arXiv subjects

Xiao Shen

Publications and source records attributed to Xiao Shen.

At least 55 records · Page 3Linked to original sources

arXiv2022

Comparison of Update and Genetic Training Algorithms in a Memristor Crossbar Perceptron

Memristor-based computer architectures are becoming more attractive as a possible choice of hardware for the implementation of neural networks. However, at present, memristor technologies are susceptible to a variety of failure modes, a serious concern in any application where regular access to the hardware may not be expected or even possible. In this study, we investigate whether certain training algorithms may be more resilient to particular hardware failure modes, and therefore more suitable for use in those applications. We implement two training algorithms -- a local update scheme and a genetic algorithm -- in a simulated memristor crossbar, and compare their ability to train for a simple image classification task as an increasing number of memristors fail to adjust their conductance. We demonstrate that there is a clear distinction between the two algorithms in several measures of the rate of failure to train.

physics.comp-ph↗

arXiv2021

Optical and Electronic Properties of SiTe_x(x=1,2) from First-Principles

The optical and electronic properties of the α-SiTe, β-SiTe, and RX-SiTe_2 are investigated. A detailed analysis of electronic properties is done using standard density functional theory (DFT) and hybrid functional (HSE06) methods. The optical dielectric properties are studied under three different methods: standard DFT, many-body Green's functions (GW), and Bethe-Salpeter equation (BSE). Our calculations show that the SiTe compounds possess extremely high static dielectric constants in their bulk forms (ε_0({\bot}) = 68.58, ε_0({\parallel}) = 127.29 for α-SiTe, and ε_0({\bot}) = 76.23, ε_0({\parallel}) = 98.15 for β-SiTe). The frequency-dependent dielectric functions Im(ε) have very large values (>100) in the optical regime, which are among the highest of layered materials, suggesting them as excellent light absorbents in the corresponding frequencies. α-SiTe exhibits a high degree of optical anisotropy as compared to the other two compounds, consistent with their structural configurations. A strong interlayer excitonic effect is observed in bulk RX-SiTe_2. In addition, an analysis of Raman intensity is also performed.

cond-mat.mtrl-sci↗

arXiv2021

Pressure-Induced Insulator-Metal Transition in Silicon Telluride from First-Principles Calculations

Silicon telluride (Si2Te3) is a two-dimensional semiconductor with unique structural properties due to the size contrast between Si and Te atoms. A recent experiment shows that the material turns metallic under hydrostatic pressure, while the lattice structure of the metallic phase remains to be identified. In this paper, we propose two metallic phases, M1 and M2, of Si2Te3 using the evolution algorithm and first-principles density functional theory (DFT) calculations. Unlike the presence of Si-Si dimers in the semiconducting (SC) phase, both M1 and M2 phases have individual Si atoms, which play important roles in the metallicity. Analysis of structural properties, electronic properties, dynamical as well as thermal stability is performed. The energies of these new structures are compared with the SC phase under the subsequent hydrostatic pressure up to 12 GPa. The results show that M1 and M2 phases have lower energies under high pressure, thus elucidating the appearance of the metallic phase of Si2Te3. In addition, the external pressure causes the SC phase to have an indirect-direct-indirect bandgap transition. Analysis of Raman spectra of the SC phase at a different pressure shows the shifting of the major Raman peaks, and finally disappearing confirms the phase transition. The results are in good agreement with the experimental observations. The understanding of the insulator-metal phase transition increases the potential usefulness of the material system.

cond-mat.mtrl-sci↗

arXiv2020

Against Adversarial Learning: Naturally Distinguish Known and Unknown in Open Set Domain Adaptation

Open set domain adaptation refers to the scenario that the target domain contains categories that do not exist in the source domain. It is a more common situation in the reality compared with the typical closed set domain adaptation where the source domain and the target domain contain the same categories. The main difficulty of open set domain adaptation is that we need to distinguish which target data belongs to the unknown classes when machine learning models only have concepts about what they know. In this paper, we propose an "against adversarial learning" method that can distinguish unknown target data and known data naturally without setting any additional hyper parameters and the target data predicted to the known classes can be classified at the same time. Experimental results show that the proposed method can make significant improvement in performance compared with several state-of-the-art methods.

