arXiv subjects

Yong He

Publications and source records attributed to Yong He.

At least 91 records · Page 5Linked to original sources

arXiv2020

Self-supervised Learning with Fully Convolutional Networks

Although deep learning based methods have achieved great success in many computer vision tasks, their performance relies on a large number of densely annotated samples that are typically difficult to obtain. In this paper, we focus on the problem of learning representation from unlabeled data for semantic segmentation. Inspired by two patch-based methods, we develop a novel self-supervised learning framework by formulating the Jigsaw Puzzle problem as a patch-wise classification process and solving it with a fully convolutional network. By learning to solve a Jigsaw Puzzle problem with 25 patches and transferring the learned features to semantic segmentation task on Cityscapes dataset, we achieve a 5.8 percentage point improvement over the baseline model that initialized from random values. Moreover, experiments show that our self-supervised learning method can be applied to different datasets and models. In particular, we achieved competitive performance with the state-of-the-art methods on the PASCAL VOC2012 dataset using significant fewer training images.

cs.CV↗

arXiv2020

Projected Estimation for Large-dimensional Matrix Factor Models

In this study, we propose a projection estimation method for large-dimensional matrix factor models with cross-sectionally spiked eigenvalues. By projecting the observation matrix onto the row or column factor space, we simplify factor analysis for matrix series to that for a lower-dimensional tensor. This method also reduces the magnitudes of the idiosyncratic error components, thereby increasing the signal-to-noise ratio, because the projection matrix linearly filters the idiosyncratic error matrix. We theoretically prove that the projected estimators of the factor loading matrices achieve faster convergence rates than existing estimators under similar conditions. Asymptotic distributions of the projected estimators are also presented. A novel iterative procedure is given to specify the pair of row and column factor numbers. Extensive numerical studies verify the empirical performance of the projection method. Two real examples in finance and macroeconomics reveal factor patterns across rows and columns, which coincides with financial, economic, or geographical interpretations.

stat.ME↗

arXiv2020

Enable an Open Software Defined Mobility Ecosystem through VEC-OF

OEMs and new entrants can take the Mobility as a Service market (MaaS) as the entry point, upgrade its E/E (Electric and Electronic) architecture to be C/C (Computing and Communication) architecture, build one open software defined and data driven software platform for its production and service model, use efficient and collaborative ways of vehicles, roads, cloud and network to continuously improve core technologies such as autonomous driving, provide MaaS operators with an affordable and agile platform. In this paper we present one new framework, VEC-OF (Vehicle-Edge-Cloud Open Framework), which is a new data and AI centric vehicle software framework enabling a much safer, more efficient, connected and trusted MaaS through cooperative vehicle, infrastructure and cloud capabilities and intelligence

cs.CY↗

arXiv2020

Alveolar mimics with periodic strain and its effect on the cell layer formation

We report on the development of a new model of alveolar air-tissue interface on a chip. The model consists of an array of suspended hexagonal monolayers of gelatin nanofibers supported by microframes and a microfluidic device for the patch integration. The suspended monolayers are deformed to a central displacement of 40-80 um at the air-liquid interface by application of air pressure in the range of 200-1000 Pa. With respect to the diameter of the monolayers that is 500 um, this displacement corresponds to a linear strain of 2-10% in agreement with the physiological strain range in the lung alveoli. The culture of A549 cells on the monolayers for an incubation time 1-3 days showed viability in the model. We exerted a periodic strain of 5% at a frequency of 0.2 Hz during 1 hour to the cells. We found that the cells were strongly coupled to the nanofibers, but the strain reduced the coupling and induced remodeling of the actin cytoskeleton, which led to a better tissue formation. Our model can serve as a versatile tool in lung investigations such as in inhalation toxicology and therapy.

