arXiv Science⌕ Search

arXiv subjects

Jin Zhang

Publications and source records attributed to Jin Zhang.

At least 289 records · Page 16Linked to original sources
arXiv2021

A Value-Function-based Interior-point Method for Non-convex Bi-level Optimization

Bi-level optimization model is able to capture a wide range of complex learning tasks with practical interest. Due to the witnessed efficiency in solving bi-level programs, gradient-based methods have gained popularity in the machine learning community. In this work, we propose a new gradient-based solution scheme, namely, the Bi-level Value-Function-based Interior-point Method (BVFIM). Following the main idea of the log-barrier interior-point scheme, we penalize the regularized value function of the lower level problem into the upper level objective. By further solving a sequence of differentiable unconstrained approximation problems, we consequently derive a sequential programming scheme. The numerical advantage of our scheme relies on the fact that, when gradient methods are applied to solve the approximation problem, we successfully avoid computing any expensive Hessian-vector or Jacobian-vector product. We prove the convergence without requiring any convexity assumption on either the upper level or the lower level objective. Experiments demonstrate the efficiency of the proposed BVFIM on non-convex bi-level problems.

math.OC↗
arXiv2021

Multiobjective Bilevel Evolutionary Approach for Off-Grid Direction-of-Arrival Estimation

The source number identification is an essential step in direction-of-arrival (DOA) estimation. Existing methods may provide a wrong source number due to inferior statistical properties (in low SNR or limited snapshots) or modeling errors (caused by relaxing sparse penalties), especially in impulsive noise. To address this issue, we propose a novel idea of simultaneous source number identification and DOA estimation. We formulate a multiobjective off-grid DOA estimation model to realize this idea, by which the source number can be automatically identified together with DOA estimation. In particular, the source number is properly exploited by the $l_0$ norm of impinging signals without relaxations, guaranteeing accuracy. Furthermore, we design a multiobjective bilevel evolutionary algorithm to solve the proposed model. The source number identification and sparse recovery are simultaneously optimized at the on-grid (lower) level. A forward search strategy is developed to further refine the grid at the off-grid (upper) level. This strategy does not need linear approximations and can eliminate the off-grid gap with low computational complexity. Simulation results demonstrate the outperformance of our method in terms of source number and root mean square error.

cs.NE↗
arXiv2021

Gridless Evolutionary Approach for Line Spectral Estimation with Unknown Model Order

Gridless methods show great superiority in line spectral estimation. These methods need to solve an atomic $l_0$ norm (i.e., the continuous analog of $l_0$ norm) minimization problem to estimate frequencies and model order. Since this problem is NP-hard to compute, relaxations of atomic $l_0$ norm, such as nuclear norm and reweighted atomic norm, have been employed for promoting sparsity. However, the relaxations give rise to a resolution limit, subsequently leading to biased model order and convergence error. To overcome the above shortcomings of relaxation, we propose a novel idea of simultaneously estimating the frequencies and model order by means of the atomic $l_0$ norm. To accomplish this idea, we build a multiobjective optimization model. The measurment error and the atomic $l_0$ norm are taken as the two optimization objectives. The proposed model directly exploits the model order via the atomic $l_0$ norm, thus breaking the resolution limit. We further design a variable-length evolutionary algorithm to solve the proposed model, which includes two innovations. One is a variable-length coding and search strategy. It flexibly codes and interactively searches diverse solutions with different model orders. These solutions act as steppingstones that help fully exploring the variable and open-ended frequency search space and provide extensive potentials towards the optima. Another innovation is a model order pruning mechanism, which heuristically prunes less contributive frequencies within the solutions, thus significantly enhancing convergence and diversity. Simulation results confirm the superiority of our approach in both frequency estimation and model order selection.

math.OC↗
arXiv2021

Parallel Scale-wise Attention Network for Effective Scene Text Recognition

The paper proposes a new text recognition network for scene-text images. Many state-of-the-art methods employ the attention mechanism either in the text encoder or decoder for the text alignment. Although the encoder-based attention yields promising results, these schemes inherit noticeable limitations. They perform the feature extraction (FE) and visual attention (VA) sequentially, which bounds the attention mechanism to rely only on the FE final single-scale output. Moreover, the utilization of the attention process is limited by only applying it directly to the single scale feature-maps. To address these issues, we propose a new multi-scale and encoder-based attention network for text recognition that performs the multi-scale FE and VA in parallel. The multi-scale channels also undergo regular fusion with each other to develop the coordinated knowledge together. Quantitative evaluation and robustness analysis on the standard benchmarks demonstrate that the proposed network outperforms the state-of-the-art in most cases.

