged as a prominent technique to provide mobile services with high computation requirement, by migrating the computation-intensive tasks from the mobile devices to the nearby MEC servers. To reduce the execution latency and device energy consumption, in this paper, we jointly optimize task offloading scheduling and transmit power allocation for MEC systems with multiple independent tasks. A low-complexity sub-optimal algorithm is proposed to minimize the weighted sum of the execution delay and device energy consumption based on alternating minimization. Specifically, given the transmit power allocation, the optimal task offloading scheduling, i.e., to determine the order of offloading, is obtained with the help of flow shop scheduling theory. Besides, the optimal transmit power allocation with a given task offloading scheduling decision will be determined using convex optimization techniques. Simulation results show that task offloading scheduling is more critical when the available radio and computational resources in MEC systems are relatively balanced. In addition, it is shown that the proposed algorithm achieves near-optimal execution delay along with a substantial device energy saving.

cs.IT↗

arXiv2017

Reliability of analog quantum simulation

Analog quantum simulators (AQS) will likely be the first nontrivial application of quantum technology for predictive simulation. However, there remain questions regarding the degree of confidence that can be placed in the results of AQS since they do not naturally incorporate error correction. Specifically, how do we know whether an analog simulation of a quantum model will produce predictions that agree with the ideal model in the presence of inevitable imperfections? At the same time, there is a widely held expectation that certain quantum simulation questions will be robust to errors and perturbations in the underlying hardware. Resolving these two points of view is a critical step in making the most of this promising technology. In this work we formalize the notion of AQS reliability by determining sensitivity of AQS outputs to underlying parameters, and formulate conditions for robust simulation. Our approach naturally reveals the importance of model symmetries in dictating the robust properties. To demonstrate the approach, we characterize the robust features of a variety of quantum many-body models.

quant-ph↗

arXiv2017

Alternating Minimization for Hybrid Precoding in Multiuser OFDM mmWave Systems

Hybrid precoding is a cost-effective approach to support directional transmissions for millimeter wave (mmWave) communications. While existing works on hybrid precoding mainly focus on single-user single-carrier transmission, in practice multicarrier transmission is needed to combat the much increased bandwidth, and multiuser MIMO can provide additional spatial multiplexing gains. In this paper, we propose a new hybrid precoding structure for multiuser OFDM mmWave systems, which greatly simplifies the hybrid precoder design and is able to approach the performance of the fully digital precoder. In particular, two groups of phase shifters are combined to map the signals from radio frequency (RF) chains to antennas. Then an effective hybrid precoding algorithm based on alternating minimization (AltMin) is proposed, which will alternately optimize the digital and analog precoders. A major algorithmic innovation is a LASSO formulation for the analog precoder, which yields computationally efficient algorithms. Simulation results will show the performance gain of the proposed algorithm. Moreover, it will reveal that canceling the interuser interference is critical in multiuser OFDM hybrid precoding systems.

cs.IT↗

arXiv2017

A robust approach for tree segmentation in deciduous forests using small-footprint airborne LiDAR data

This paper presents a non-parametric approach for segmenting trees from airborne LiDAR data in deciduous forests. Based on the LiDAR point cloud, the approach collects crown information such as steepness and height on-the-fly to delineate crown boundaries, and most importantly, does not require a priori assumptions of crown shape and size. The approach segments trees iteratively starting from the tallest within a given area to the smallest until all trees have been segmented. To evaluate its performance, the approach was applied to the University of Kentucky Robinson Forest, a deciduous closed-canopy forest with complex terrain and vegetation conditions. The approach identified 94% of dominant and co-dominant trees with a false detection rate of 13%. About 62% of intermediate, overtopped, and dead trees were also detected with a false detection rate of 15%. The overall segmentation accuracy was 77%. Correlations of the segmentation scores of the proposed approach with local terrain and stand metrics was not significant, which is likely an indication of the robustness of the approach as results are not sensitive to the differences in terrain and stand structures.

cs.CV↗

arXiv2016

Stronger uncertainty relations with improvable upper and lower bounds

We utilize quantum superposition principle to establish the improvable upper and lower bounds on the stronger uncertainty relation, i.e., the "weighted-like" sum of the variances of observables. Our bounds include some free parameters which not only guarantee the nontrivial bounds but also can effectively control the bounds as tightly as one expects. Especially, these parameters don't obviously depend on the state and observables. It also implies one advantage of our method that any nontrivial bound can always be more improvable. In addition, we generalize both bounds to the uncertainty relation with multiple observables, but the perfect tightness is not changed. Examples are given to illustrate the improvability of our bounds in each case.

