arXiv subjects

Chao Zhang

Publications and source records attributed to Chao Zhang.

At least 1,081 records · Page 60Linked to original sources

arXiv2018

Cross-domain Deep Feature Combination for Bird Species Classification with Audio-visual Data

In recent decade, many state-of-the-art algorithms on image classification as well as audio classification have achieved noticeable successes with the development of deep convolutional neural network (CNN). However, most of the works only exploit single type of training data. In this paper, we present a study on classifying bird species by exploiting the combination of both visual (images) and audio (sounds) data using CNN, which has been sparsely treated so far. Specifically, we propose CNN-based multimodal learning models in three types of fusion strategies (early, middle, late) to settle the issues of combining training data cross domains. The advantage of our proposed method lies on the fact that We can utilize CNN not only to extract features from image and audio data (spectrogram) but also to combine the features across modalities. In the experiment, we train and evaluate the network structure on a comprehensive CUB-200-2011 standard data set combing our originally collected audio data set with respect to the data species. We observe that a model which utilizes the combination of both data outperforms models trained with only an either type of data. We also show that transfer learning can significantly increase the classification performance.

cs.CV↗

arXiv2018

Characterization of temperatures associated to Schrödinger operators with initial data in Morrey spaces

Let $\mathcal{L}$ be a Schrödinger operator of the form $\mathcal{L} = -Δ+V$ acting on $L^2(\mathbb R^n)$ where the nonnegative potential $V$ belongs to the reverse Hölder class $B_q$ for some $q\geq n.$ Let $L^{p,λ}(\mathbb{R}^{n})$, $0\le λ<n$ denote the Morrey space on $\mathbb{R}^{n}$. In this paper, we will show that a function $f\in L^{2,λ}(\mathbb{R}^{n})$ is the trace of the solution of ${\mathbb L}u=u_{t}+{\mathcal{L}}u=0, u(x,0)= f(x),$ where $u$ satisfies a Carleson-type condition \begin{eqnarray*} \sup_{x_B, r_B} r_B^{-λ}\int_0^{r_B^2}\int_{B(x_B, r_B)} |\nabla u(x,t)|^2 {dx dt} \leq C <\infty. \end{eqnarray*} Conversely, this Carleson-type condition characterizes all the ${\mathbb L}$-carolic functions whose traces belong to the Morrey space $L^{2,λ}(\mathbb{R}^{n})$ for all $0\le λ<n$. This result extends the analogous characterization founded by Fabes and Neri for the classical BMO space of John and Nirenberg.

math.AP↗

arXiv2018

Generalization Bounds for Vicinal Risk Minimization Principle

The vicinal risk minimization (VRM) principle, first proposed by \citet{vapnik1999nature}, is an empirical risk minimization (ERM) variant that replaces Dirac masses with vicinal functions. Although there is strong numerical evidence showing that VRM outperforms ERM if appropriate vicinal functions are chosen, a comprehensive theoretical understanding of VRM is still lacking. In this paper, we study the generalization bounds for VRM. Our results support Vapnik's original arguments and additionally provide deeper insights into VRM. First, we prove that the complexity of function classes convolving with vicinal functions can be controlled by that of the original function classes under the assumption that the function class is composed of Lipschitz-continuous functions. Then, the resulting generalization bounds for VRM suggest that the generalization performance of VRM is also effected by the choice of vicinity function and the quality of function classes. These findings can be used to examine whether the choice of vicinal function is appropriate for the VRM-based learning setting. Finally, we provide a theoretical explanation for existing VRM models, e.g., uniform distribution-based models, Gaussian distribution-based models, and mixup models.

cs.LG↗

arXiv2018

Angular momentum loss for eccentric compact binary in screened modified gravity

Gravitational wave (GW) observations provide insight into the gravity regime. These observations relies on the comparison of the data to the GWs model, which depends on the orbital evolution of the binary systems. In this paper, we study the orbital evolution of eccentric compact binaries within screened modified gravity, which is a kind of scalar-tensor theory with screening mechanisms. According to the Noether's theorem, from the theory's GW action, we compute the angular momentum flux both as a function of the fields in the theory and as a function of the multipole moments of a N-body system. We specialize to the orbital parameters of eccentric compact binaries to express its orbital energy and angular momentum decay, and then derive the decay rates of its orbital eccentricity and semimajor axis. We find that these quantities decay faster than in General Relativity due to the presence of dipole radiation.

