arXiv subjects

Qiao Wang

Publications and source records attributed to Qiao Wang.

At least 91 records · Page 5Linked to original sources

arXiv2019

Unsupervised Classification of Street Architectures Based on InfoGAN

Street architectures play an essential role in city image and streetscape analysing. However, existing approaches are all supervised which require costly labeled data. To solve this, we propose a street architectural unsupervised classification framework based on Information maximizing Generative Adversarial Nets (InfoGAN), in which we utilize the auxiliary distribution $Q$ of InfoGAN as an unsupervised classifier. Experiments on database of true street view images in Nanjing, China validate the practicality and accuracy of our framework. Furthermore, we draw a series of heuristic conclusions from the intrinsic information hidden in true images. These conclusions will assist planners to know the architectural categories better.

cs.CV↗

arXiv2019

A Deep, Information-theoretic Framework for Robust Biometric Recognition

Deep neural networks (DNN) have been a de facto standard for nowadays biometric recognition solutions. A serious, but still overlooked problem in these DNN-based recognition systems is their vulnerability against adversarial attacks. Adversarial attacks can easily cause the output of a DNN system to greatly distort with only tiny changes in its input. Such distortions can potentially lead to an unexpected match between a valid biometric and a synthetic one constructed by a strategic attacker, raising security issue. In this work, we show how this issue can be resolved by learning robust biometric features through a deep, information-theoretic framework, which builds upon the recent deep variational information bottleneck method but is carefully adapted to biometric recognition tasks. Empirical evaluation demonstrates that our method not only offers stronger robustness against adversarial attacks but also provides better recognition performance over state-of-the-art approaches.

cs.CV↗

arXiv2018

Discrete and continuous coupled nonlinear integrable systems via the dressing method

A discrete analogue of the dressing method is presented and used to derive integrable nonlinear evolution equations, including two infinite families of novel continuous and discrete coupled integrable systems of equations of nonlinear Schrodinger type. First, a demonstration is given of how discrete nonlinear integrable equations can be derived starting from their linear counterparts. Then, starting from two uncoupled, discrete one-directional linear wave equations, an appropriate matrix Riemann-Hilbert problem is constructed, and a discrete matrix nonlinear Schrodinger system of equations is derived, together with its Lax pair. The corresponding compatible vector reductions admitted by these systems are also discussed, as well as their continuum limits. Finally, by increasing the size of the problem, three-component discrete and continuous integrable discrete systems are derived, as well as their generalizations to systems with an arbitrary number of components.

nlin.SI↗

arXiv2018

Resolution of the apparent discrepancy between the number of massive subhaloes in Abell 2744 and ΛCDM

Schwinn et al. (2017) have recently compared the abundance and distribution of massive substructures identified in a gravitational lensing analysis of Abell 2744 by Jauzac et al. (2016) and N-body simulation and found no cluster in ΛCDM simulation that is similar to Abell 2744. Schwinn et al.(2017) identified the measured projected aperture masses with the actual masses associated with subhaloes in the MXXL N-body simulation. We have used the high resolution Phoenix cluster simulations to show that such an identification is incorrect: the aperture mass is dominated by mass in the body of the cluster that happens to be projected along the line-of-sight to the subhalo. This enhancement varies from factors of a few to factors of more than 100, particularly for subhaloes projected near the centre of the cluster. We calculate aperture masses for subhaloes in our simulation and compare them to the measurements for Abell 2744. We find that the data for Abell 2744 are in excellent agreement with the matched predictions from ΛCDM. We provide further predictions for aperture mass functions of subhaloes in idealized surveys with varying mass detection thresholds.

astro-ph.GA↗

arXiv2018

Projection effects of large-scale structures on weak-lensing peak abundances

High peaks in weak lensing (WL) maps originate dominantly from the lensing effects of single massive halos. Their abundance is therefore closely related to the halo mass function and thus a powerful cosmological probe. On the other hand, however, besides individual massive halos, large-scale structures (LSS) along lines of sight also contribute to the peak signals. In this paper, with ray tracing simulations, we investigate the LSS projection effects. We show that for current surveys with a large shape noise, the stochastic LSS effects are subdominant. For future WL surveys with source galaxies having a median redshift $z_{\mathrm{med}}\sim1$ or higher, however, they are significant. For the cosmological constraints derived from observed WL high peak counts, severe biases can occur if the LSS effects are not taken into account properly. We extend the model of \citet{Fan2010} by incorporating the LSS projection effects into the theoretical considerations. By comparing with simulation results, we demonstrate the good performance of the improved model and its applicability in cosmological studies.