quant-ph↗

arXiv2016

A fast iterative scheme for the linearized Boltzmann equation

An iterative scheme can be used to find a steady-state solution to the Boltzmann equation, however, it is very slow to converge in the near-continuum flow regime. In this paper, a synthetic iterative scheme is developed to speed up the solution of the linearized Boltzmann equation. The velocity distribution function is first solved by the conventional iterative scheme, then it is corrected such that the macroscopic flow velocity is governed by a diffusion equation which is asymptotic-preserving in the Navier-Stokes limit. The efficiency of the new scheme is verified by calculating the eigenvalue of the iteration, as well as solving for Poiseuille and thermal transpiration flows. The synthetic iterative scheme is significantly faster than the conventional iterative scheme in both the transition and the near-continuum flow regimes. Moreover, due to the asymptotic-preserving properties, the SIS needs less spatial resolution in the near-continuum flow regimes, which makes it even faster than the conventional iterative scheme. Using this synthetic iterative scheme, and the fast spectral approximation of the linearized Boltzmann collision operator, Poiseuille and thermal transpiration flows between two parallel plates, through channels of circular/rectangular cross sections, and various porous media are calculated over the whole range of gas rarefaction. Finally, the flow of a Ne-Ar gas mixture is solved based on the linearized Boltzmann equation with the Lennard-Jones potential for the first time, and the difference between these results and those using hard-sphere intermolecular potential is discussed.

physics.comp-ph↗

arXiv2016

Experimental measurement-device-independent quantum random number generation

The randomness from a quantum random number generator (QRNG) relies on the accurate characterization of its devices. However, device imperfections and inaccurate characterizations can result in wrong entropy estimation and bias in practice, which highly affects the genuine randomness generation and may even induce the disappearance of quantum randomness in an extreme case. Here we experimentally demonstrate a measurement-device-independent (MDI) QRNG based on time-bin encoding to achieve certified quantum randomness even when the measurement devices are uncharacterized and untrusted. The MDI-QRNG is randomly switched between the regular randomness generation mode and a test mode, in which four quantum states are randomly prepared to perform measurement tomography in real-time. With a clock rate of 25 MHz, the MDI-QRNG generates a final random bit rate of 5.7 Kbps. Such implementation with an all-fiber setup provides an approach to construct a fully-integrated MDI-QRNG with trusted but error-prone devices in practice.

quant-ph↗

arXiv2016

Preserving coherent spin and squeezed spin states of a spin-1 Bose-Einstein condensate with rotary echoes

A challenge in precision measurement with squeezed spin state arises from the spin dephasing due to stray magnetic fields. To suppress such environmental noises, we employ a continuous driving protocol, rotary echo, to enhance the spin coherence of a spin-1 Bose-Einstein condensate in stray magnetic fields. Our analytical and numerical results show that the coherent and the squeezed spin states are preserved for a significantly long time, compared to the free induction decay time, if the condition $hτ= mπ$ is met with $h$ the pulse amplitude and $τ$ pulse width. In particular, both the spin average and the spin squeezing, including the direction and the amplitude, are simultaneously fixed for a squeezed spin state. Our results point out a practical way to implement quantum measurements based on a spin-1 condensate beyond the standard quantum limit.

cond-mat.quant-gas↗

arXiv2016

Enhancement of a sunspot light wall with external disturbances

Based on the \emph{Interface Region Imaging Spectrograph} observations, we study the response of a solar sunspot light wall to external disturbances. A flare occurrence near the light wall caused material to erupt from the lower solar atmosphere into the corona. Some material falls back to the solar surface, and hits the light bridge (i.e., the base of the light wall), then sudden brightenings appear at the wall base followed by the rise of wall top, leading to an increase of the wall height. Once the brightness of the wall base fades, the height of the light wall begins to decrease. Five hours later, another nearby flare takes place, a bright channel is formed that extends from the flare towards the light bridge. Although no obvious material flow along the bright channel is found, some ejected material is conjectured to reach the light bridge. Subsequently, the wall base brightens and the wall height begins to increase again. Once more, when the brightness of the wall base decays, the wall top fluctuates to lower heights. We suggest, based on the observed cases, that the interaction of falling material and ejected flare material with the light wall results in the brightenings of wall base and causes the height of the light wall to increase. Our results reveal that the light wall can be not only powered by the linkage of \emph{p}-mode from below the photosphere, but may also be enhanced by external disturbances, such as falling material.

astro-ph.SR↗

arXiv2016

A Quantum Hamiltonian Identification Algorithm: Computational Complexity and Error Analysis

Quantum Hamiltonian identification is important for characterizing the dynamics of quantum systems, calibrating quantum devices and achieving precise quantum control. In this paper, an effective two-step optimization (TSO) quantum Hamiltonian identification algorithm is developed within the framework of quantum process tomography. In the identification method, different probe states are inputted into quantum systems and the output states are estimated using the quantum state tomography protocol via linear regression estimation. The time-independent system Hamiltonian is reconstructed based on the experimental data for the output states. The Hamiltonian identification method has computational complexity O(d^6) where d is the dimension of the system Hamiltonian. An error upper bound O(d^3/N^(1/2))$ is also established, where N is the resource number for the tomography of each output state, and several numerical examples demonstrate the effectiveness of the proposed TSO Hamiltonian identification method.

quant-ph↗