gr-qc↗

arXiv2018

Joint Sub-bands Learning with Clique Structures for Wavelet Domain Super-Resolution

Convolutional neural networks (CNNs) have recently achieved great success in single-image super-resolution (SISR). However, these methods tend to produce over-smoothed outputs and miss some textural details. To solve these problems, we propose the Super-Resolution CliqueNet (SRCliqueNet) to reconstruct the high resolution (HR) image with better textural details in the wavelet domain. The proposed SRCliqueNet firstly extracts a set of feature maps from the low resolution (LR) image by the clique blocks group. Then we send the set of feature maps to the clique up-sampling module to reconstruct the HR image. The clique up-sampling module consists of four sub-nets which predict the high resolution wavelet coefficients of four sub-bands. Since we consider the edge feature properties of four sub-bands, the four sub-nets are connected to the others so that they can learn the coefficients of four sub-bands jointly. Finally we apply inverse discrete wavelet transform (IDWT) to the output of four sub-nets at the end of the clique up-sampling module to increase the resolution and reconstruct the HR image. Extensive quantitative and qualitative experiments on benchmark datasets show that our method achieves superior performance over the state-of-the-art methods.

cs.CV↗

arXiv2018

Small-Deviation Inequalities for Sums of Random Matrices

Random matrices have played an important role in many fields including machine learning, quantum information theory and optimization. One of the main research focuses is on the deviation inequalities for eigenvalues of random matrices. Although there are intensive studies on the large-deviation inequalities for random matrices, only a few of works discuss the small-deviation behavior of random matrices. In this paper, we present the small-deviation inequalities for the largest eigenvalues of sums of random matrices. Since the resulting inequalities are independent of the matrix dimension, they are applicable to the high-dimensional and even the infinite-dimensional cases.

math.PR↗

arXiv2018

The Importance of Norm Regularization in Linear Graph Embedding: Theoretical Analysis and Empirical Demonstration

Learning distributed representations for nodes in graphs is a crucial primitive in network analysis with a wide spectrum of applications. Linear graph embedding methods learn such representations by optimizing the likelihood of both positive and negative edges while constraining the dimension of the embedding vectors. We argue that the generalization performance of these methods is not due to the dimensionality constraint as commonly believed, but rather the small norm of embedding vectors. Both theoretical and empirical evidence are provided to support this argument: (a) we prove that the generalization error of these methods can be bounded by limiting the norm of vectors, regardless of the embedding dimension; (b) we show that the generalization performance of linear graph embedding methods is correlated with the norm of embedding vectors, which is small due to the early stopping of SGD and the vanishing gradients. We performed extensive experiments to validate our analysis and showcased the importance of proper norm regularization in practice.

cs.LG↗

arXiv2018

Waveforms of compact binary inspiral gravitational radiation in screened modified gravity

Scalar-tensor gravity, with the screening mechanisms to avoid the severe constraints of the fifth force in the Solar System, can be described with a unified theoretical framework, the so-called screened modified gravity (SMG). Within this framework, in this paper we calculate the waveforms of gravitational-waves (GWs) emitted by inspiral compact binaries, which include four polarization modes, the plus $h_{+}$, cross $h_{\times}$, breathing $h_{b}$, and longitudinal $h_{L}$ modes. The scalar polarizations $h_b$ and $h_L$ are both caused by the scalar field of SMG, and satisfy a simple linear relation. With the stationary phase approximations, we obtain their Fourier transforms, and derive the correction terms in the amplitude, phase, and polarizations of GWs, relative to the corresponding results in general relativity. The corresponding parametrized post-Einsteinian parameters in the general SMG are also identified. Imposing the noise level of the ground-based Einstein Telescope, we find that GW detections from inspiral compact binaries composed of a neutron star and a black hole can place stringent constraints on the sensitivities of neutron stars, and the bound is applicable to any SMG theory. Finally, we apply these results to some specific theories of SMG, including chameleon, symmetron, dilaton and $f(R)$.

