arXiv subjects

Zhuo Wang

Publications and source records attributed to Zhuo Wang.

At least 91 records · Page 5Linked to original sources

arXiv2019

ColluEagle: Collusive review spammer detection using Markov random fields

Product reviews are extremely valuable for online shoppers in providing purchase decisions. Driven by immense profit incentives, fraudsters deliberately fabricate untruthful reviews to distort the reputation of online products. As online reviews become more and more important, group spamming, i.e., a team of fraudsters working collaboratively to attack a set of target products, becomes a new fashion. Previous works use review network effects, i.e. the relationships among reviewers, reviews, and products, to detect fake reviews or review spammers, but ignore time effects, which are critical in characterizing group spamming. In this paper, we propose a novel Markov random field (MRF)-based method (ColluEagle) to detect collusive review spammers, as well as review spam campaigns, considering both network effects and time effects. First we identify co-review pairs, a review phenomenon that happens between two reviewers who review a common product in a similar way, and then model reviewers and their co-review pairs as a pairwise-MRF, and use loopy belief propagation to evaluate the suspiciousness of reviewers. We further design a high quality yet easy-to-compute node prior for ColluEagle, through which the review spammer groups can also be subsequently identified. Experiments show that ColluEagle can not only detect collusive spammers with high precision, significantly outperforming state-of-the-art baselines --- FraudEagle and SpEagle, but also identify highly suspicious review spammer campaigns.

cs.IR↗

arXiv2018

PACT: Parameterized Clipping Activation for Quantized Neural Networks

Deep learning algorithms achieve high classification accuracy at the expense of significant computation cost. To address this cost, a number of quantization schemes have been proposed - but most of these techniques focused on quantizing weights, which are relatively smaller in size compared to activations. This paper proposes a novel quantization scheme for activations during training - that enables neural networks to work well with ultra low precision weights and activations without any significant accuracy degradation. This technique, PArameterized Clipping acTivation (PACT), uses an activation clipping parameter $α$ that is optimized during training to find the right quantization scale. PACT allows quantizing activations to arbitrary bit precisions, while achieving much better accuracy relative to published state-of-the-art quantization schemes. We show, for the first time, that both weights and activations can be quantized to 4-bits of precision while still achieving accuracy comparable to full precision networks across a range of popular models and datasets. We also show that exploiting these reduced-precision computational units in hardware can enable a super-linear improvement in inferencing performance due to a significant reduction in the area of accelerator compute engines coupled with the ability to retain the quantized model and activation data in on-chip memories.

cs.CV↗

arXiv2018

Bridging the Accuracy Gap for 2-bit Quantized Neural Networks (QNN)

Deep learning algorithms achieve high classification accuracy at the expense of significant computation cost. In order to reduce this cost, several quantization schemes have gained attention recently with some focusing on weight quantization, and others focusing on quantizing activations. This paper proposes novel techniques that target weight and activation quantizations separately resulting in an overall quantized neural network (QNN). The activation quantization technique, PArameterized Clipping acTivation (PACT), uses an activation clipping parameter $α$ that is optimized during training to find the right quantization scale. The weight quantization scheme, statistics-aware weight binning (SAWB), finds the optimal scaling factor that minimizes the quantization error based on the statistical characteristics of the distribution of weights without the need for an exhaustive search. The combination of PACT and SAWB results in a 2-bit QNN that achieves state-of-the-art classification accuracy (comparable to full precision networks) across a range of popular models and datasets.

cs.CV↗

arXiv2018

Enabling Quality Control for Entity Resolution: A Human and Machine Cooperation Framework

Even though many machine algorithms have been proposed for entity resolution, it remains very challenging to find a solution with quality guarantees. In this paper, we propose a novel HUman and Machine cOoperation (HUMO) framework for entity resolution (ER), which divides an ER workload between the machine and the human. HUMO enables a mechanism for quality control that can flexibly enforce both precision and recall levels. We introduce the optimization problem of HUMO, minimizing human cost given a quality requirement, and then present three optimization approaches: a conservative baseline one purely based on the monotonicity assumption of precision, a more aggressive one based on sampling and a hybrid one that can take advantage of the strengths of both previous approaches. Finally, we demonstrate by extensive experiments on real and synthetic datasets that HUMO can achieve high-quality results with reasonable return on investment (ROI) in terms of human cost, and it performs considerably better than the state-of-the-art alternatives in quality control.