astro-ph.CO↗

arXiv2018

PHoToNs--A Parallel Heterogeneous & Threads oriented code for cosmological N-body simulation

We introduce a new code for cosmological simulations, PHoToNs, which has features on performing massive cosmological simulations on heterogeneous high performance Computer (HPC) and threads oriented programming. PHoToNs adopts a hybrid scheme to compute gravity force, with the conventional PM to compute the long-range force, the Tree algorithm to compute the short range force, and the direct summation PP to compute the gravity from very close particles. A self-similar space filling Peano-Hilbert curve is used to decompose computing domain. Threads programming is highly used to more flexibly manage the domain communication, PM calculation and synchronization, as well as Dual Tree Traversal on the CPU+MIC platform. The scalability of the PHoToNs performs well and the efficiency of PP kernel achieves 68.6% of peak performance on MIC and 74.4% on CPU platforms. We also test the accuracy of the code against the much used Gadget-2 in the community and found excellent agreement.

astro-ph.CO↗

arXiv2017

Equivalence between LINE and Matrix Factorization

LINE [1], as an efficient network embedding method, has shown its effectiveness in dealing with large-scale undirected, directed, and/or weighted networks. Particularly, it proposes to preserve both the local structure (represented by First-order Proximity) and global structure (represented by Second-order Proximity) of the network. In this study, we prove that LINE with these two proximities (LINE(1st) and LINE(2nd)) are actually factoring two different matrices separately. Specifically, LINE(1st) is factoring a matrix M (1), whose entries are the doubled Pointwise Mutual Information (PMI) of vertex pairs in undirected networks, shifted by a constant. LINE(2nd) is factoring a matrix M (2), whose entries are the PMI of vertex and context pairs in directed networks, shifted by a constant. We hope this finding would provide a basis for further extensions and generalizations of LINE.

cs.LG↗

arXiv2017

KiDS-450: Cosmological Constraints from Weak Lensing Peak Statistics-I: Inference from Analytical Prediction of High Signal-to-Noise Ratio Convergence Peaks

This paper is the first of a series of papers constraining cosmological parameters with weak lensing peak statistics using $\sim 450~\rm deg^2$ of imaging data from the Kilo Degree Survey (KiDS-450). We measure high signal-to-noise ratio (SNR: $ν$) weak lensing convergence peaks in the range of $3<ν<5$, and employ theoretical models to derive expected values. These models are validated using a suite of simulations. We take into account two major systematic effects, the boost factor and the effect of baryons on the mass-concentration relation of dark matter haloes. In addition, we investigate the impacts of other potential astrophysical systematics including the projection effects of large scale structures, intrinsic galaxy alignments, as well as residual measurement uncertainties in the shear and redshift calibration. Assuming a flat $Λ$CDM model, we find constraints for $S_{\rm 8}=σ_{\rm 8}(Ω_{\rm m}/0.3)^{0.5}=0.746^{+0.046}_{-0.107}$ according to the degeneracy direction of the cosmic shear analysis and $Σ_{\rm 8}=σ_{\rm 8}(Ω_{\rm m}/0.3)^{0.38}=0.696^{+0.048}_{-0.050}$ based on the derived degeneracy direction of our high-SNR peak statistics. The difference between the power index of $S_{\rm 8}$ and in $Σ_{\rm 8}$ indicates that combining the two probes has the potential to break the degeneracy in $σ_{\rm 8}$ and $Ω_{\rm m}$. Our results are consistent with the cosmic shear tomographic correlation analysis of the same dataset and $\sim 2σ$ lower than the Planck 2016 results.

astro-ph.CO↗

arXiv2017

Whitham modulation theory for the two-dimensional Benjamin-Ono equation

Whitham modulation theory for the two dimensional Benjamin-Ono (2DBO) equation is presented. A system of five quasi-linear first-order partial differential equations is derived. The system describes modulations of the traveling wave solutions of the 2DBO equation. These equations are transformed to a singularity-free hyrdodynamic-like system referred to here as the 2DBO-Whitham system. Exact reductions of this system are discussed, the formulation of initial value problems is considered, and the system is used to study the transverse stability of traveling wave solutions of the 2DBO equation.