gr-qc↗

arXiv2018

Weakly-Supervised Neural Text Classification

Deep neural networks are gaining increasing popularity for the classic text classification task, due to their strong expressive power and less requirement for feature engineering. Despite such attractiveness, neural text classification models suffer from the lack of training data in many real-world applications. Although many semi-supervised and weakly-supervised text classification models exist, they cannot be easily applied to deep neural models and meanwhile support limited supervision types. In this paper, we propose a weakly-supervised method that addresses the lack of training data in neural text classification. Our method consists of two modules: (1) a pseudo-document generator that leverages seed information to generate pseudo-labeled documents for model pre-training, and (2) a self-training module that bootstraps on real unlabeled data for model refinement. Our method has the flexibility to handle different types of weak supervision and can be easily integrated into existing deep neural models for text classification. We have performed extensive experiments on three real-world datasets from different domains. The results demonstrate that our proposed method achieves inspiring performance without requiring excessive training data and outperforms baseline methods significantly.

cs.IR↗

arXiv2018

Blur-Countering Keypoint Detection via Eigenvalue Asymmetry

Well-known corner or local extrema feature based detectors such as FAST and DoG have achieved noticeable successes. However, detecting keypoints in the presence of blur has remained to be an unresolved issue. As a matter of fact, various kinds of blur (e.g., motion blur, out-of-focus, and space-variant) remarkably increase challenges for keypoint detection. As a result, those methods have limited performance. To settle this issue, we propose a blur-countering method for detecting valid keypoints for various types and degrees of blurred images. Specifically, we first present a distance metric for derivative distributions, which preserves the distinctiveness of patch pairs well under blur. We then model the asymmetry by utilizing the difference of squared eigenvalues based on the distance metric. To make it scale-robust, we also extend it to scale space. The proposed detector is efficient as the main computational cost is the square of derivatives at each pixel. Extensive visual and quantitative results show that our method outperforms current approaches under different types and degrees of blur. Without any parallelization, our implementation\footnote{We will make our code publicly available upon the acceptance.} achieves real-time performance for low-resolution images (e.g., $320\times240$ pixel).

cs.CV↗

arXiv2018

Charge compensation at the interface between the polar NaCl(111) surface and a NaCl aqueous solution

Periodic supercell models of electric double layers formed at the interface between a charged surface and an electrolyte are subject to serious finite size errors and require certain adjustments in the treatment of the long-range electrostatic interactions. In a previous publication (C. Zhang, M. Sprik, Phys. Rev. B 94, 245309 (2016)) we have shown how this can be achieved using finite field methods. The test system was the familiar simple point charge model of a NaCl aqueous solution confined between two oppositely charged walls. Here this method is extended to the interface between the (111) polar surface of a NaCl crystal and a high concentration NaCl aqueous solution. The crystal is kept completely rigid and the compensating charge screening the polarization can only be provided by the electrolyte. We verify that the excess electrolyte ionic charge at the interface conforms to the Tasker 1/2 rule for compensating charge in the theory of polar rocksalt (111) surfaces. The interface can be viewed as an electric double layer with a net charge. We define a generalized Helmholtz capacitance $C_\text{H}$ which can be computed by varying the applied electric field. We find $C_\text{H} = 8.23 \, μ\mathrm{Fcm}^{-2}$, which should be compared to the $4.23 \, μ\mathrm{Fcm}^{-2}$ for the (100) non-polar surface of the same NaCl crystal. This is rationalized by the observation that compensating ions shed their first solvation shell adsorbing as contact ions pairs on the polar surface.

physics.chem-ph↗

arXiv2018

Finite electric displacement simulations of polar ionic solid-electrolyte interfaces: Application to NaCl(111)/aqueous NaCl solution

