arXiv subjects
Publications and source records attributed to Xin Zhang.
Machine reading comprehension aims to teach machines to understand a text like a human and is a new challenging direction in Artificial Intelligence. This article summarizes recent advances in MRC, mainly focusing on two aspects (i.e., corpus and techniques). The specific characteristics of various MRC corpus are listed and compared. The main ideas of some typical MRC techniques are also described.
In this paper, we investigate the observational constraints on the scenario of vacuum energy interacting with cold dark matter. We consider eight typical interaction forms in such an interacting vacuum energy scenario. The observational data used in this work to constrain these models include the JLA sample of type Ia supernovae observation, the Planck 2015 distance priors data of cosmic microwave background anisotropies observation, the baryon acoustic oscillations data, and the Hubble constant direct measurement. We find that the current observational data almost equally favor these interacting vacuum energy models. We also find that for all these models of vacuum energy interacting with cold dark matter the case of no interaction is actually well consistent with the current observational data within 1$σ$ range.
This paper studies the ergodic mutual information of hybrid selection/maximal-ratio combining (H-S/MRC) diversity system under BPSK/QPSK modulations. We consider a simple single-input multiple-output (SIMO) channel, where a subset of branches are selected and combined using maximal-ratio combining (MRC) to maximize the instantaneous Signal to Noise Ratio (SNR) at the receiver. For independent and identically distributed (i.i.d.) Rayleigh flat fading, a general recursive expression is developed to estimate the ergodic input-output mutual information of the whole system. Besides analytical derivations, simulations are provided to demonstrate the feasibility and validity of the derived results.
High quality metal thin films and nanostructures are critical building blocks for next generation nanotechnologies. They comprise low-loss circuit elements in nanodevices, provide new catalytic pathways for water splitting and $CO_2$ reduction technologies, and enable the confinement of spatially extended electromagnetic waves to be harnessed for application in information processing, energy harvesting, engineered metamaterials, and new technologies that will operate in the quantum plasmonics limit. However, the controlled fabrication of high-definition single-crystal subwavelength metal nanostructures remains a significant hurdle, due to the tendency for polycrystalline metal growth using conventional physical vapor deposition methods, and the challenges associated with placing solution-grown nanocrystals in desired orientations and locations on a surface to fabricate functional devices. Here, we introduce a new scalable, green, wet chemical approach to monocrystalline noble metals that enables the fabrication of ultrasmooth, epitaxial, single-crystal films of controllable thickness. They are ideal for the subtractive manufacture of nanostructure through ion beam milling, and additive crystalline nanostructure via lithographic patterning to enable large area, single-crystal metamaterials and high aspect ratio nanowires. Our single-crystal nanostructures demonstrate improved feature quality and pattern transfer yield, reduced optical and resistive losses, tailored local fields, and greatly improved stability compared to polycrystalline structures, supporting greater local field enhancements and enabling new practical advances at the nanoscale.
Let $(Z,κ)$ be a Walsh Brownian motion with spinning measure $κ$. Suppose $μ$ is a probability measure on $\mathbb{R}^n$. We characterize all the $κ$ such that $μ$ is a stopping distribution of $(Z,κ)$. If we further restrict the solution to be integrable, we show that there would be only one choice of $κ$. We also generalize Vallois' embedding, and prove that it minimizes the expectation $\mathbb{E}[Ψ(L^Z_τ)]$ among all the admissible solutions $τ$, where $Ψ$ is a strictly convex function and $(L_t^Z)_{t \geq 0}$ is the local time of the Walsh Brownian motion at the origin.