nlin.PS↗

arXiv2017

Whitham modulation theory for the Kadomtsev-Petviashvili equation

The genus-1 KP-Whitham system is derived for both variants of the Kadomtsev-Petviashvili (KP) equation (namely, the KPI and KPII equations). The basic properties of the KP-Whitham system, including symmetries, exact reductions, and its possible complete integrability, together with the appropriate generalization of the one-dimensional Riemann problem for the Korteweg-deVries equation are discussed. Finally, the KP-Whitham system is used to study the linear stability properties of the genus-1 solutions of the KPI and KPII equations; it is shown that all genus-1 solutions of KPI are linearly unstable while all genus-1 solutions of KPII {are linearly stable within the context of Whitham theory.

nlin.SI↗

arXiv2017

Globular clusters vs dark matter haloes in strong lensing observations

Small distortions in the images of Einstein rings or giant arcs offer the exciting prospect of detecting dark matter haloes or subhaloes of mass below $10^9$M$_{\odot}$, most of which are too small to have made a visible galaxy. A very large number of such haloes are predicted to exist in the cold dark matter model of cosmogony; in contrast other models, such as warm dark matter, predict no haloes below a mass of this order which depends on the properties of the warm dark matter particle. Attempting to detect these small perturbers could therefore discriminate between different kinds of dark matter particles, and even rule out the cold dark matter model altogether. Globular clusters in the lens galaxy also induce distortions in the image which could, in principle, contaminate the test. Here, we investigate the population of globular clusters in six early type galaxies in the Virgo cluster. We find that the number density of globular clusters of mass $\sim10^6$M$_{\odot}$ is comparable to that of the dark matter perturbers (including subhaloes in the lens and haloes along the line-of-sight). We show that the very different degrees of mass concentration in globular clusters and dark matter haloes result in different lensing distortions. These are detectable with milli-arcsecond resolution imaging which can distinguish between globular cluster and dark matter halo signals.

astro-ph.GA↗

arXiv2017

Torus Models of the Outer Disc of the Milky Way using LAMOST Survey Data

With a sample of 48,161 K giant stars selected from the LAMOST DR 2 catalogue, we construct torus models in a large volume extending, for the first time, from the solar vicinity to a Galactocentric distance of $\sim 20$ kpc, reaching the outskirts of the Galactic disc. We show that the kinematics of the K giant stars match conventional models, e.g. as created by Binney in 2012, in the Solar vicinity. However such two-disc models fail if they are extended to the outer regions, even if an additional disc component is utilised. If we loosen constraints in the Sun's vicinity, we find that an effective thick disc model could explain the anti-centre of the MW. The LAMOST data imply that the sizes of the Galactic discs are much larger, and that the outer disc is much thicker, than previously thought, or alternatively that the outer structure is not a conventional disc at all. However, the velocity dispersion $σ_{0z}$ of the kinematically thick disc in the best-fitting model is about 80 km s$^{-1}$ and has a scale parameter $R_σ$ for an exponential distribution function of $\sim 19$ kpc. Such a height $σ_{0z}$ is strongly rejected by current measurements in the solar neighbourhood, and thus a model beyond quasi-thermal, two or three thin or thick discs is required.

astro-ph.GA↗

arXiv2017

Calibration of a two-state pitch-wise HMM method for note segmentation in Automatic Music Transcription systems

Many methods for automatic music transcription involves a multi-pitch estimation method that estimates an activity score for each pitch. A second processing step, called note segmentation, has to be performed for each pitch in order to identify the time intervals when the notes are played. In this study, a pitch-wise two-state on/off firstorder Hidden Markov Model (HMM) is developed for note segmentation. A complete parametrization of the HMM sigmoid function is proposed, based on its original regression formulation, including a parameter alpha of slope smoothing and beta? of thresholding contrast. A comparative evaluation of different note segmentation strategies was performed, differentiated according to whether they use a fixed threshold, called "Hard Thresholding" (HT), or a HMM-based thresholding method, called "Soft Thresholding" (ST). This evaluation was done following MIREX standards and using the MAPS dataset. Also, different transcription scenarios and recording natures were tested using three units of the Degradation toolbox. Results show that note segmentation through a HMM soft thresholding with a data-based optimization of the {alpha,beta} parameter couple significantly enhances transcription performance.