cs.CV↗
arXiv2021

Scalar one-loop 4-point integral with one massless vertex in loop regularization

The scalar one-loop 4-point function with one massless vertex is evaluated analytically by employing the loop regularization method. According to the method a characteristic scale $μ_{s}$ is introduced to regularize the divergent integrals. The infrared divergent parts, which take the form of $\ln^{2}(λ^{2}/μ^{2}_{s})$ and $\ln(λ^{2}/μ^{2}_{s})$ as $μ_{s}\rightarrow 0$ where $λ$ is a constant and expressed in terms of masses and Mandelstam variables, and the infrared stable parts are well separated. The result is shown explicitly via $44$ dilogarithms in the kinematic sector in which our evaluation is valid.

hep-ph↗
arXiv2021

Temporal Consistency Two-Stream CNN for Human Motion Prediction

Fusion is critical for a two-stream network. In this paper, we propose a novel temporal fusion (TF) module to fuse the two-stream joints' information to predict human motion, including a temporal concatenation and a reinforcement trajectory spatial-temporal (TST) block, specifically designed to keep prediction temporal consistency. In particular, the temporal concatenation keeps the temporal consistency of preliminary predictions from two streams. Meanwhile, the TST block improves the spatial-temporal feature coupling. However, the TF module can increase the temporal continuities between the first predicted pose and the given poses and between each predicted pose. The fusion is based on a two-stream network that consists of a dynamic velocity stream (V-Stream) and a static position stream (P-Stream) because we found that the joints' velocity information improves the short-term prediction, while the joints' position information is better at long-term prediction, and they are complementary in motion prediction. Finally, our approach achieves impressive results on three benchmark datasets, including H3.6M, CMU-Mocap, and 3DPW in both short-term and long-term predictions, confirming its effectiveness and efficiency.

cs.CV↗
arXiv2021

Highly Variable $γ$-Ray Emission of CTD 135 and Implications for its Compact Symmetric Structure

The $γ$-ray emission properties of CTD 135, a typical compact symmetric object (CSO), are investigated with $\sim$11-year Fermi/LAT observations. We show that it has bright and significantly variable GeV emission, with the $γ$-ray luminosity of $L_γ\sim10^{47}$ erg s$^{-1}$ and a variation index of TS$_{\rm var}=1002$. A quasi-periodic oscillation (QPO) with a periodicity of $\sim$460 days is detected in the global 95\% false-alarm level. These $γ$-ray emission features are similar to that of blazars. Its broadband spectral energy distribution (SED) can be attributed to the radiations of the relativistic electrons accelerated in the core region and the extended region. The SED modeling shows that the $γ$-rays are from the core region, which has a Doppler boosting factor of $δ\sim10.8$ and relativistically moves with a small viewing angle, being similar to blazar jets. On the base of the analysis results, we propose that the episodic activity of the central engine in CTD 135 results in a blazar-like jet and the bubble-like lobes as the Fermi bubbles in the Galaxy. The strong $γ$-ray emission with obvious variability is from the jet radiations and the symmetric radio structure is attributed to the bubbles. The jet radiation power and disk luminosity in units of Eddington luminosity of CTD 135 follow the same relation as other young radio galaxies, indicating that its jet radiation may be also driven by the Eddington ratio.

astro-ph.HE↗
arXiv2021

Identification of the Mott insulating CDW state in 1T-TaS$_2$

We investigate the low-temperature charge-density-wave (CDW) state of bulk TaS$_2$ with a fully self-consistent DFT+U approach, over which the controversy has remained unresolved regarding the out-of-plane metallic band. By examining the innate structure of the Hubbard U potential, we reveal that the conventional use of atomic-orbital basis could seriously misevaluate the electron correlation in the CDW state. By adopting a generalized basis, covering the whole David star, we successfully reproduce the Mott insulating nature with the layer-by-layer antiferromagnetic order. Similar consideration should be applied for description of the electron correlation in molecular solid.