cond-mat.soft↗

arXiv2020

Large-dimensional Factor Analysis without Moment Constraints

Large-dimensional factor model has drawn much attention in the big-data era, in order to reduce the dimensionality and extract underlying features using a few latent common factors. Conventional methods for estimating the factor model typically requires finite fourth moment of the data, which ignores the effect of heavy-tailedness and thus may result in unrobust or even inconsistent estimation of the factor space and common components. In this paper, we propose to recover the factor space by performing principal component analysis to the spatial Kendall's tau matrix instead of the sample covariance matrix. In a second step, we estimate the factor scores by the ordinary least square (OLS) regression. Theoretically, we show that under the elliptical distribution framework the factor loadings and scores as well as the common components can be estimated consistently without any moment constraint. The convergence rates of the estimated factor loadings, scores and common components are provided. The finite sample performance of the proposed procedure is assessed through thorough simulations. An analysis of a financial data set of asset returns shows the superiority of the proposed method over the classical PCA method.

stat.ME↗

arXiv2020

Multiscale Modeling and Analysis for High-fidelity Interferometric Scattering Microscopy

Interferometric scattering microscopy (iSCAT), as an ultrasensitive fluorescence-free imaging modality, has recently gain enormous attention and been rapidly developing from demonstration of principle to quantitative sensing. Here we report on a theoretical and experimental study for iSCAT with samples having structural dimensions that differ by 4-5 orders of magnitude. In particular, we demonstrate and intuitively explain the profound effects of sub-nanometer surface roughness of a glass coverslip and of a mica surface on the absolute signal and the shape of the point spread function of a gold nanoparticle. These quantities significantly affect the accuracies for determining the target size and position in all three dimensions. Moreover, we investigate a sample system mimicking a gold nanoparticle in a simplified cell environment and show position-dependent and even asymmetric point spread function of the nanoparticle. The multiscale study will facilitate the development of high fidelity iSCAT in real applications.

physics.optics↗

arXiv2020

Robust Covariance Estimation for High-dimensional Compositional Data with Application to Microbial Communities Analysis

Microbial communities analysis is drawing growing attention due to the rapid development of high-throughput sequencing techniques nowadays. The observed data has the following typical characteristics: it is high-dimensional, compositional (lying in a simplex) and even would be leptokurtic and highly skewed due to the existence of overly abundant taxa, which makes the conventional correlation analysis infeasible to study the co-occurrence and co-exclusion relationship between microbial taxa. In this article, we address the challenges of covariance estimation for this kind of data. Assuming the basis covariance matrix lying in a well-recognized class of sparse covariance matrices, we adopt a proxy matrix known as centered log-ratio covariance matrix in the literature, which is approximately indistinguishable from the real basis covariance matrix as the dimensionality tends to infinity. We construct a Median-of-Means (MOM) estimator for the centered log-ratio covariance matrix and propose a thresholding procedure that is adaptive to the variability of individual entries. By imposing a much weaker finite fourth moment condition compared with the sub-Gaussianity condition in the literature, we derive the optimal rate of convergence under the spectral norm. In addition, we also provide theoretical guarantee on support recovery. The adaptive thresholding procedure of the MOM estimator is easy to implement and gains robustness when outliers or heavy-tailedness exist. Thorough simulation studies are conducted to show the advantages of the proposed procedure over some state-of-the-arts methods. At last, we apply the proposed method to analyze a microbiome dataset in human gut. The R script for implementing the method is available at https://github.com/heyongstat/RCEC.

stat.ME↗

arXiv2020

Network-Assisted Estimation for Large-dimensional Factor Model with Guaranteed Convergence Rate Improvement

Network structure is growing popular for capturing the intrinsic relationship between large-scale variables. In the paper we propose to improve the estimation accuracy for large-dimensional factor model when a network structure between individuals is observed. To fully excavate the prior network information, we construct two different penalties to regularize the factor loadings and shrink the idiosyncratic errors. Closed-form solutions are provided for the penalized optimization problems. Theoretical results demonstrate that the modified estimators achieve faster convergence rates and lower asymptotic mean squared errors when the underlying network structure among individuals is correct. An interesting finding is that even if the priori network is totally misleading, the proposed estimators perform no worse than conventional state-of-art methods. Furthermore, to facilitate the practical application, we propose a data-driven approach to select the tuning parameters, which is computationally efficient. We also provide an empirical criterion to determine the number of common factors. Simulation studies and application to the S&P100 weekly return dataset convincingly illustrate the superiority and adaptivity of the new approach.

