Search arXiv⌕ Search

arXiv subjects

Kai Zhang

Publications and source records attributed to Kai Zhang.

At least 559 records · Page 31Linked to original sources
arXiv2021

Convolutional neural networks for fluid flow analysis: toward effective metamodeling and low-dimensionalization

We focus on a convolutional neural network (CNN), which has recently been utilized for fluid flow analyses, from the perspective on the influence of various operations inside it by considering some canonical regression problems with fluid flow data. We consider two types of CNN-based fluid flow analyses; 1. CNN metamodeling and 2. CNN autoencoder. For the first type of CNN with additional scalar inputs, which is one of the common forms of CNN for fluid flow analysis, we investigate the influence of input placements in the CNN training pipeline. As an example, estimation of force coefficients of an inclined flat plate and two side-by-side cylinders in laminar flows is considered. We find that care should be taken for the placement of additional scalar inputs depending on the problem setting and the complexity of flows that users handle. We then discuss the influence of various parameters and operations on the CNN performance, with the utilization of autoencoder (AE). A two-dimensional turbulence is considered for the demonstration of AE. The results of AE highly rely on the decaying nature. Investigation on the influence of padding operation at a convolutional layer is also performed. The zero padding shows reasonable ability compared to other methods which account for the boundary conditions assumed in the numerical data. Moreover, the effect of the dimensional reduction/extension methods inside CNN is also examined. The CNN model is robust against the difference in dimension reduction operations, while it is sensitive to the dimensional extension methods. The findings of this paper will help us better design a CNN architecture for practical fluid flow analysis

physics.flu-dyn↗
arXiv2021

Switching magnon chirality in artificial antiferromagnet

Magnons in antiferromagnets can support both right-handed and left-handed chiralities, which shed a light on the chirality-based spintronics. Here we demonstrate the switching and reading of magnon chirality in an artificial antiferromagnet. The coexisting antiferromagnetic and ferromagnetic characteristic resonance modes are discovered, which permits a high tunability in the modulation of magnon chirality. The reading of the chirality is accomplished via the chirality-dependent spin pumping as well as spin rectification effect. Our result illustrates an ideal antiferromagnetic platform for handling magnon chirality and paves the way for chirality-based spintronics.

physics.app-ph↗
arXiv2021

Plug-and-Play Image Restoration with Deep Denoiser Prior

Recent works on plug-and-play image restoration have shown that a denoiser can implicitly serve as the image prior for model-based methods to solve many inverse problems. Such a property induces considerable advantages for plug-and-play image restoration (e.g., integrating the flexibility of model-based method and effectiveness of learning-based methods) when the denoiser is discriminatively learned via deep convolutional neural network (CNN) with large modeling capacity. However, while deeper and larger CNN models are rapidly gaining popularity, existing plug-and-play image restoration hinders its performance due to the lack of suitable denoiser prior. In order to push the limits of plug-and-play image restoration, we set up a benchmark deep denoiser prior by training a highly flexible and effective CNN denoiser. We then plug the deep denoiser prior as a modular part into a half quadratic splitting based iterative algorithm to solve various image restoration problems. We, meanwhile, provide a thorough analysis of parameter setting, intermediate results and empirical convergence to better understand the working mechanism. Experimental results on three representative image restoration tasks, including deblurring, super-resolution and demosaicing, demonstrate that the proposed plug-and-play image restoration with deep denoiser prior not only significantly outperforms other state-of-the-art model-based methods but also achieves competitive or even superior performance against state-of-the-art learning-based methods. The source code is available at https://github.com/cszn/DPIR.

eess.IV↗
arXiv2021

Effect of pseudo datasets for the classification-based engineering design

Machine learning classification techniques have been used widely to recognize the feasible design domain and discover hidden patterns in engineering design. An accurate classification model needs a large dataset; however, generating a large dataset is costly for complex simulation-based problems. After training by a small dataset, surrogate models can generate a large pseudo dataset efficiently. Errors, however, may be introduced by surrogate modeling. This paper investigates the mutual effect of a large pseudo dataset and surrogate modeling uncertainty. Four widely used methods, i.e., Naive Bayes classifier, support vector machine, random forest regression, and artificial neural network for classification, are studied on four benchmark problems. Kriging is used as the basic surrogate model method. The results show that a large pseudo dataset improves the classification accuracy, which depends on both design problems and classification algorithms. Except for the Naive Bayes, the other three methods are recommended for using pseudo data to improve classification performance. Also, a wind turbine design problem is used to illustrate the effect of the pseudo dataset on feasible subspace recognition. The large pseudo dataset improves the recognized subspace bound greatly, which can be reproduced by classification models well except for the Naive Bayes classifier. Under the uncertainty of surrogate modeling, the random forest presents high robustness to recognize the feasible design domain bound, while the artificial neural network demonstrates a high sensitivity to uncertainty with the recognized bound deteriorated.