stat.ME↗

arXiv2016

Projection effects in the strong lensing study of subhaloes

The defining characteristic of the cold dark matter (CDM) hypothesis is the presence of a very large number of low-mass haloes, too small to have made a visible galaxy. Other hypotheses for the nature of the dark matter, such as warm dark matter (WDM), predict a much smaller number of such low-mass haloes. Strong lensing systems offer the possibility of detecting small-mass haloes through the distortions they induce in the lensed image. Here we show that the main contribution to the image distortions comes from haloes along the line of sight rather than subhaloes in the lens as has normally been assumed so far. These interlopers enhance the differences between the predictions of CDM and WDM models. We derive the total perturber mass function, including both subhaloes and interlopers, and show that measurements of approximately 20 strong lens systems with a detection limit of $M_{\rm low}=10^7 h^{-1} M_{\odot}$ would distinguish (at 3 sigma) between CDM and a WDM model consisting of 7 keV sterile neutrinos such as those required to explain the recently detected 3.5 keV X-ray emission line from the centres of galaxies and clusters.

astro-ph.CO↗

arXiv2016

Constraining $f(R)$ Gravity Theory Using Weak Lensing Peak Statistics from the Canada-France-Hawaii-Telescope Lensing Survey

In this Letter, we report the observational constraints on the Hu-Sawicki $f(R)$ theory derived from weak lensing peak abundances, which are closely related to the mass function of massive halos. In comparison with studies using optical or x-ray clusters of galaxies, weak lensing peak analyses have the advantages of not relying on mass-baryonic observable calibrations. With observations from the Canada-France-Hawaii-Telescope Lensing Survey, our peak analyses give rise to a tight constraint on the model parameter $|f_{R0}|$ for $n=1$. The $95\%$ CL limit is $\log_{10}|f_{R0}| < -4.82$ given WMAP9 priors on $(Ω_{\rm m}, A_{\rm s})$. With Planck15 priors, the corresponding result is $\log_{10}|f_{R0}| < -5.16$.

astro-ph.CO↗

arXiv2016

The Sum Rate of Vector Gaussian Multiple Description Coding with Tree-Structured Covariance Distortion Constraints

A single-letter lower bound on the sum rate of multiple description coding with tree-structured distortion constraints is established by generalizing Ozarow's celebrated converse argument through the introduction of auxiliary random variables that form a Markov tree. For the quadratic vector Gaussian case, this lower bound is shown to be achievable by an extended version of the El Gamal-Cover scheme, yielding a complete sum-rate characterization.

cs.IT↗

arXiv2016

Virtual Worlds as Proxy for Multi-Object Tracking Analysis

Modern computer vision algorithms typically require expensive data acquisition and accurate manual labeling. In this work, we instead leverage the recent progress in computer graphics to generate fully labeled, dynamic, and photo-realistic proxy virtual worlds. We propose an efficient real-to-virtual world cloning method, and validate our approach by building and publicly releasing a new video dataset, called Virtual KITTI (see http://www.xrce.xerox.com/Research-Development/Computer-Vision/Proxy-Virtual-Worlds), automatically labeled with accurate ground truth for object detection, tracking, scene and instance segmentation, depth, and optical flow. We provide quantitative experimental evidence suggesting that (i) modern deep learning algorithms pre-trained on real data behave similarly in real and virtual worlds, and (ii) pre-training on virtual data improves performance. As the gap between real and virtual worlds is small, virtual worlds enable measuring the impact of various weather and imaging conditions on recognition performance, all other things being equal. We show these factors may affect drastically otherwise high-performing deep models for tracking.

cs.CV↗

arXiv2015

Distinguishing general relativity and $f(R)$ gravity with the gravitational lensing Minkowski functionals

We explore the Minkowski functionals of weak lensing convergence map to distinguish between $f(R)$ gravity and the general relativity (GR). The mock weak lensing convergence maps are constructed with a set of high-resolution simulations assuming different gravity models. It is shown that the lensing MFs of $f(R)$ gravity can be considerably different from that of GR because of the environmentally dependent enhancement of structure formation. We also investigate the effect of lensing noise on our results, and find that it is likely to distinguish F5, F6 and GR gravity models with a galaxy survey of $\sim3000$ degree$^2$ and with a background source number density of $n_g=30~{\rm arcmin}^{-2}$, comparable to an upcoming survey dark energy survey (DES). We also find that the $f(R)$ signal can be partially degenerate with the effect of changing cosmology, but combined use of other observations, such as the cosmic microwave background (CMB) data, can help break this degeneracy.

astro-ph.CO↗