stat.ME↗

arXiv2019

Appending Adversarial Frames for Universal Video Attack

There have been many efforts in attacking image classification models with adversarial perturbations, but the same topic on video classification has not yet been thoroughly studied. This paper presents a novel idea of video-based attack, which appends a few dummy frames (e.g., containing the texts of `thanks for watching') to a video clip and then adds adversarial perturbations only on these new frames. Our approach enjoys three major benefits, namely, a high success rate, a low perceptibility, and a strong ability in transferring across different networks. These benefits mostly come from the common dummy frame which pushes all samples towards the boundary of classification. On the other hand, such attacks are easily to be concealed since most people would not notice the abnormality behind the perturbed video clips. We perform experiments on two popular datasets with six state-of-the-art video classification models, and demonstrate the effectiveness of our approach in the scenario of universal video attacks.

cs.CV↗

arXiv2019

Single-Molecule Doped Crystalline Nanosheets for Delicate Photophysics Studies and Directional Single-Photon Emitting Devices

Single molecules in solids have been considered as an attractive class of solid-state single quantum systems because they can be chemically synthesized at low cost to have stable narrow transitions at desired wavelengths. Here we report and demonstrate single dibenzoterrylene molecules in crystalline anthracene nanosheets as a robust and versatile solid-state platform for delicate photophysics studies and building blocks of single-photon devices. The high-quality nanosheet sample enables robust studies of delicate single-molecule photophysics at room temperature, including the first real-time observation of single molecule insertion site jump, quantitative measurements of the associated changes of dipole moment orientation and magnitude, unambiguous determination of excitation-power dependent intersystem crossing rate and triplet lifetime. Moreover, we demonstrate the flexible assembly of the nanosheets into a planar antenna device to achieve bright single-molecule emission with a Gaussian emission pattern. The thin thickness, good photostability and mechanical stability make the dibenzoterrylene-in-anthracene nanosheet system an excellent candidate of static quantum nodes in integrated photonic circuit.

physics.app-ph↗

arXiv2019

Robust Factor Number Specification for Large-dimensional Elliptical Factor Model

The accurate specification of the number of factors is critical to the validity of factor models and the topic almost occupies the central position in factor analysis. Plenty of estimators are available under the restrictive condition that the fourth moments of the factors and idiosyncratic errors are bounded. In this paper we propose efficient and robust estimators for the factor number via considering a more general static Elliptical Factor Model (EFM) framework. We innovatively propose to exploit the multivariate Kendall's tau matrix, which captures the correlation structure of elliptical random vectors. Theoretically we show that the proposed estimators are consistent without exerting any moment condition when both cross-sections N and time dimensions T go to infinity. Simulation study shows that the new estimators perform much better in heavy-tailed data setting while performing comparably with the state-of-the-art methods in the light-tailed Gaussian setting. At last, a real macroeconomic data example is given to illustrate its empirical advantages and usefulness.

stat.ME↗

arXiv2018

High-dimensional Two-sample Precision Matrices Test: An Adaptive Approach through Multiplier Bootstrap

Precision matrix, which is the inverse of covariance matrix, plays an important role in statistics, as it captures the partial correlation between variables. Testing the equality of two precision matrices in high dimensional setting is a very challenging but meaningful problem, especially in the differential network modelling. To our best knowledge, existing test is only powerful for sparse alternative patterns where two precision matrices differ in a small number of elements. In this paper we propose a data-adaptive test which is powerful against either dense or sparse alternatives. Multiplier bootstrap approach is utilized to approximate the limiting distribution of the test statistic. Theoretical properties including asymptotic size and power of the test are investigated. Simulation study verifies that the data-adaptive test performs well under various alternative scenarios. The practical usefulness of the test is illustrated by applying it to a gene expression data set associated with lung cancer.