physics.data-an↗
arXiv2021

Learning-to-Dispatch: Reinforcement Learning Based Flight Planning under Emergency

The effectiveness of resource allocation under emergencies especially hurricane disasters is crucial. However, most researchers focus on emergency resource allocation in a ground transportation system. In this paper, we propose Learning-to-Dispatch (L2D), a reinforcement learning (RL) based air route dispatching system, that aims to add additional flights for hurricane evacuation while minimizing the airspace's complexity and air traffic controller's workload. Given a bipartite graph with weights that are learned from the historical flight data using RL in consideration of short- and long-term gains, we formulate the flight dispatch as an online maximum weight matching problem. Different from the conventional order dispatch problem, there is no actual or estimated index that can evaluate how the additional evacuation flights influence the air traffic complexity. Then we propose a multivariate reward function in the learning phase and compare it with other univariate reward designs to show its superior performance. The experiments using the real-world dataset for Hurricane Irma demonstrate the efficacy and efficiency of our proposed schema.

cs.MA↗
arXiv2021

The Heterogeneity Hypothesis: Finding Layer-Wise Differentiated Network Architectures

In this paper, we tackle the problem of convolutional neural network design. Instead of focusing on the design of the overall architecture, we investigate a design space that is usually overlooked, i.e. adjusting the channel configurations of predefined networks. We find that this adjustment can be achieved by shrinking widened baseline networks and leads to superior performance. Based on that, we articulate the heterogeneity hypothesis: with the same training protocol, there exists a layer-wise differentiated network architecture (LW-DNA) that can outperform the original network with regular channel configurations but with a lower level of model complexity. The LW-DNA models are identified without extra computational cost or training time compared with the original network. This constraint leads to controlled experiments which direct the focus to the importance of layer-wise specific channel configurations. LW-DNA models come with advantages related to overfitting, i.e. the relative relationship between model complexity and dataset size. Experiments are conducted on various networks and datasets for image classification, visual tracking and image restoration. The resultant LW-DNA models consistently outperform the baseline models. Code is available at https://github.com/ofsoundof/Heterogeneity_Hypothesis.

cs.CV↗
arXiv2021

Effect of near-field distribution on transmission characteristics of fiber-fed Fabry-Perot etalons

Fiber-fed etalons are widely employed in advanced interferometric instruments such as gravitational-wave detectors, ultrastable lasers and calibration reference for high-precision spectrographs. We demonstrate that variation in near-field distribution of the feeding fiber would deteriorate the spectrum precision of the fiber-fed Fabry-Perot etalon, especially when precision at the order of 3 * 10-10 or higher is required. The octagonal fiber reinforced with double scrambler could greatly improve the steadiness and uniformness of the near-field distribution. When building wavelength calibrators for sub-m s-1 precision radial-velocity instruments, the double scrambler should be considered meticulously.

astro-ph.IM↗
arXiv2021

Fully-staggered-array bulk Re-Ba-Cu-O short-period undulator: large-scale 3D electromagnetic modelling and design optimization using A-V and H-formulation methods

The development of a new hard x-ray beamline I-TOMCAT equipped with a 1-meter-long short-period bulk high-temperature superconductor undulator (BHTSU) has been scheduled for the upgrade of the Swiss Light Source (SLS 2.0) at the Paul Scherrer Institute (PSI). The very hard x-ray source generated by the BHTSU will increase the brilliance at the beamline by over one order of magnitude in comparison to other state-of-the-art undulator technologies and allow experiments to be carried out with photon energies in excess of 60 keV. One of the key challenges for designing a 1-meter-long (100 periods) BHTSU is the large-scale simulation of the magnetization currents inside 200 staggered-array bulk superconductors. A feasible approach to simplify the electromagnetic model is to retain five periods from both ends of the 1-meter-long BHTSU, reducing the number of degrees of freedom (DOFs) to the scale of millions. In this paper, the theory of the recently-proposed 2D A-V formulation-based backward computation method is extended to calculate the critical state magnetization currents in the ten-period staggered-array BHTSU in 3D. The simulation results of the magnetization currents and the associated undulator field along the electron beam axis are compared with the well-known 3D H-formulation and the highly efficient 3D H-ϕ formulation method, all methods showing excellent agreement with each other as well as with experimental results. The mixed H-ϕ formulation avoids computing the eddy currents in the air subdomain and is significantly faster than the full H-formulation method, but is slower in comparison to the A-V formulation-based backward computation. Finally, the fastest and the most efficient A-V formulation in ANSYS 2020R1 Academic is adopted to optimize the integrals of the undulator field along the electron beam axis by optimizing the sizes of the end bulks.

