arXiv subjects

Feng Dong

Publications and source records attributed to Feng Dong.

28 records · Page 2Linked to original sources

arXiv2020

MadDroid: Characterising and Detecting Devious Ad Content for Android Apps

Advertisement drives the economy of the mobile app ecosystem. As a key component in the mobile ad business model, mobile ad content has been overlooked by the research community, which poses a number of threats, e.g., propagating malware and undesirable contents. To understand the practice of these devious ad behaviors, we perform a large-scale study on the app contents harvested through automated app testing. In this work, we first provide a comprehensive categorization of devious ad contents, including five kinds of behaviors belonging to two categories: \emph{ad loading content} and \emph{ad clicking content}. Then, we propose MadDroid, a framework for automated detection of devious ad contents. MadDroid leverages an automated app testing framework with a sophisticated ad view exploration strategy for effectively collecting ad-related network traffic and subsequently extracting ad contents. We then integrate dedicated approaches into the framework to identify devious ad contents. We have applied MadDroid to 40,000 Android apps and found that roughly 6\% of apps deliver devious ad contents, e.g., distributing malicious apps that cannot be downloaded via traditional app markets. Experiment results indicate that devious ad contents are prevalent, suggesting that our community should invest more effort into the detection and mitigation of devious ads towards building a trustworthy mobile advertising ecosystem.

cs.CR↗

arXiv2018

FraudDroid: Automated Ad Fraud Detection for Android Apps

Although mobile ad frauds have been widespread, state-of-the-art approaches in the literature have mainly focused on detecting the so-called static placement frauds, where only a single UI state is involved and can be identified based on static information such as the size or location of ad views. Other types of fraud exist that involve multiple UI states and are performed dynamically while users interact with the app. Such dynamic interaction frauds, although now widely spread in apps, have not yet been explored nor addressed in the literature. In this work, we investigate a wide range of mobile ad frauds to provide a comprehensive taxonomy to the research community. We then propose, FraudDroid, a novel hybrid approach to detect ad frauds in mobile Android apps. FraudDroid analyses apps dynamically to build UI state transition graphs and collects their associated runtime network traffics, which are then leveraged to check against a set of heuristic-based rules for identifying ad fraudulent behaviours. We show empirically that FraudDroid detects ad frauds with a high precision (93%) and recall (92%). Experimental results further show that FraudDroid is capable of detecting ad frauds across the spectrum of fraud types. By analysing 12,000 ad-supported Android apps, FraudDroid identified 335 cases of fraud associated with 20 ad networks that are further confirmed to be true positive results and are shared with our fellow researchers to promote advanced ad fraud detection

cs.CR↗

arXiv2011

An Optical Catalog of Galaxy Clusters Obtained from an Adaptive Matched Filter Finder Applied to SDSS DR6

We present a new cluster catalog extracted from the Sloan Digital Sky Survey Data Release 6 (SDSS DR6) using an adaptive matched filter (AMF) cluster finder. We identify 69,173 galaxy clusters in the redshift range 0.045 $\le z <$ 0.78 in 8420 sq. deg. of the sky. We provide angular position, redshift, richness, core and virial radii estimates for these clusters, as well as an error analysis for each of these quantities. We also provide a catalog of more than 205,000 galaxies representing the three brightest galaxies in the $r$ band which are possible BCG candidates. We show basic properties of the BCG candidates and study how their luminosity scales in redshift and cluster richness. We compare our catalog with the maxBCG and GMBCG catalogs, as well as with that of Wen, Han, and Liu. We match between 30% and 50% of clusters between catalogs over all overlapping redshift ranges. We find that the percentage of matches increases with the richness for all catalogs. We cross match the AMF catalog with available X-ray data in the same area of the sky and find 539 matches, 119 of which with temperature measurements. We present scaling relations between optical and X-ray properties and cluster center comparison. We find that both $Λ_{200}$ and $R_{200}$ correlate well with both $L_X$ and $T_X$, with no significant difference in trend if we restrict the matches to flux-limited X-ray samples.