cs.LG↗

arXiv2020

Deep Adversarial Domain Adaptation Based on Multi-layer Joint Kernelized Distance

Domain adaptation refers to the learning scenario that a model learned from the source data is applied on the target data which have the same categories but different distribution. While it has been widely applied, the distribution discrepancy between source data and target data can substantially affect the adaptation performance. The problem has been recently addressed by employing adversarial learning and distinctive adaptation performance has been reported. In this paper, a deep adversarial domain adaptation model based on a multi-layer joint kernelized distance metric is proposed. By utilizing the abstract features extracted from deep networks, the multi-layer joint kernelized distance (MJKD) between the $j$th target data predicted as the $m$th category and all the source data of the $m'$th category is computed. Base on MJKD, a class-balanced selection strategy is utilized in each category to select target data that are most likely to be classified correctly and treat them as labeled data using their pseudo labels. Then an adversarial architecture is used to draw the newly generated labeled training data and the remaining target data close to each other. In this way, the target data itself provide valuable information to enhance the domain adaptation. An analysis of the proposed method is also given and the experimental results demonstrate that the proposed method can achieve a better performance than a number of state-of-the-art methods.

cs.LG↗

arXiv2020

Coalescence estimates for the corner growth model with exponential weights

We establish estimates for the coalescence time of semi-infinite directed geodesics in the planar corner growth model with i.i.d. exponential weights. There are four estimates: upper and lower bounds on the probabilities of both fast and slow coalescence on the correct spatial scale with exponent $3/2$. Our proofs utilize a geodesic duality introduced by Pimentel and properties of the increment-stationary last-passage percolation process. For fast coalescence our bounds are new and they have matching optimal exponential order of magnitude. For slow coalescence we reproduce bounds proved earlier with integrable probability inputs, except that our upper bound misses the optimal order by a logarithmic factor.

math.PR↗

arXiv2020

Temperature- and Polarization- Dependent Optical Properties of Single Si2Te3 Nanoplates

We report a combined experimental and computational study of the optical properties of individual silicon telluride (Si2Te3) nanoplates. The p-type semiconductor Si2Te3 has a unique layered crystal structure with hexagonal closed-packed Te sublattices and Si-Si dimers occupying octahedral intercalation sites. The orientation of the silicon dimers leads to unique optical and electronic properties. Two-dimensional Si2Te3 nanoplates with thicknesses of hundreds of nanometers and lateral sizes of tens of micrometers are synthesized by a chemical vapor deposition technique. At temperatures below 150 K, the Si2Te3 nanoplates exhibit a direct band structure with a band gap energy of 2.394 eV at 7 K and an estimated free exciton binding energy of 150 meV. Polarized reflection measurements at different temperatures show anisotropy in the absorption coefficient due to an anisotropic orientation of the silicon dimers, which is in excellent agreement with theoretical calculations of the dielectric functions. Polarized Raman measurements of single Si2Te3 nanoplates at different temperatures reveal various vibrational modes, which agree with density functional perturbation theory calculations. The unique structural and optical properties of nanostructured Si2Te3 hold great potential applications in optoelectronics and chemical sensing.