cond-mat.str-el↗
arXiv2021

Laser Driven Fluorescence Emission in a Nitrogen Gas Jet at 100 MHz Repetition Rate

We report the fluorescence emission which is driven by femtosecond laser pulses with a repetition rate of 100 MHz and a center wavelength of 1040 nm in a nitrogen gas jet. The experiment is performed in a femtosecond enhancement cavity coupled with high repetition rate laser for the first time to the best of our knowledge. In contrast to previous observation at low repetition rate with a nitrogen gas jet, where the 391 nm radiation was observed but the 337 nm emission was missing, the 337 nm emission is 3 times stronger than the 391 nm emission in our experiment. By examining the dependence of the radiation intensity on the flow rate of the nitrogen gas and the polarization of the pump pulse, the formation mechanism of the N2(C3Πu) triplet excited state, i.e., the upper state of the 337 nm emission, is investigated. We attribute the main excitation process to the inelastic collision excitation process, and exclude the possibility of the dissociative recombination as the dominate pathway. The role of the steady state plasma that is generated under our experimental conditions is also discussed.

physics.atom-ph↗
arXiv2021

Supercloseness of finite element method on a Bakhvalov-type mesh for a singularly perturbed problem with two parameters

In this paper, the linear finite element method on a Bakhvalov-type mesh is applied to a singularly perturbed problem with two parameters. The solution of the problem exists two exponential boundary layers. A new interpolation, which is simple in construction and analysis, is introduced for convergence analysis. Furthermore, we find a subtle relationship between the Bakhvalov-type mesh itself and the weaker exponential layer and obtain an interesting result. Finally, we prove a supercloseness result between the Lagrange interpolation and the numerical solution. Numerical tests confirm our theoretical results.

math.NA↗
arXiv2020

Calibration of the instrumental polarization effects of SCExAO-CHARIS' spectropolarimetric mode

SCExAO at the Subaru telescope is a visible and near-infrared high-contrast imaging instrument employing extreme adaptive optics and coronagraphy. The instrument feeds the near-infrared light (JHK) to the integral field spectrograph CHARIS. Recently, a Wollaston prism was added to CHARIS' optical path, giving CHARIS a spectropolarimetric capability that is unique among high-contrast imaging instruments. We present a detailed Mueller matrix model describing the instrumental polarization effects of the complete optical path, thus the telescope and instrument. The 22 wavelength bins of CHARIS provide a unique opportunity to investigate in detail the wavelength dependence of the instrumental polarization effects. From measurements with the internal light source, we find that the image derotator (K-mirror) produces strong wavelength-dependent crosstalk, in the worst case converting ~95% of the incident linear polarization to circularly polarized light that cannot be measured. Theoretical calculations show that the magnitude of the instrumental polarization of the telescope varies with wavelength between approximately 0.5% and 0.7%, and that its angle is exactly equal to the altitude angle of the telescope. We plan to more accurately determine the instrumental polarization of the telescope with observations of a polarization standard star, and fit more comprehensive physical models to all experimental data. In addition, we plan to integrate the complete Mueller matrix model into the existing CHARIS post-processing pipeline, with the aim to achieve a polarimetric accuracy of <0.1% in the degree of linear polarization. Our calibrations of CHARIS' spectropolarimetric mode will enable unique quantitative polarimetric studies of circumstellar disks and planetary and brown dwarf companions.

astro-ph.IM↗
arXiv2020

Finite element method for singularly perturbed problems with two parameters on a Bakhvalov-type mesh in 2D

For a singularly perturbed elliptic model problem with two small parameters, we analyze finite element methods of any order on a Bakhvalov-type mesh. For convergence analysis, we construct a new interpolation by using the characteristics of layers. Besides, a more subtle analysis of the mesh scale near the exponential layer is carried out. Based on the interpolation and new analysis of the mesh scale, we prove the optimal convergence order.

math.NA↗
arXiv2020

Convergence and supercloseness in a balanced norm of finite element methods on Bakhvalov-type meshes for reaction-diffusion problems