stat.ME↗

arXiv2018

Doubly Robust Sure Screening for Elliptical Copula Regression Model

Regression analysis has always been a hot research topic in statistics. We propose a very flexible semi-parametric regression model called Elliptical Copula Regression (ECR) model, which covers a large class of linear and nonlinear regression models such as additive regression model,single index model. Besides, ECR model can capture the heavy-tail characteristic and tail dependence between variables, thus it could be widely applied in many areas such as econometrics and finance. In this paper we mainly focus on the feature screening problem for ECR model in ultra-high dimensional setting. We propose a doubly robust sure screening procedure for ECR model, in which two types of correlation coefficient are involved: Kendall tau correlation and Canonical correlation. Theoretical analysis shows that the procedure enjoys sure screening property, i.e., with probability tending to 1, the screening procedure selects out all important variables and substantially reduces the dimensionality to a moderate size against the sample size. Thorough numerical studies are conducted to illustrate its advantage over existing sure independence screening methods and thus it can be used as a safe replacement of the existing procedures in practice. At last, the proposed procedure is applied on a gene-expression real data set to show its empirical usefulness.

stat.ME↗

arXiv2018

Two-dimensional InSe/WS$_2$ heterostructure with enhanced optoelectronic performance in the visible region

Two-dimensional (2D) InSe and WS$_2$ exhibit promising characteristics for optoelectronic and photoelectrochemical applications, e.g. photodetection and photocatalytic water splitting. However, both of them have poor absorption of visible light due to wide band gaps. 2D InSe has high electron mobility but low hole mobility, while 2D WS$_2$ is on the opposite. Here, we design a 2D heterostructure composed of their monolayers and study its optoelectronic properties by first-principles calculations. Our results show that the heterostructure has a direct band gap of 2.19 eV, which is much smaller than those of the monolayers mainly due to a type-II band alignment: the valence band maximum and the conduction band minimum of monolayer InSe are lower than those of monolayer WS$_2$, respectively. The visible-light absorption is enhanced considerably, e.g. about fivefold (threefold) increase at the wavelength of 490 nm in comparison to monolayer InSe (WS$_2$). The type-II band alignment also facilitates the spatial separation of photogenerated electron-hole pairs, i.e., electrons (holes) reside preferably in the InSe (WS$_2$) layer. The two layers complement each other in carrier mobilities of the heterostructure: the photogenerated electrons and holes inherit the large mobilities from the InSe and WS$_2$ monolayers, respectively.

cond-mat.mtrl-sci↗

arXiv2018

Large Time Behavior and Convergence for the Camassa-Holm Equations with Fractional Laplacian Viscosity

In this paper, we consider the $n$-dimensional ($n=2,3$) Camassa-Holm equations with fractional Laplacian viscosity in the whole space. In stark contrast to the Camassa-Holm equations without any nonlocal effect, to our best knowledge, little has been known on the large time behavior and convergence for the nonlocal equations under study. We first study the large time behavior of solutions. We then discuss the relation between the equations under consideration and the imcompressible Navier-Stokes equations with fractional Laplacian viscosity (INSF). The main difficulty to achieve them lies in the fractional Laplacian viscosity. Fortunately, by employing some properties of fractional Laplacian, in particular, the fractional Leibniz chain rule and the fractional Gagliardo-Nirenberg-Sobolev type estimates, the high and low frequency splitting method and the Fourier splitting method, we first establish the large time behavior concerning non-uniform decay and algebraic decay of solutions to the nonlocal equations under study. In particular, under the critical case $s=\dfrac{n}{4}$, the nonlocal version of Ladyzhenskaya's inequality is skillfully used, and the smallness of initial data in several Sobolev spaces is required to gain the non-uniform decay and algebraic decay. On the other hand, by means of the fractional heat kernel estimates, we figure out the relation between the nonlocal equations under consideration and the equations (INSF). Specifically, we prove that the solution to the Camassa-Holm equations with nonlocal viscosity converges strongly as the filter parameter $α\rightarrow~0$ to a solution of the equations (INSF).