cond-mat.mtrl-sci↗

arXiv2020

Network Together: Node Classification via Cross network Deep Network Embedding

Network embedding is a highly effective method to learn low-dimensional node vector representations with original network structures being well preserved. However, existing network embedding algorithms are mostly developed for a single network, which fail to learn generalized feature representations across different networks. In this paper, we study a cross-network node classification problem, which aims at leveraging the abundant labeled information from a source network to help classify the unlabeled nodes in a target network. To succeed in such a task, transferable features should be learned for nodes across different networks. To this end, a novel cross-network deep network embedding (CDNE) model is proposed to incorporate domain adaptation into deep network embedding so as to learn label-discriminative and network-invariant node vector representations. On one hand, CDNE leverages network structures to capture the proximities between nodes within a network, by mapping more strongly connected nodes to have more similar latent vector representations. On the other hand, node attributes and labels are leveraged to capture the proximities between nodes across different networks by making the same labeled nodes across networks have aligned latent vector representations. Extensive experiments have been conducted, demonstrating that the proposed CDNE model significantly outperforms the state-of-the-art network embedding algorithms in cross-network node classification.

cs.SI↗

arXiv2020

Predicting A Novel Phase of 2D SiTe$_2$

Layered IV-VI$_2$ compounds often exist in the CdI$_2$ structure. Using the evolution algorithm and first-principles calculations, we predict a novel layered structure of silicon ditelluride (SiTe$_2$) that is more stable than the CdI$_2$ phase. The structure has a triclinic unit cell in its bulk form and exhibits the competition between the Si atoms' tendency to form tetrahedral bonds and the Te atoms' tendency to form hexagonal close-packing. The electronic and vibrational properties of the predicted phase are investigated. The effective mass of electron is small among 2D semiconductors, which is beneficial for applications such as field-effect transistors. The vibrational Raman and IR spectra are calculated to facilitate future experimental investigations

cond-mat.mtrl-sci↗

arXiv2020

Metallic ferroelectric-ferromagnetic multiferroics in strained EuTiO$_{3-x}$H$_x$

Polar metals are defined by the coexistence of metallicity and polar crystal structure. They have potential applications in non-linear optics, ferroelectric devices, and quantum devices. Meanwhile, ferroelectric-ferromagnetic (FE-FM) multiferroics display simultaneous ferroelectricity and ferromagnetism, leading to new technologies in information storage. It remains an open question whether metallicity, ferroelectricity, and ferromagnetism can coexist in a single domain of a material. EuTiO$_3$ is actively studied for potential applications in magnetic sensors, memories, magneto-optical devices, and energy conversion devices. It stabilizes at a multi-critical equilibrium and exhibits a rich range of intriguing properties. Here, using the results from hybrid density functional theory calculations, we report metallic FE-FM multiferroics in strain-engineered epitaxial EuTiO$_3$ with H doping. The emergence of the magnetism in polar metals provides a new degree of freedom to control these materials in applications. The underlying mechanism for the coexistence of metallicity, ferroelectricity, and ferromagnetism is discussed. The ferromagnetism in metallic EuTiO$_{3-x}$H$_x$ is explained by the Ruderman-Kittel-Kasuya- Yosida (RKKY) interaction, which agrees with experiments. The coexistence of metallicity and ferroelectricity is allowed because the electrons at the Fermi level are weakly coupled to the ferroelectric distortion. Our results suggest that the combined effect of strain and doping is responsible for achieving EuTiO$_3$-based metallic FE-FM multiferroics and may provide a new way for obtaining metallic FE-FM multiferroics in other materials.

cond-mat.str-el↗

arXiv2020

Adversarial Deep Network Embedding for Cross-network Node Classification

In this paper, the task of cross-network node classification, which leverages the abundant labeled nodes from a source network to help classify unlabeled nodes in a target network, is studied. The existing domain adaptation algorithms generally fail to model the network structural information, and the current network embedding models mainly focus on single-network applications. Thus, both of them cannot be directly applied to solve the cross-network node classification problem. This motivates us to propose an adversarial cross-network deep network embedding (ACDNE) model to integrate adversarial domain adaptation with deep network embedding so as to learn network-invariant node representations that can also well preserve the network structural information. In ACDNE, the deep network embedding module utilizes two feature extractors to jointly preserve attributed affinity and topological proximities between nodes. In addition, a node classifier is incorporated to make node representations label-discriminative. Moreover, an adversarial domain adaptation technique is employed to make node representations network-invariant. Extensive experimental results demonstrate that the proposed ACDNE model achieves the state-of-the-art performance in cross-network node classification.