physics.acc-ph↗
arXiv2021

Size-Sieving Separation of Hard-Sphere Mixtures through Cylindrical Pores

The collision dynamics of hard spheres and cylindrical pores is solved exactly, which is the minimal model for a regularly porous membrane. Nonequilibrium event-driven molecular dynamics simulations are used to show that the permeability $P$ of hard spheres of size $σ$ through cylinderical pores of size $d$ follow the hindered diffusion mechanism due to size exclusion as $P \propto (1-σ/d)^2$. Under this law, the separation of binary mixtures of large and small particles exhibits a linear relationship between $α^{-1/2}$ and $P^{-1/2}$, where $α$ and $P$ are the selectivity and permeability of the smaller particle, respectively. The mean permeability through polydisperse pores is the sum of permeabilities of individual pores, weighted by the fraction of the single pore area over the total pore area.

cond-mat.soft↗
arXiv2021

PQA: Perceptual Question Answering

Perceptual organization remains one of the very few established theories on the human visual system. It underpinned many pre-deep seminal works on segmentation and detection, yet research has seen a rapid decline since the preferential shift to learning deep models. Of the limited attempts, most aimed at interpreting complex visual scenes using perceptual organizational rules. This has however been proven to be sub-optimal, since models were unable to effectively capture the visual complexity in real-world imagery. In this paper, we rejuvenate the study of perceptual organization, by advocating two positional changes: (i) we examine purposefully generated synthetic data, instead of complex real imagery, and (ii) we ask machines to synthesize novel perceptually-valid patterns, instead of explaining existing data. Our overall answer lies with the introduction of a novel visual challenge -- the challenge of perceptual question answering (PQA). Upon observing example perceptual question-answer pairs, the goal for PQA is to solve similar questions by generating answers entirely from scratch (see Figure 1). Our first contribution is therefore the first dataset of perceptual question-answer pairs, each generated specifically for a particular Gestalt principle. We then borrow insights from human psychology to design an agent that casts perceptual organization as a self-attention problem, where a proposed grid-to-grid mapping network directly generates answer patterns from scratch. Experiments show our agent to outperform a selection of naive and strong baselines. A human study however indicates that ours uses astronomically more data to learn when compared to an average human, necessitating future research (with or without our dataset).

cs.CV↗
arXiv2021

GMOT-40: A Benchmark for Generic Multiple Object Tracking

Multiple Object Tracking (MOT) has witnessed remarkable advances in recent years. However, existing studies dominantly request prior knowledge of the tracking target, and hence may not generalize well to unseen categories. In contrast, Generic Multiple Object Tracking (GMOT), which requires little prior information about the target, is largely under-explored. In this paper, we make contributions to boost the study of GMOT in three aspects. First, we construct the first public GMOT dataset, dubbed GMOT-40, which contains 40 carefully annotated sequences evenly distributed among 10 object categories. In addition, two tracking protocols are adopted to evaluate different characteristics of tracking algorithms. Second, by noting the lack of devoted tracking algorithms, we have designed a series of baseline GMOT algorithms. Third, we perform a thorough evaluation on GMOT-40, involving popular MOT algorithms (with necessary modifications) and the proposed baselines. We will release the GMOT-40 benchmark, the evaluation results, as well as the baseline algorithm to the public upon the publication of the paper.

cs.CV↗
arXiv2021

PhySG: Inverse Rendering with Spherical Gaussians for Physics-based Material Editing and Relighting

We present PhySG, an end-to-end inverse rendering pipeline that includes a fully differentiable renderer and can reconstruct geometry, materials, and illumination from scratch from a set of RGB input images. Our framework represents specular BRDFs and environmental illumination using mixtures of spherical Gaussians, and represents geometry as a signed distance function parameterized as a Multi-Layer Perceptron. The use of spherical Gaussians allows us to efficiently solve for approximate light transport, and our method works on scenes with challenging non-Lambertian reflectance captured under natural, static illumination. We demonstrate, with both synthetic and real data, that our reconstructions not only enable rendering of novel viewpoints, but also physics-based appearance editing of materials and illumination.

cs.CV↗
arXiv2021

Unifying the Concepts of Scattering and Structure Factor in Ordered and Disordered Samples

Scattering methods are widely used in many research areas to analyze and resolve material structures. Given the importance, a large number of full textbooks are devoted to this topic. However, technical details in experiments and disconnection between explanations from different perspectives often confuse and frustrate beginner students and researchers. To create an effective learning path, we review the core concepts about scattering and structure factor in this article in a self-contained way. Classical examples of scattering photography and spectroscopy are calculated. Sample CPU and GPU codes are provided to facilitate the understanding and application of these methods.