We proposed a novel visual stimulus for brain-computer interface. The stimulus is in the form gaiting sequence of a human. The hypothesis is that observing such a visual stimulus would simultaneously induce 1) steady-state motion visual evoked potential (SSMVEP) in the occipital area, similarly to an SSVEP stimulus; and 2) sensorimotor rhythm (SMR) in the primary sensorimotor area, because such action observation (AO) could activate the mirror neuron system. Canonical correlation analysis (CCA) was used to detect SSMVEP from occipital EEG, and event-related spectral perturbations (ERSP) were used to identify SMR in the EEG from the sensorimotor area. The results showed that the proposed visual gaiting stimulus-induced SSMVEP, with classification accuracies of 88.9 $\pm$ 12.0% in a four-class scenario. More importantly, it induced clear and sustained event-related desynchronization/synchronization (ERD/ERS) in the EEG from the sensorimotor area, while no ERD/ERS in the sensorimotor area could be observed when the other two SSVEP stimuli were used. Further, for participants with sufficiently clear SSMVEP pattern (classification accuracy > 85%), the ERD index values in mu-beta band induced by the proposed gaiting stimulus were statistically different from that of the other two types of stimulus. Therefore, a novel BCI based on the proposed stimulus has potential in neurorehabilitation applications because it simultaneously has the high accuracy of an SSMVEP (~90% accuracy in a four-class setup) and the ability to activate sensorimotor cortex. And such potential will be further explored in future studies.
In this work, we consider to improve the model estimation efficiency by aggregating the neighbors' information as well as identify the subgroup membership for each node in the network. A tree-based $l_1$ penalty is proposed to save the computation and communication cost. We design a decentralized generalized alternating direction method of multiplier algorithm for solving the objective function in parallel. The theoretical properties are derived to guarantee both the model consistency and the algorithm convergence. Thorough numerical experiments are also conducted to back up our theory, which also show that our approach outperforms in the aspects of the estimation accuracy, computation speed and communication cost.
We present a deep neural network based method for the retrieval of watermarks from images of 3D printed objects. To deal with the variability of all possible 3D printing and image acquisition settings we train the network with synthetic data. The main simulator parameters such as texture, illumination and camera position are dynamically randomized in non-realistic ways, forcing the neural network to learn the intrinsic features of the 3D printed watermarks. At the end of the pipeline, the watermark, in the form of a two-dimensional bit array, is retrieved through a series of simple image processing and statistical operations applied on the confidence map generated by the neural network. The results demonstrate that the inclusion of synthetic DR data in the training set increases the generalization power of the network, which performs better on images from previously unseen 3D printed objects. We conclude that in our application domain of information retrieval from 3D printed objects, where access to the exact CAD files of the printed objects can be assumed, one can use inexpensive synthetic data to enhance neural network training, reducing the need for the labour intensive process of creating large amounts of hand labelled real data or the need to generate photorealistic synthetic data.
We investigate how the dark energy properties impact the constraints on the total neutrino mass in interacting dark energy (IDE) models. In this study, we focus on two typical interacting dynamical dark energy models, i.e., the interacting $w$ cold dark matter (I$w$CDM) model and the interacting holographic dark energy (IHDE) model. To avoid the large-scale instability problem in IDE models, we apply the parameterized post-Friedmann approach to calculate the perturbation of dark energy. We employ the Planck 2015 cosmic microwave background temperature and polarization data, combined with low-redshift measurements on baryon acoustic oscillation distance scales, type Ia supernovae, and the Hubble constant, to constrain the cosmological parameters. We find that the dark energy properties could influence the constraint limits on the total neutrino mass. Once dynamical dark energy is considered in the IDE models, the upper bounds of $\sum m_ν$ will be changed. By considering the values of $χ^2_{\rm min}$, we find that in these IDE models the normal hierarchy case is slightly preferred over the inverted hierarchy case; for example, $Δχ^2=2.720$ is given in the IHDE+$\sum m_ν$ model. In addition, we also find that in the I$w$CDM+$\sum m_ν$ model $β=0$ is consistent with current observational data inside the 1$σ$ range, and in the IHDE+$\sum m_ν$ model $β>0$ is favored at more than 2$σ$ level.