Tasker type III polar terminations of ionic crystals carry a net surface charge as well as a dipole moment and are fundamentally unstable. In contact with electrolytes, such polar surfaces can be stabilized by adsorption of counter ions from solution to form electric double layers (EDLs). In a previous work (J. Chem. Phys 147, 104702 (2017)) we reported on a classical force field based molecular dynamics study of a prototype model system namely a NaCl(111) slab interfaced with an aqueous NaCl solution on both sides. A serious hurdle in the simulation is that the finite width of the slab admits an electric field in the solid perturbing the theoretical charge balance at the interface of semi-infinite systems (half the surface charge density for NaCl(111)). It was demonstrated that the application of a finite macroscopic field $E$ cancelling the internal electric field can recover the correct charge compensation at the interface. In the present work, we expand this method by applying a conjugate electric displacement field $D$. The benefits of using $D$ instead of $E$ as the control variable are two fold: it does not only speed up the convergence of the polarization in the simulation but also leads to a succinct expression for the biasing displacement field involving only structural parameters which are known in advance. This makes it feasible to study the charge compensating phenomenon of this prototype system with density functional theory based molecular dynamics (DFTMD), as shown in this work.

physics.chem-ph↗

arXiv2018

Factorizable Net: An Efficient Subgraph-based Framework for Scene Graph Generation

Generating scene graph to describe all the relations inside an image gains increasing interests these years. However, most of the previous methods use complicated structures with slow inference speed or rely on the external data, which limits the usage of the model in real-life scenarios. To improve the efficiency of scene graph generation, we propose a subgraph-based connection graph to concisely represent the scene graph during the inference. A bottom-up clustering method is first used to factorize the entire scene graph into subgraphs, where each subgraph contains several objects and a subset of their relationships. By replacing the numerous relationship representations of the scene graph with fewer subgraph and object features, the computation in the intermediate stage is significantly reduced. In addition, spatial information is maintained by the subgraph features, which is leveraged by our proposed Spatial-weighted Message Passing~(SMP) structure and Spatial-sensitive Relation Inference~(SRI) module to facilitate the relationship recognition. On the recent Visual Relationship Detection and Visual Genome datasets, our method outperforms the state-of-the-art method in both accuracy and speed.

cs.CV↗

arXiv2018

Computing the Helmholtz Capacitance of Charged Insulator-Electrolyte Interfaces from the Supercell Polarization

Supercell modelling of an electrical double layer (EDL) at electrified solid-electrolyte interfaces is a challenge. The net polarization of EDLs arising from the fixed chemical composition setup leads to uncompensated EDLs under periodic boundary condition and convolutes the calculation of the Helmholtz capacitance [Zhang and Sprik, Phys. Rev. B, 94, 245309 (2016)]. Here we provide a new formula based on the supercell polarization at zero electric field $\bar{E} = 0$ (i.e. standard Ewald boundary condition) to calculate the Helmholtz capacitance of charged insulator-electrolyte interfaces and validate it using atomistic simulations. Results are shown to be independent of the supercell size. This formula gives a shortcut to compute the Helmholtz capacitance without locating the zero net charge state of EDL and applies directly to any standard molecular dynamics code where the electrostatic interactions are treated by the Ewald summation or its variants.

cond-mat.soft↗

arXiv2018

Easing Embedding Learning by Comprehensive Transcription of Heterogeneous Information Networks

Heterogeneous information networks (HINs) are ubiquitous in real-world applications. In the meantime, network embedding has emerged as a convenient tool to mine and learn from networked data. As a result, it is of interest to develop HIN embedding methods. However, the heterogeneity in HINs introduces not only rich information but also potentially incompatible semantics, which poses special challenges to embedding learning in HINs. With the intention to preserve the rich yet potentially incompatible information in HIN embedding, we propose to study the problem of comprehensive transcription of heterogeneous information networks. The comprehensive transcription of HINs also provides an easy-to-use approach to unleash the power of HINs, since it requires no additional supervision, expertise, or feature engineering. To cope with the challenges in the comprehensive transcription of HINs, we propose the HEER algorithm, which embeds HINs via edge representations that are further coupled with properly-learned heterogeneous metrics. To corroborate the efficacy of HEER, we conducted experiments on two large-scale real-words datasets with an edge reconstruction task and multiple case studies. Experiment results demonstrate the effectiveness of the proposed HEER model and the utility of edge representations and heterogeneous metrics. The code and data are available at https://github.com/GentleZhu/HEER.