cs.DB↗

arXiv2018

Liquid-solid surface phase transformation of fluorinated fullerene on monolayer tungsten diselenide

Hybrid van der Waals (vdW) heterostructures constructed by the integration of organic molecules and two-dimensional (2D) transition metal dichalcogenide (TMD) materials have useful tunable properties for flexible electronic devices. Due to the chemically inert and atomically smooth nature of the TMD surface, well-defined crystalline organic films form atomically sharp interfaces facilitating optimal device performance. Here, the surface phase transformation of the supramolecular packing structure of fluorinated fullerene (C60F48) on single-layer (SL) tungsten diselenide (WSe2) is revealed by low-temperature (LT) scanning tunnelling microscopy (STM), from thermally stable liquid to solid phases as the coverage increases. Statistical analysis of the intermolecular interaction potential reveals that the repulsive dipole-dipole interaction induced by interfacial charge transfer and substrate-mediated interactions play important roles in stabilizing the liquid C60F48 phases. Theoretical calculations further suggest that the dipole moment per C60F48 molecule varies with the surface molecule density, and the liquid-solid transformation could be understood from the perspective of the thermodynamic free energy for open systems. This study offers new insights into the growth behaviour at 2D organic/TMD hybrid heterointerfaces.

physics.app-ph↗

arXiv2018

Vapor-Liquid-Solid Growth of Monolayer MoS2 Nanoribbons

Chemical vapor deposition (CVD) of two-dimensional (2D) materials such as monolayer MoS2 typically involves the conversion of vapor-phase precursors to a solid product in a process that may be described as a vapor-solid-solid (VSS) mode. Here, we report the first demonstration of vapor-liquid-solid (VLS) growth of monolayer MoS2 yielding highly crystalline ribbon-shaped structures with a width of a few tens of nanometers to a few micrometers. The VLS growth mode is triggered by the reaction between molybdenum oxide and sodium chloride, which results in the formation of molten Na-Mo-O droplets. These droplets mediate the growth of MoS2 ribbons in the "crawling mode" when saturated with sulfur on a crystalline substrate. Our growth yields straight and kinked ribbons with a locally well-defined orientation, reflecting the regular horizontal motion of the liquid droplets during growth. Using atomic-resolution scanning transmission electron microscopy (STEM) and second harmonic generation (SHG) microscopy, we show that the ribbons are homoepitaxially on monolayer MoS2 surface with predominantly 2H- or 3R-type stacking. These findings pave the way to novel devices with structures of mixed dimensionalities.

cond-mat.mtrl-sci↗

arXiv2017

Polarization-multiplexed, dual-beam swept source optical coherence tomography angiography

A polarization-multiplexed, dual-beam setup is proposed to expand the field of view for a swept source optical coherence tomography angiography (OCTA) system. This method used a Wollaston prism to split sample path light into two orthogonal-polarized beams. This allowed two beams to shine on the cornea at an angle separation of ~ 14 degrees, which led to a separation of ~ 4.2 mm on the retina. A 3-mm glass plate was inserted into one of the beam paths to set a constant path length difference between the two polarized beams so the interferogram from the two beams are coded at different frequency bands. The resulting OCTA images from the two beams were coded with a depth separation of ~ 2 mm. 5x5 mm2 angiograms from the two beams were obtained simultaneously in 4 seconds. The two angiograms then were montaged to get a wider field of view (FOV) of ~ 5x9.2 mm2.

physics.med-ph↗

arXiv2017

From anti-perovskite to double anti-perovskite: lattice chemistry basis for super-fast transportation of Li+ ions in cubic solid lithium halogen-chalcogenides

Using a materials genome approach on the basis of the density functional theory, we have formulated a new class of inorganic electrolytes for fast diffusion of Li+ ions, through fine-tuning of lattice chemistry of anti-perovskite structures. Systematic modelling has been carried out to elaborate the structural stability and ion transportation characteristics in Li3AX based cubic anti-perovskite, through alloying on the chalcogen lattice site (A) and alternative occupancy of the halogen site (X). In addition to identifying effective ways for reduction of diffusion barriers for Li+ ions in anti-perovskite phases via suitable designation of lattice occupancy, the current theoretical study leads to discovery and synthesis of a new phase with a double-anti-perovskite structure, Li6OSI2 (or Li3O0.5S0.5I). Such a new compound is of fairly low activation barrier for Li+ diffusion, together with a wide energy band gap to hinder conduction of electrons.