Semiconductors, a significant type of material in the information era, are becoming more and more powerful in the field of quantum information. In the last decades, semiconductor quantum computation was investigated thoroughly across the world and developed with a dramatically fast speed. The researches vary from initialization, control and readout of qubits, to the architecture of fault tolerant quantum computing. Here, we first introduce the basic ideas for quantum computing, and then discuss the developments of single- and two- qubit gate control in semiconductor. Till now, the qubit initialization, control and readout can be realized with relatively high fidelity and a programmable two-qubit quantum processor was even demonstrated. However, to further improve the qubit quality and scale it up, there are still some challenges to resolve such as the improvement of readout method, material development and scalable designs. We discuss these issues and introduce the forefronts of progress. Finally, considering the positive trend of the research on semiconductor quantum devices and recent theoretical work on the applications of quantum computation, we anticipate that semiconductor quantum computation may develop fast and will have a huge impact on our lives in the near future.
This paper studies the secrecy performance of multiple-input multiple-output (MIMO) wiretap channels, also termed as multiple-input multiple-output multiple-eavesdropper (MIMOME) channels, under transmit antenna selection (TAS) and BPSK/QPSK modulations. In the main channel between the transmitter and the legitimate receiver, a single transmit antenna is selected to maximizes the instantaneous Signal to Noise Ratio (SNR) at the receiver. At the receiver and the eavesdropper, selection combination (SC) is utilized. By assuming Rayleigh flat fading, we first derive the closed-form approximated expression for the ergodic secrecy rate when the channel state information of the eavesdropper (CSIE) is available at the transmitter. Next, analytical formulas for the approximated and asymptotic secrecy outage probability (SOP) are also developed when CSIE is unavailable. Besides theoretical derivations, simulation results are provided to demonstrate the approximation precision of the derived results. Furthermore, the asymptotic results reveal that the secrecy diversity order degrades into 0 due to the finitealphabet inputs, which is totally different from that driven by the Gaussian inputs.
We investigate whether the current Hubble parameter $H(z)$ measurements could help improve the constraints on dark energy on the basis of the mainstream cosmological probes including the type Ia supernovae (SN) observation, the cosmic microwave background anisotropies (CMB) observation, and the baryon acoustic oscillations (BAO) observation. For the current $H(z)$ data, we use 30 data points measured by using a differential age method. Furthermore, we also consider the future $H(z)$ measurements based on the Sandage-Loeb (SL) test by means of the E-ELT in construction, and thus we also use 30 simulated $H(z)$ data according to a 10-year SL test observation. In this work, we choose four typical dark energy models as examples, i.e., the $Λ$CDM model, the $w$CDM model, the $α$DE model, and the GCG model, to complete the analysis. We find that, when only the current $H(z)$ data are added, the constraints on these models are not improved compared to the cases using the SN+CMB+BAO data; but when further adding the 10-year SL test data, the constraint results are tremendously improved for all the four models. Therefore, we conclude that, although the current $H(z)$ measurements could not provide an evident improvement on the basis of the current mainstream cosmological probes, the future $H(z)$ measurements from the SL test would have enormous potential to change the status of the Hubble parameter measurements in constraining dark energy.
Linear discriminant analysis (LDA) is a powerful tool in building classifiers with easy computation and interpretation. Recent advancements in science technology have led to the popularity of datasets with high dimensions, high orders and complicated structure. Such datasetes motivate the generalization of LDA in various research directions. The R package TULIP integrates several popular high-dimensional LDA-based methods and provides a comprehensive and user-friendly toolbox for linear, semi-parametric and tensor-variate classification. Functions are included for model fitting, cross validation and prediction. In addition, motivated by datasets with diverse sources of predictors, we further include functions for covariate adjustment. Our package is carefully tailored for low storage and high computation efficiency. Moreover, our package is the first R package for many of these methods, providing great convenience to researchers in this area.