cs.SI↗

arXiv2018

Beam energy dependence of squeeze-out effect on the directed and elliptic flow in Au+Au collisions at high baryon density region

We present a detailed analysis of the beam energy dependence of the mechanisms for the generation of directed and elliptic flows in Au+Au collisions focusing on the role of hadronic rescattering and spectator shadowing within a microscopic transport model JAM with different equation of state. A systematic study of the beam energy dependence is performed for Au+Au collisions at $\sqrt{s_{NN}} =2.3 - 62.4$ GeV. The transition of the dynamical origin of the directed flow is observed. We find that the initial Glauber type nucleon-nucleon collisions generate negative $v_1$ for nucleons at midrapidity due to the presence of spectator matter, and this negative nucleon $v_1$ is turned to be positive by the meson-baryon interactions at the beam energy region of $\sqrt{s_{NN}} < 30$ GeV. In contrast, above 30 GeV there is no spectator shadowing at midrapidity, and initial nucleon-nucleon collisions do not generate directed flow, but subsequent rescatterings among produced particles generate negative $v_1$ for nucleons. It is demonstrated that negative pion-directed flows are mostly generated by the interaction with the spectator matter. It is also shown that the squeeze-out effect is largely suppressed in the case of softening, which leads to the enhancement of elliptic flow around $\sqrt{s_{NN}}=5-7$ GeV. The elliptic flow at midrapidity above 10 GeV is not influenced by the squeeze-out due to spectator matter, while its effect is seen at the forward rapidity range of $y/y_\mathrm{c.m.}>0.5$, which decreases as beam energy increases.

nucl-ex↗

arXiv2018

Semi-tied Units for Efficient Gating in LSTM and Highway Networks

Gating is a key technique used for integrating information from multiple sources by long short-term memory (LSTM) models and has recently also been applied to other models such as the highway network. Although gating is powerful, it is rather expensive in terms of both computation and storage as each gating unit uses a separate full weight matrix. This issue can be severe since several gates can be used together in e.g. an LSTM cell. This paper proposes a semi-tied unit (STU) approach to solve this efficiency issue, which uses one shared weight matrix to replace those in all the units in the same layer. The approach is termed "semi-tied" since extra parameters are used to separately scale each of the shared output values. These extra scaling factors are associated with the network activation functions and result in the use of parameterised sigmoid, hyperbolic tangent, and rectified linear unit functions. Speech recognition experiments using British English multi-genre broadcast data showed that using STUs can reduce the calculation and storage cost by a factor of three for highway networks and four for LSTMs, while giving similar word error rates to the original models.

cs.CL↗

arXiv2018

Evidence of deviations from general relativity in binary pulsars?

Testing gravitational theories by binary pulsars nowadays becomes a key issue. For the general screened modified gravity (SMG), the post-Keplerian parameters in the neutron star (NS) - white dwarf (WD) binaries differ from those of general relativity (GR), and the differences are quantified by the scalar charge $ε_{\scriptscriptstyle\rm WD}$ of WD. After deriving the constraints on $ε_{\scriptscriptstyle\rm WD}$ from four different NS-WD binaries, we find that $ε_{\scriptscriptstyle\rm WD}$ is different from zero at the 2$σ$ level in all the cases studied, and there exists an inverse correlation between masses and scalar charges of low-mass WDs, which is consistent with the screening mechanisms. In particular, two independent binaries with measured radii of WDs follow the coincident constraints on the vacuum expectation value of the scalar field. These self-consistent results indicate that the observations in NS-WD binary pulsars seem in favor of SMG, rather than GR.

gr-qc↗