physics.chem-ph↗

arXiv2017

Theoretical Design of Solid Electrolytes with Superb Ionic Conductivity: Alloying Effect on Li+ Transportation in Cubic Li6PA5X Chalcogenides

It is of great importance to develop inorganic solid electrolytes with high ionic conductivity, thus enabling solid state Li-ion batteries to address the notorious safety issue about the current technology due to use of highly flammable liquid organic electrolytes. On the basis of systematic first principles modelling, we have formulated new inorganic electrolytes with ultra-low activation energies for long-distance diffusion of Li+ ions, through alloying in the cubic Li6PA5X chalcogenides (chalcogen A; halogen X). We find that the long-distance transportation of Li+ is dictated by inter-octahedral diffusion, as the activation energy for Li+ to migrate over a Li6A octahedron is minimal. The inter-octahedral diffusion barrier for Li+ is largely dependent on the interaction with chalcogen anions in the compound. Radical reduction of diffusion barrier for Li+ ions can be realized through isovalent substitution of S using elements of lower electronegativity, together with smaller halogen ions on the X site.

physics.chem-ph↗

arXiv2016

Modulational instability in nonlocal Kerr media with sine-oscillatory response

We discuss the modulational instability (MI) of plane waves in nonlocal Kerr media with the sine-oscillation response function, which can model the nematic liquid crystal with negative dielectric anisotropy. The results in the framework of the nonlocal nonlinear Schrödinger equation show that the MI in this system has two extraordinary properties. First, MI exists in both cases of negative Kerr coefficient $(n_{2}<0)$ and positive Kerr coefficient $(n_{2}>0)$. Second, the maximum gain of MI is not affected by the intensity of the plane wave, and the wave number %spatial frequency of the maximum gain depends on the oscillation period of the response function. We also explore, for the first time to the best of our knowledge, the physical mechanism of MI in the framework of the four-wave mixing (FWM). %FWM interaction. By introducing a phase mismatch term $\triangle k$ and a (nonlinear) growth factor $γ$ (both of them have clear physical meanings), we find that the necessary and sufficient condition for the MI appearance is that the phase mismatching during the FWM process should be small enough such that $|\triangle k|<2|γ|$. Basing on the discover, we present a new scenario to uniformly and consistently explain why the MI can happen for the focusing local Kerr media and cannot for the defocusing local Kerr media, as well as the MI phenomena in the nonlocal cases.

nlin.PS↗

arXiv2016

Comparative study of microwave radiation-induced magnetoresistive oscillations induced by circularly- and linearly- polarized photo-excitation

A comparative study of the radiation-induced magnetoresistance oscillations in the high mobility GaAs/AlGaAs heterostructure two dimensional electron system (2DES) under linearly- and circularlypolarized microwave excitation indicates a profound difference in the response observed upon rotating the microwave launcher for the two cases, although circularly polarized microwave radiation induced magnetoresistance oscillations observed at low magnetic fields are similar to the oscillations observed with linearly polarized radiation. For the linearly polarized radiation, the magnetoresistive response is a strong sinusoidal function of the launcher rotation (or linear polarization) angle, θ. For circularly polarized radiation, the oscillatory magnetoresistive response is hardly sensitive to θ.

cond-mat.mes-hall↗

arXiv2016

Superconducting-contact-induced resistance-anomalies in the 3D topological insulator Bi2Te3

This study examines the magnetotransport response observed in flakes of the 3D topological insulator (TI) Bi2Te3, including indium superconducting electrodes, and demonstrates two critical transitions in the magnetoresistive response with decreasing temperatures below T = 3.4K. The first transition is attributed to superconductivity in the indium electrodes, and the second transition, with a critical field exceeding the transition field of indium, is attributed to a proximity effect at the 2D planar interface of this hybrid TI/superconductor structure.

cond-mat.mes-hall↗

arXiv2015

Transition between self-focusing and self-defocusing in nonlocally nonlinear media

We reveal the relevance between the nonlocality and the focusing/defocusing states in nonlocally nonlinear media, and predict a novel phenomenon that the self-focusing/self-defocusing property of the optical beam in the nonlocally nonlinear medium with a sine-oscillation response function depends on its degree of nonlocality. The transition from the focusing nonlinearity to the defocusing nonlinearity of the nonlinear refractive index will happen when the degree of nonlocality of the system goes cross a critical value, and vise verse. Bright and dark soliton solutions are obtained, respectively, in the focusing state and in the defocusing state, and their stabilities are also discussed. It is mentioned that such a phenomenon might be experimentally realized in the nematic liquid crystal with negative dielectric anisotropy or in the quadratic nonlinear medium.