cs.SI↗

arXiv2019

Adversarial Training Methods for Network Embedding

Network Embedding is the task of learning continuous node representations for networks, which has been shown effective in a variety of tasks such as link prediction and node classification. Most of existing works aim to preserve different network structures and properties in low-dimensional embedding vectors, while neglecting the existence of noisy information in many real-world networks and the overfitting issue in the embedding learning process. Most recently, generative adversarial networks (GANs) based regularization methods are exploited to regularize embedding learning process, which can encourage a global smoothness of embedding vectors. These methods have very complicated architecture and suffer from the well-recognized non-convergence problem of GANs. In this paper, we aim to introduce a more succinct and effective local regularization method, namely adversarial training, to network embedding so as to achieve model robustness and better generalization performance. Firstly, the adversarial training method is applied by defining adversarial perturbations in the embedding space with an adaptive $L_2$ norm constraint that depends on the connectivity pattern of node pairs. Though effective as a regularizer, it suffers from the interpretability issue which may hinder its application in certain real-world scenarios. To improve this strategy, we further propose an interpretable adversarial training method by enforcing the reconstruction of the adversarial examples in the discrete graph domain. These two regularization methods can be applied to many existing embedding models, and we take DeepWalk as the base model for illustration in the paper. Empirical evaluations in both link prediction and node classification demonstrate the effectiveness of the proposed methods.

cs.LG↗

arXiv2019

Ultra-High Mechanical Flexibility of 2D Silicon Telluride

Silicon telluride (Si2Te3) is a two-dimensional material with a unique variable structure where the silicon atoms form Si-Si dimers to fill the "metal" sites between the Te layers. The Si-Si dimers have four possible orientations: three in-plane and one out-of-the plane directions. The structural variability of Si2Te3 allows unusual properties especially the mechanical properties. Using results from first-principles calculations, we show that the Si2Te3 monolayer can sustain a uniaxial tensile strain up to 38%, highest among all two-dimensional materials reported. The high mechanical flexibility allows applying mechanical strain to reduce the band gap by 1.4 eV. With increasing strain, the band gap undergoes an unusual indirect-direct-indirect-direct transition. We also show that the uniaxial strain can effectively control the Si-Si dimer alignment, which is beneficial for practical applications.

cond-mat.mtrl-sci↗

arXiv2019

Anisotropic Optical Properties of 2D Silicon Telluride From Ab Initio Calculations

Silicon telluride (Si2Te3) is a silicon-based 2D chalcogenide with potential applications in optoelectronics. It has a unique crystal structure where Si atoms form Si-Si dimers to occupy the "metal" sites. In this paper, we report an ab initio computational study of its optical dielectric properties using the GW approximation and the Bethe-Salpeter equation (BSE). Strong optical anisotropy is discovered. The imaginary part of the dielectric constant in the direction parallel to the Si-Si dimers is found to be much lower than that perpendicular to the dimers. We show this effect originates from the particular compositions of the wavefunctions in the valence and conduction bands. BSE calculations reduce GW quasiparticle band gap by 0.3 eV in bulk and 0.6 eV in monolayer, indicating a large excitonic effect in Si2Te3. Furthermore, including electron-hole interaction in bulk calculations significantly reduces the imaginary part of the dielectric constant in the out-of-plane direction, suggesting strong interlayer exciton effect in Si2Te3 multilayers.