astro-ph.CO↗

arXiv2010

Alignment of Brightest Cluster Galaxies with their Host Clusters

We examine the alignment between Brightest Cluster Galaxies (BCGs) and their host clusters in a sample of 7031 clusters with 0.08 0.65 mag brighter than the mean of the second and third ranked galaxies) show stronger alignment than do clusters with less dominant BCGs at the 4.4 sigma level. Rich clusters show a stronger alignment than do poor clusters at the 2.3 sigma level. Low redshift clusters (z<0.26) show more alignment than do high redshift (z>0.26) clusters, with a difference significant at the 3.0 sigma level. Our results do not depend on the algorithm used to select the cluster sample, suggesting that they are not biased by systematics of either algorithm. The correlation between BCG dominance and cluster alignment may be a consequence of the hierarchical merging process which forms the cluster. The observed redshift evolution may follow from secondary infall at late redshifts.

astro-ph.CO↗

arXiv2007

Optical Cluster-Finding with An Adaptive Matched-Filter Technique: Algorithm and Comparison with Simulations

We present a modified adaptive matched filter algorithm designed to identify clusters of galaxies in wide-field imaging surveys such as the Sloan Digital Sky Survey. The cluster-finding technique is fully adaptive to imaging surveys with spectroscopic coverage, multicolor photometric redshifts, no redshift information at all, and any combination of these within one survey. It works with high efficiency in multi-band imaging surveys where photometric redshifts can be estimated with well-understood error distributions. Tests of the algorithm on realistic mock SDSS catalogs suggest that the detected sample is ~85% complete and over 90% pure for clusters with masses above 1.0*10^{14} h^{-1} M_solar and redshifts up to z=0.45. The errors of estimated cluster redshifts from maximum likelihood method are shown to be small (typically less that 0.01) over the whole redshift range with photometric redshift errors typical of those found in the Sloan survey. Inside the spherical radius corresponding to a galaxy overdensity of Delta=200, we find the derived cluster richness Lambda_{200} a roughly linear indicator of its virial mass M_{200}, which well recovers the relation between total luminosity and cluster mass of the input simulation.

astro-ph↗

arXiv2004

Properties of Cold Dark Matter Halos at z>6

We compute the properties of dark matter halos with mass $10^{6.5}-10^9\msun$ at redshift $z=6-11$ in the standard cold dark matter cosmological model, utilizing a very high resolution N-body simulation. We find that dark matter halos in these mass and redshift ranges are significantly biased over matter with a bias factor in the range 2-6. The dark matter halo mass function displays a slope of $2.05\pm 0.15$ at the small mass end. We do not find a universal dark matter density profile. Instead, we find a significant dependence of the central density profile of dark matter halos on halo mass and epoch with $α_0=0.4-1.0$; the high-mass ($M\ge 10^8\msun$) low-redshift ($z\sim 6$) halos occupy the high end of the range and low-mass ($M\sim 10^{7}\msun$) high-redshift ($z\sim 11$) halos occupy the low end. Additionally, for fixed mass and epoch there is a significant dispersion in $α_0$ due to the stochastic assembly of halos. Our results fit a relationship of the form $α_0=0.75((1+z)/7.0)^{-1.25}(M/10^7\msun)^{0.11(1+z)/7.0}$ with a dispersion about this fit of $\pm 0.5$ and no systematic dependence of variance correlated with environment. The median spin parameter of dark matter halos is $0.03-0.04$ but with a large lognormal dispersion of $\sim 0.4$. Various quantities are tabulated or fitted with empirical formulae.

astro-ph↗

arXiv2003

The Richness-Dependent Cluster Correlation Function: Early SDSS Data

The cluster correlation function and its richness dependence are determined from 1108 clusters of galaxies -- the largest sample of clusters studied so far -- found in 379 deg^2 of Sloan Digital Sky Survey early data. The results are compared with previous samples of optically and X-ray selected clusters. The richness-dependent correlation function increases monotonically from an average correlation scale of ~ 12 h^{-1} Mpc for poor clusters to ~ 25 h^{-1} Mpc for the richer, more massive clusters with a mean separation of ~ 90 h^{-1} Mpc. X-ray selected clusters suggest slightly stronger correlations than optically selected clusters (~ 2-σ). The results are compared with large-scale cosmological simulations. The observed richness-dependent cluster correlation function is well represented by the standard flat LCDM model (Ω_m ~= 0.3, h ~= 0.7), and is inconsistent with the considerably weaker correlations predicted by Ω_m = 1 models. An analytic relation for the correlation scale versus cluster mean separation, r_0 - d, that best describes the observations and the LCDM prediction is r_0 ~= 2.6 \sqrt{d} (for d ~= 20 - 90 h^{-1} Mpc). Data from the complete Sloan Digital Sky Survey, when available, will greatly enhance the accuracy of the results and allow a more precise determination of cosmological parameters.