physics.optics↗

arXiv2015

Optimization and analysis of large scale data sorting algorithm based on Hadoop

When dealing with massive data sorting, we usually use Hadoop which is a framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models. A common approach in implement of big data sorting is to use shuffle and sort phase in MapReduce based on Hadoop. However, if we use it directly, the efficiency could be very low and the load imbalance can be a big problem. In this paper we carry out an experimental study of an optimization and analysis of large scale data sorting algorithm based on hadoop. In order to reach optimization, we use more than 2 rounds MapReduce. In the first round, we use a MapReduce to take sample randomly. Then we use another MapReduce to order the data uniformly, according to the results of the first round. If the data is also too big, it will turn back to the first round and keep on. The experiments show that, it is better to use the optimized algorithm than shuffle of MapReduce to sort large scale data.

cs.DC↗

arXiv2015

Observation of Optical Solitons and Abnormal Modulation Instability in Liquid Crystals with Negative Dielectric Anisotropy

We investigate theoretically and experimentally the optical beam propagation in the nematic liquid crystal with negative dielectric anisotropy, which is aligned homeotropically in a $80μm$-thickness planar cell in the presence of an externally voltage. It is predicted that the nonlocal nonlinearity of liquid crystal undergo an oscillatory response function with a negative nonlinear refractive index coefficient. We found that the oscillatory nonlocal nonlinearity can support stable bright solitons, which are observed in experiment. We also found that abnormal modulation instability occurs with infinity gain coefficient at a fixed spatial frequency, which is no depend on the beam intensity. We observed the modulation instability in the liquid crystal at a very low intensity ($0.26W/cm^2$), and the maximum gain frequency were found kept unchange when beam power changes over 2-3 orders of magnitude.

nlin.PS↗

arXiv2012

1/2+1/2>1 for quantum error correction

Quantum error correction assisted by entanglement helps to transmit the encoded qudits through quantum channels with some of them being noiseless. Here we consider a more realistic scheme for experiments what we called as partial-noisy quantum channels in which, instead of completely free of noise, only part of the qudit suffers no noise. In this scenario we show by an explicit example that two half-noisy channels work better than one noiseless channel, a queer phenomenon showing 1/2+1/2>1. Our example also saturates a unified quantum Singleton bound, valid for the standard and entanglement-assisted codes as well. Furthermore, as illustrated by a mixed-alphabet code with half-noisy channels, a higher dimensional physical qudit can so improve the performance of a partial-noisy channel that it even outperforms a noiseless channel.

quant-ph↗

arXiv2012

Quantum Error-Correcting Codes over Mixed Alphabets

Errors are inevitable during all kinds quantum informational tasks and quantum error-correcting codes (QECCs) are powerful tools to fight various quantum noises. For standard QECCs physical systems have the same number of energy levels. Here we shall propose QECCs over mixed alphabets, i.e., physical systems of different dimensions, and investigate their constructions as well as their quantum Singleton bound. We propose two kinds of constructions: a graphical construction based a graph-theoretical object composite coding clique and a projection-based construction. We illustrate our ideas using two alphabets by finding out some 1-error correcting or detecting codes over mixed alphabets, e.g., optimal $((6,8,3))_{4^52^1}$, $((6,4,3))_{4^42^2}$ and $((5,16,2))_{4^32^2}$ code and suboptimal $((5,9,2))_{3^42^1}$ code. Our methods also shed light to the constructions of standard QECCs, e.g., the construction of the optimal $((6,16,3))_4$ code as well as the optimal $((2n+3,p^{2n+1},2))_{p}$ codes with $p=4k$.

quant-ph↗

arXiv2011

Single Crystal Growth of Magnesium Diboride by Using Low Melting Point Alloy Flux

Magnesium diboride crystals were grown with low melting alloy flux. The sample with superconductivity at 38.3 K was prepared in the zinc-magnesium flux and another sample with superconductivity at 34.11 K was prepared in the cadmium-magnesium flux. MgB4 was found in both samples, and MgB4's magnetization curve being above zero line had not superconductivity. Magnesium diboride prepared by these two methods is all flaky and irregular shape and low quality of single crystal. However, these two metals should be the optional flux for magnesium diboride crystal growth. The study of single crystal growth is very helpful for the future applications.

cond-mat.supr-con↗