cond-mat.soft↗
arXiv2021

Flow-based Kernel Prior with Application to Blind Super-Resolution

Kernel estimation is generally one of the key problems for blind image super-resolution (SR). Recently, Double-DIP proposes to model the kernel via a network architecture prior, while KernelGAN employs the deep linear network and several regularization losses to constrain the kernel space. However, they fail to fully exploit the general SR kernel assumption that anisotropic Gaussian kernels are sufficient for image SR. To address this issue, this paper proposes a normalizing flow-based kernel prior (FKP) for kernel modeling. By learning an invertible mapping between the anisotropic Gaussian kernel distribution and a tractable latent distribution, FKP can be easily used to replace the kernel modeling modules of Double-DIP and KernelGAN. Specifically, FKP optimizes the kernel in the latent space rather than the network parameter space, which allows it to generate reasonable kernel initialization, traverse the learned kernel manifold and improve the optimization stability. Extensive experiments on synthetic and real-world images demonstrate that the proposed FKP can significantly improve the kernel estimation accuracy with less parameters, runtime and memory usage, leading to state-of-the-art blind SR results.

cs.CV↗
arXiv2021

Evidence of wind signatures in the gas velocity profiles of Red Geysers

Spatially resolved spectroscopy from SDSS-IV MaNGA survey has revealed a class of quiescent, relatively common early-type galaxies, termed "red geysers", that possibly host large scale active galactic nuclei driven winds. Given their potential importance in maintaining low level of star formation at late times, additional evidence confirming that winds are responsible for the red geyser phenomenon is critical. In this work, we present follow-up observations with the Echellette Spectrograph and Imager (ESI) at the Keck telescope of two red geysers (z$<$0.1) using multiple long slit positions to sample different regions of each galaxy. Our ESI data with a spectral resolution (R) $\sim$ 8000 improves upon MaNGA's resolution by a factor of four, allowing us to resolve the ionized gas velocity profiles along the putative wind cone with an instrumental resolution of $\rm σ= 16~km~s^{-1}$. The line profiles of H$α$ and [NII]$\rm λ6584$ show asymmetric shapes that depend systematically on location $-$ extended blue wings on the red-shifted side of the galaxy and red wings on the opposite side. We construct a simple wind model and show that our results are consistent with geometric projections through an outflowing conical wind oriented at an angle towards the line of sight. An alternative hypothesis that assigns the asymmetric pattern to "beam-smearing" of a rotating, ionized gas disk does a poor job matching the line asymmetry profiles. While our study features just two sources, it lends further support to the notion that red geysers are the result of galaxy-scale winds.

astro-ph.GA↗
arXiv2021

Universal non-Hermitian skin effect in two and higher dimensions

Skin effect, experimentally discovered in one dimension, describes the physical phenomenon that on an open chain, an extensive number of eigenstates of a non-Hermitian hamiltonian are localized at the end(s) of the chain. Here in two and higher dimensions, we establish a theorem that the skin effect exists, if and only if periodic-boundary spectrum of the hamiltonian covers a finite area on the complex plane. This theorem establishes the universality of the effect, because the above condition is satisfied in almost every generic non-Hermitian hamiltonian, and, unlike in one dimension, is compatible with all spatial symmetries. We propose two new types of skin effect in two and higher dimensions: the corner-skin effect where all eigenstates are localized at one corner of the system, and the geometry-dependent-skin effect where skin modes disappear for systems of a particular shape, but appear on generic polygons. An immediate corollary of our theorem is that any non-Hermitian system having exceptional points (lines) in two (three) dimensions exhibits skin effect, making this phenomenon accessible to experiments in photonic crystals, Weyl semimetals, and Kondo insulators.

cond-mat.mes-hall↗
arXiv2021

Spatio-Temporal Data Mining for Aviation Delay Prediction

To accommodate the unprecedented increase of commercial airlines over the next ten years, the Next Generation Air Transportation System (NextGen) has been implemented in the USA that records large-scale Air Traffic Management (ATM) data to make air travel safer, more efficient, and more economical. A key role of collaborative decision making for air traffic scheduling and airspace resource management is the accurate prediction of flight delay. There has been a lot of attempts to apply data-driven methods such as machine learning to forecast flight delay situation using air traffic data of departures and arrivals. However, most of them omit en-route spatial information of airlines and temporal correlation between serial flights which results in inaccuracy prediction. In this paper, we present a novel aviation delay prediction system based on stacked Long Short-Term Memory (LSTM) networks for commercial flights. The system learns from historical trajectories from automatic dependent surveillance-broadcast (ADS-B) messages and uses the correlative geolocations to collect indispensable features such as climatic elements, air traffic, airspace, and human factors data along posterior routes. These features are integrated and then are fed into our proposed regression model. The latent spatio-temporal patterns of data are abstracted and learned in the LSTM architecture. Compared with previous schemes, our approach is demonstrated to be more robust and accurate for large hub airports.

cs.LG↗