Detecting weak clustered signal in spatial data is important but challenging in applications such as medical image and epidemiology. A more efficient detection algorithm can provide more precise early warning, and effectively reduce the decision risk and cost. To date, many methods have been developed to detect signals with spatial structures. However, most of the existing methods are either too conservative for weak signals or computationally too intensive. In this paper, we consider a novel method named Spatial CUSUM (SCUSUM), which employs the idea of the CUSUM procedure and false discovery rate controlling. We develop theoretical properties of the method which indicates that asymptotically SCUSUM can reach high classification accuracy. In the simulation study, we demonstrate that SCUSUM is sensitive to weak spatial signals. This new method is applied to a real fMRI dataset as illustration, and more irregular weak spatial signals are detected in the images compared to some existing methods, including the conventional FDR, FDR$_L$ and scan statistics.
The need to condition distributional properties such as expectation, variance, and entropy arises in algorithmic fairness, model simplification, robustness and many other areas. At face value however, distributional properties are not random variables, and hence conditioning them is a semantic error and type error in probabilistic programming languages. On the other hand, distributional properties are contingent on other variables in the model, change in value when we observe more information, and hence in a precise sense are random variables too. In order to capture the uncertain over distributional properties, we introduce a probability construct -- the random conditional distribution -- and incorporate it into a probabilistic programming language Omega. A random conditional distribution is a higher-order random variable whose realizations are themselves conditional random variables. In Omega we extend distributional properties of random variables to random conditional distributions, such that for example while the expectation a real valued random variable is a real value, the expectation of a random conditional distribution is a distribution over expectations. As a consequence, it requires minimal syntax to encode inference problems over distributional properties, which so far have evaded treatment within probabilistic programming systems and probabilistic modeling in general. We demonstrate our approach case studies in algorithmic fairness and robustness.
Restart strategy helps the covariance matrix adaptation evolution strategy (CMA-ES) to increase the probability of finding the global optimum in optimization, while a single run CMA-ES is easy to be trapped in local optima. In this paper, the continuous non-revisiting genetic algorithm (cNrGA) is used to help CMA-ES to achieve multiple restarts from different sub-regions of the search space. The CMA-ES with on-line search history-assisted restart strategy (HR-CMA-ES) is proposed. The entire on-line search history of cNrGA is stored in a binary space partitioning (BSP) tree, which is effective for performing local search. The frequently sampled sub-region is reflected by a deep position in the BSP tree. When leaf nodes are located deeper than a threshold, the corresponding sub-region is considered a region of interest (ROI). In HR-CMA-ES, cNrGA is responsible for global exploration and suggesting ROI for CMA-ES to perform an exploitation within or around the ROI. CMA-ES restarts independently in each suggested ROI. The non-revisiting mechanism of cNrGA avoids to suggest the same ROI for a second time. Experimental results on the CEC 2013 and 2017 benchmark suites show that HR-CMA-ES performs better than both CMA-ES and cNrGA. A positive synergy is observed by the memetic cooperation of the two algorithms.
There is a growing need for the ability to analyse interval-valued data. However, existing descriptive frameworks to achieve this ignore the process by which interval-valued data are typically constructed; namely by the aggregation of real-valued data generated from some underlying process. In this article we develop the foundations of likelihood based statistical inference for random intervals that directly incorporates the underlying generative procedure into the analysis. That is, it permits the direct fitting of models for the underlying real-valued data given only the random interval-valued summaries. This generative approach overcomes several problems associated with existing methods, including the rarely satisfied assumption of within-interval uniformity. The new methods are illustrated by simulated and real data analyses.
A graph is 1-planar if it can be drawn on a plane so that each edge is crossed by at most one other edge. In this paper, we first give a useful structural theorem for 1-planar graphs, and then apply it to the list edge and list total coloring, the $(p,1)$-total labelling, and the equitable edge coloring of 1-planar graphs. More precisely, we verify the well-known List Edge Coloring Conjecture and List Total Coloring Conjecture for 1-planar graph with maximum degree at least 18, prove that the $(p,1)$-total labelling number of every 1-planar graph $G$ is at most $Δ(G)+2p-2$ provided that $Δ(G)\geq 8p+2$ and $p\geq 2$, and show that every 1-planar graph has an equitable edge coloring with $k$ colors for any integer $k\geq 18$. These three results respectively generalize the main theorems of three different previously published papers.