math.AP↗

arXiv2018

Regularity of Solutions of the Camassa-Holm Equations with Fractional Laplacian Viscosity

We study the existence, uniqueness and regularity of solutions to the $n$-dimensional ($n=2,3$) Camassa-Holm equations with fractional Laplacian viscosity with smooth initial data. It is a coupled system between the Navier-Stokes equations with nonlocal viscosity and a Helmholtz equation. The main difficulty lies in establishing some a priori estimates for the fractional Laplacian viscosity. To achieve this, we need to explore suitable fractional-power Sobolev-type estimates, and bilinear estimates for fractional derivatives. Especially, for the critical case $\displaystyle s=\frac{n}{4}$ with $n=2,3$, we will make extra efforts for acquiring the expected estimates as obtained in the case $\displaystyle \frac{n}{4}<s<1$. By the aid of the fractional Leibniz rule and the nonlocal version of Ladyzhenskaya's inequality, we prove the existence, uniqueness and regularity to the Camassa-Holm equations under study by the energy method and a bootstrap argument, which rely crucially on the fractional Laplacian viscosity. In particular, under the critical case $s=\dfrac{n}{4}$, the nonlocal version of Ladyzhenskaya's inequality is skillfully used, and the smallness of initial data in several Sobolev spaces is required to gain the desired results concernig existence, uniqueness and regularity.

math.AP↗

arXiv2017

Confined subdiffusion in three dimensions

The three-dimensional (3D) Fick's diffusion equation and fractional diffusion equation are solved for different reflecting boundaries. We use the continuous time random walk model (CTRW) to investigate the time averaged mean square displacement (MSD) of 3D single particle trajectory. Theoretical results show the ensemble average of the time averaged MSD can be expressed analytically by a Mittag-Leffler function. Our new expression is in agreement with previous formulas in two limiting cases which are $ < \overline{δ^2} > \simΔ$ in short lag time and $ < \overline{δ^2}> \simΔ^{1-α}$ in long lag time. We also simulate the experimental data of mRNA diffusion in living E. coli using 3D CTRW model under confined and crowded conditions. The simulated results are well consistent with experimental results. The calculations of power spectral density (PSD) indicate further the subdiffsive behavior of individual trajectory.

cond-mat.soft↗

arXiv2016

White matter deficits underlie the loss of consciousness level and predict recovery outcome in disorders of consciousness

This study aimed to identify white matter (WM) deficits underlying the loss of consciousness in disorder of consciousness (DOC) patients using Diffusion Tensor Imaging (DTI) and to demonstrate the potential value of DTI parameters in predicting recovery outcomes of DOC patients. With 30 DOC patients (8 comatose, 8 unresponsive wakefulness syndrome/vegetative state, and 14 minimal conscious state) and 25 patient controls, we performed group comparison of DTI parameters across 48 core WM regions of interest (ROIs) using Analysis of Covariance. Compared with controls, DOC patients had decreased Fractional anisotropy (FA) and increased diffusivities in widespread WM area.The corresponding DTI parameters of those WM deficits in DOC patients significantly correlated with the consciousness level evaluated by Coma Recovery Scale Revised (CRS-R) and Glasgow Coma Scale (GCS). As for predicting the recovery outcomes (i.e., regaining consciousness or not, grouped by their Glasgow Outcome Scale more than 2 or not) at 3 months post scan, radial diffusivity of left superior cerebellar peduncle and FA of right sagittal stratum reached an accuracy of 87.5% and 75% respectively. Our findings showed multiple WM deficits underlying the loss of consciousness level, and demonstrated the potential value of these WM areas in predicting the recovery outcomes of DOC patients who have lost awareness of the environment and themselves.

q-bio.NC↗