cond-mat.mtrl-sci↗

arXiv2019

Deep Network Embedding for Graph Representation Learning in Signed Networks

Network embedding has attracted an increasing attention over the past few years. As an effective approach to solve graph mining problems, network embedding aims to learn a low-dimensional feature vector representation for each node of a given network. The vast majority of existing network embedding algorithms, however, are only designed for unsigned networks, and the signed networks containing both positive and negative links, have pretty distinct properties from the unsigned counterpart. In this paper, we propose a deep network embedding model to learn the low-dimensional node vector representations with structural balance preservation for the signed networks. The model employs a semi-supervised stacked auto-encoder to reconstruct the adjacency connections of a given signed network. As the adjacency connections are overwhelmingly positive in the real-world signed networks, we impose a larger penalty to make the auto-encoder focus more on reconstructing the scarce negative links than the abundant positive links. In addition, to preserve the structural balance property of signed networks, we design the pairwise constraints to make the positively connected nodes much closer than the negatively connected nodes in the embedding space. Based on the network representations learned by the proposed model, we conduct link sign prediction and community detection in signed networks. Extensive experimental results in real-world datasets demonstrate the superiority of the proposed model over the state-of-the-art network embedding algorithms for graph representation learning in signed networks.

cs.SI↗

arXiv2018

Temperature dependent dynamics of photoexcited carriers of Si2Te3 nanowires

We report an optical study of the dynamics of photoexcited carriers in Si2Te3 nanowires at various temperatures and excitation powers. Si2Te3 nanowires were synthesized, by using gold as a catalyst, on a silicon substrate by the chemical vapor deposition method. The photoluminescence spectrum of Si2Te3 nanowires was primary dominated by defect and surface states related emission at both low and room temperatures. We observed that the decay time of photoexcited carries was very long (> 10 ns) at low temperatures and became shorter (< 2 ns) at room temperature. Further, the carrier decay time became faster at high excitation rates. The acceleration of the photoexcited carrier decay rates indicate the thermal quenching along with the non-radiative recombination at high temperature and excitation power. Our results have quantitatively elucidated decay mechanisms that are important towards understanding and controlling of the electronic states in Si2Te3 nanostructures for optoelectronic applications.

physics.optics↗

arXiv2016

Unified Band Theoretic Description of Electronic and Magnetic Properties of Vanadium Dioxide Phases

The debate about whether the insulating phases of vanadium dioxide (VO2) can be described by band theory or must be described by a theory of strong electron correlations remains unresolved even after decades of research. Energy-band calculations using hybrid exchange functionals or including self-energy corrections account for the insulating or metallic nature of different phases, but have not yet successfully accounted for the observed magnetic orderings. Strongly-correlated theories have had limited quantitative success. Here we report that, by using hard pseudopotentials and an optimized hybrid exchange functional, the energy gaps and magnetic orderings of both monoclinic VO2 phases and the metallic nature of the high-temperature rutile phase are consistent with available experimental data, obviating an explicit role for strong correlations. We also report a potential candidate for the newly-found metallic monoclinic phase and present a detailed magnetic structure of the M2 monoclinic phase.

cond-mat.mtrl-sci↗

arXiv2015

High On/Off Ratio Memristive Switching of Manganite-Cuprate Bilayer by Interfacial Magnetoelectricity

Memristive switching serves as the basis for a new generation of electronic devices. Memristors are two-terminal devices in which the current is turned on and off by redistributing point defects, e.g., vacancies, which is difficult to control. Memristors based on alternative mechanisms have been explored, but achieving both the high On/Off ratio and the low switching energy desirable for use in electronics remains a challenge. Here we report memristive switching in a La0.7Ca0.3MnO3/PrBa2Cu3O7 bilayer with an On/Off ratio greater than 10^3 and demonstrate that the phenomenon originates from a new type of interfacial magnetoelectricity. Using results from first-principles calculations, we show that an external electric-field induces subtle displacements of the interfacial Mn ions, which switches on/off an interfacial magnetic "dead" layer, resulting in memristive behavior for spin-polarized electron transport across the bilayer. The interfacial nature of the switching entails low energy cost about of a tenth of atto Joule for write/erase a "bit". Our results indicate new opportunities for manganite/cuprate systems and other transition-metal-oxide junctions in memristive applications.

cond-mat.str-el↗