In convergence analysis of finite element methods for singularly perturbed reaction--diffusion problems, balanced norms have been successfully introduced to replace standard energy norms so that layers can be captured. In this article, we focus on the convergence analysis in a balanced norm on Bakhvalov-type rectangular meshes. In order to achieve our goal, a novel interpolation operator, which consists of a local weighted $L^2$ projection operator and the Lagrange interpolation operator, is introduced for a convergence analysis of optimal order in the balanced norm. The analysis also depends on the stabilities of the $L^2$ projection and the characteristics of Bakhvalov-type meshes. Furthermore, we obtain a supercloseness result in the balanced norm, which appears in the literature for the first time. This result depends on another novel interpolant, which consists of the local weighted $L^2$ projection operator, a vertices-edges-element operator and some corrections on the boundary.

math.NA↗
arXiv2020

Integrated Plasmonics: Broadband Dirac Plasmons in Borophene

The past decade has witnessed numerous discoveries of two-dimensional (2D) semimetals and insulators, whereas 2D metals are rarely identified. Borophene, a monolayer boron sheet, has recently emerged as a perfect 2D metal with unique structure and electronic properties. Here we study collective excitations in borophene, which exhibit two major plasmon modes with low damping rates extending from infrared to ultraviolet regime. The anisotropic 1D plasmon originates from electronic excitations of tilted Dirac cones in borophene, analogous to that in heavily doped Dirac semimetals. These features make borophene promising to realize directional polariton transportation and broadband optical communications for next-generation optoelectronic devices.

cond-mat.mtrl-sci↗
arXiv2020

Research on a New Convolutional Neural Network Model Combined with Random Edges Adding

It is always a hot and difficult point to improve the accuracy of convolutional neural network model and speed up its convergence. Based on the idea of small world network, a random edge adding algorithm is proposed to improve the performance of convolutional neural network model. This algorithm takes the convolutional neural network model as a benchmark, and randomizes backwards and cross-layer connections with probability p to form a new convolutional neural network model. The proposed idea can optimize the cross layer connectivity by changing the topological structure of convolutional neural network, and provide a new idea for the improvement of the model. The simulation results based on Fashion-MINST and cifar10 data set show that the model recognition accuracy and training convergence speed are greatly improved by random edge adding reconstructed models with aprobability p = 0.1.

cs.NE↗
arXiv2020

Nonlinear twistoptics at symmetry-broken interfaces

Broken symmetries induce strong nonlinear optical responses in materials and at interfaces. Twist angle can give complete control over the presence or lack of inversion symmetry at a crystal interface, and is thus an appealing knob for tuning nonlinear optical systems. In contrast to conventional nonlinear crystals with rigid lattices, the weak interlayer coupling in van der Waals (vdW) heterostructures allows for arbitrary selection of twist angle, making nanomechanical manipulation of fundamental interfacial symmetry possible within a single device. Here we report highly tunable second harmonic generation (SHG) from nanomechanically rotatable stacks of bulk hexagonal boron nitride (BN) crystals, and introduce the term twistoptics to describe studies of optical properties in dynamically twistable vdW systems. We observe SHG intensity modulated by a factor of more than 50, polarization patterns determined by moiré interface symmetry, and enhanced conversion efficiency for bulk crystals by stacking multiple pieces of BN joined by symmetry-broken interfaces. Our study provides a foundation for compact twistoptics architectures aimed at efficient, scalable, and tunable frequency-conversion, and demonstrates SHG as a robust probe of buried vdW interfaces.

cond-mat.mes-hall↗
arXiv2020

A Novel Emergency Light Based Smart Building Solution: Design, Implementation and Use Cases

Deployment of Internet of Things (IoT) in smart buildings has received considerable interest from both the academic community and commercial sectors. Unfortunately, widespread adoption of current smart building solutions is inhibited by the high costs associated with installation and maintenance. Moreover, different types of IoT devices from different manufacturers typically form distinct networks and data silos. There is a need to use a common backbone network that facilitates interoperability and seamless data exchange in a uniform way. In this paper, we present EMIoT, a novel solution for smart buildings that breaks these barriers by leveraging existing emergency lighting systems. In EMIoT, we embed a wireless LoRa module in each emergency light to turn them into wireless routers. EMIoT has been deployed in more than 50 buildings of different types in Sydney Australia and has been successfully running over two years. We present the design and implementation of EMIoT in this paper. Moreover, we use the deployment in a residential building as a use case to show the performance of EMIoT in real-world environments and share lessons learned. Finally, we discuss the advantages and disadvantages of EMIoT. This paper provides practical insights for IoT deployment in smart buildings for practitioners and solution providers.

cs.NI↗