astro-ph↗

arXiv2003

Evolution of the Cluster Correlation Function

We study the evolution of the cluster correlation function and its richness-dependence from z = 0 to z = 3 using large-scale cosmological simulations. A standard flat LCDM model with Ω_m = 0.3 and, for comparison, a tilted Ω_m = 1 model, TSCDM, are used. The evolutionary predictions are presented in a format suitable for direct comparisons with observations. We find that the cluster correlation strength increases with redshift: high redshift clusters are clustered more strongly (in comoving scale) than low redshift clusters of the same mass. The increased correlations with redshift, in spite of the decreasing mass correlation strength, is caused by the strong increase in cluster bias with redshift: clusters represent higher density peaks of the mass distribution as the redshift increases. The richness-dependent cluster correlation function, presented as the correlation-scale versus cluster mean separation relation, R_0 - d, is found to be, remarkably, independent of redshift to z <~ 2 for LCDM and z <~ 1 for TCDM (for a fixed correlation function slope and cluster mass within a fixed comoving radius). The non-evolving R_0 - d relation implies that both the comoving clustering scale and the cluster mean separation increase with redshift for the same mass clusters so that the R_0 - d relation remains essentially unchanged. The evolution of the R_0 - d relation from z ~ 0 to z ~ 3 provides an important new tool in cosmology; it can be used to break degeneracies that exist at z ~ 0 and provide precise determination of cosmological parameters.

astro-ph↗

arXiv2003

A Merged Catalog of Clusters of Galaxies from Early SDSS Data

We present a catalog of 799 clusters of galaxies in the redshift range z_est = 0.05 - 0.3 selected from ~400 deg^2 of early SDSS commissioning data along the celestial equator. The catalog is based on merging two independent selection methods -- a color-magnitude red-sequence maxBCG technique (B), and a Hybrid Matched-Filter method (H). The BH catalog includes clusters with richness Λ>= 40 (Matched-Filter) and N_gal >= 13 (maxBCG), corresponding to typical velocity dispersion of σ_v >~ 400 km s^{-1} and mass (within 0.6 h^{-1) Mpc radius) >~ 5*10^{13} h^{-1} M_sun. This threshold is below Abell richness class 0 clusters. The average space density of these clusters is 2*10^{-5} h^3 Mpc^{-3}. All NORAS X-ray clusters and 53 of the 58 Abell clusters in the survey region are detected in the catalog; the 5 additional Abell clusters are detected below the BH catalog cuts. The cluster richness function is determined and found to exhibit a steeply decreasing cluster abundance with increasing richness. We derive observational scaling relations between cluster richness and observed cluster luminosity and cluster velocity dispersion; these scaling relations provide important physical calibrations for the clusters. The catalog can be used for studies of individual clusters, for comparisons with other sources such as X-ray clusters and AGNs, and, with proper correction for the relevant selection functions, also for statistical analyses of clusters.

astro-ph↗

arXiv2002

The Cluster Mass Function from Early SDSS Data: Cosmological Implications

The mass function of clusters of galaxies is determined from 400 deg^2 of early commissioning imaging data of the Sloan Digital Sky Survey; ~300 clusters in the redshift range z = 0.1 - 0.2 are used. Clusters are selected using two independent selection methods: a Matched Filter and a red-sequence color magnitude technique. The two methods yield consistent results. The cluster mass function is compared with large-scale cosmological simulations. We find a best-fit cluster normalization relation of sigma_8*omega_m^0.6 = 0.33 +- 0.03 (for 0.1 ~< omega_m ~< 0.4), or equivalently sigma_8 = (0.16/omega_m)^0.6. The amplitude of this relation is significantly lower than the previous canonical value, implying that either omega_m is lower than previously expected (omega_m = 0.16 if sigma_8 = 1) or sigma_8 is lower than expected (sigma_8 = 0.7 if omega_m = 0.3). The best-fit mass function parameters are omega_m = 0.19 (+0.08,-0.07) and sigma_8 = 0.9 (+0.3,-0.2). High values of omega_m (>= 0.4) and low sigma_8 (=< 0.6) are excluded at >~ 2 sigma.

astro-ph↗