arXiv subjects

Lei Xu

Publications and source records attributed to Lei Xu.

At least 235 records · Page 13Linked to original sources

arXiv2019

Revisit Lmser and its further development based on convolutional layers

Proposed in 1991, Least Mean Square Error Reconstruction for self-organizing network, shortly Lmser, was a further development of the traditional auto-encoder (AE) by folding the architecture with respect to the central coding layer and thus leading to the features of symmetric weights and neurons, as well as jointly supervised and unsupervised learning. However, its advantages were only demonstrated in a one-hidden-layer implementation due to the lack of computing resources and big data at that time. In this paper, we revisit Lmser from the perspective of deep learning, develop Lmser network based on multiple convolutional layers, which is more suitable for image-related tasks, and confirm several Lmser functions with preliminary demonstrations on image recognition, reconstruction, association recall, and so on. Experiments demonstrate that Lmser indeed works as indicated in the original paper, and it has promising performance in various applications.

cs.LG↗

arXiv2019

Spontaneous photon-pair generation at the nanoscale

Optical nanoantennas have shown a great capacity for efficient extraction of photons from the near to the far-field, enabling directional emission from nanoscale single-photon sources. However, their potential for the generation and extraction of multi-photon quantum states remains unexplored. Here we demonstrate experimentally the nanoscale generation of two-photon quantum states at telecommunication wavelengths based on spontaneous parametric down-conversion in an optical nanoantenna. The antenna is a crystalline AlGaAs nanocylinder, possessing Mie-type resonances at both the pump and the bi-photon wavelengths and when excited by a pump beam generates photonpairs with a rate of 35 Hz. Normalized to the pump energy stored by the nanoantenna, this rate corresponds to 1.4 GHz/Wm, being one order of magnitude higher than conventional on-chip or bulk photon-pair sources. Our experiments open the way for multiplexing several antennas for coherent generation of multi-photon quantum states with complex spatial-mode entanglement and applications in free-space quantum communications and sensing.

quant-ph↗

arXiv2019

Superexchange interactions between spin-orbit-coupled $j\!\approx\!1/2$ ions in oxides with face-sharing ligand octahedra

Using ab initio wave-function-based calculations, we provide valuable insights with regard to the magnetic exchange in 5$d$ and 4$d$ oxides with face-sharing ligand octahedra, BaIrO$_3$ and BaRhO$_3$. Surprisingly strong antiferromagnetic Heisenberg interactions as large as 400 meV are computed for idealized iridate structures with 90$^{\circ}$ Ir-O-Ir bond angles and in the range of 125 meV for angles of 80$^{\circ}$ as measured experimentally in BaIrO$_3$. These estimates exceed the values derived so far for corner-sharing and edge-sharing systems and motivate more detailed experimental investigations of quantum magnets with extended 5$d$/4$d$ orbitals and networks of face-sharing ligand cages. The strong electron-lattice couplings evidenced by our calculations suggest rich phase diagrams as function of strain and pressure, a research direction with much potential for materials of this type.

cond-mat.str-el↗

arXiv2019

A three-dimensional constitutive model for the martensitic transformation in polycrystalline shape memory alloys under large deformation

This work presents a three-dimensional constitutive model for the martensitic transformation in polycrystalline Shape Memory Alloys (SMAs) under large deformation. By utilizing the logarithmic strain and rate, the model is able to account for large strains and rotations that SMA-based components may undertake, but also resolves the artificial stress errors caused by the non-integrable objective rates that are widely used in current commercial finite element software. The model is developed through classical thermodynamic laws combined with the standard Coleman-Noll procedure. The scalar martensitic volume fraction and the second-order transformation strain tensor are chosen as the internal state variables to capture the material response exhibited by polycrystalline SMAs. A detailed implementation procedure of the proposed model is described through a user-defined material subroutine. Numerical experiments considering SMA components including a bar, a beam, a torque tube and a solid flexible structure under stress/thermally-induced phase transformations are investigated via the proposed model, and the results under cyclic loading are compared against the predictions provided by the Abaqus nonlinear solver. The development framework of the proposed model and its implementation procedure can be extended to incorporate other nonlinear phenomena exhibited by SMAs, such as transformation-induced plasticity, viscoplasticity, and damage under large deformation.

cond-mat.mtrl-sci↗

arXiv2019

A* Tree Search for Portfolio Management

We propose a planning-based method to teach an agent to manage portfolio from scratch. Our approach combines deep reinforcement learning techniques with search techniques like AlphaGo. By uniting the advantages in A* search algorithm with Monte Carlo tree search, we come up with a new algorithm named A* tree search in which best information is returned to guide next search. Also, the expansion mode of Monte Carlo tree is improved for a higher utilization of the neural network. The suggested algorithm can also optimize non-differentiable utility function by combinatorial search. This technique is then used in our trading system. The major component is a neural network that is trained by trading experiences from tree search and outputs prior probability to guide search by pruning away branches in turn. Experimental results on simulated and real financial data verify the robustness of the proposed trading system and the trading system produces better strategies than several approaches based on reinforcement learning.

cs.AI↗

arXiv2019

Controllable SH wave radiation patterns generated by finite-size face-shear piezoelectric transducers for structural health monitoring

The fundamental shear horizontal (SH0) wave in plate-like structures is of practical importance in structural health monitoring due to its non-dispersive characteristics. However, compared with Lamb wave, SH0 wave is less used in practice since it is difficult to be excited by using conventional piezoelectric transducers. Recently, face-shear piezoelectric mode has been realized in PZT ceramics and several face-shear piezoelectric transducers have been developed to excite pure SH0 wave in plates. However, when controllable radiation pattern of SH0 wave is required, it is found that no mathematical basis is available for designers to optimize the transducer's parameters. In this work, based on the reciprocal theorem of elastodynamics and Huygens' principle, a theoretical model is proposed to describe the SH0 wave field generated by arbitrary-shaped face-shear transducers bonded on an infinite isotropic plate. It is proved that the wave field generated by a face-shear transducer is equivalent to that generated by uniform in-plane tractions along its perimeter. Then the proposed model is applied to two specific cases of a single square face-shear (d24) PZT wafer and a bidirectional SH0 wave piezoelectric transducer(BSH-PT).Explicit analytical expressions for both cases are successfully obtained. The accuracy of the obtained analytical model is verified by both experiments and FEM simulations. Furthermore, some general guidelines based on the analytical model are presented for optimization of the transducer's dimensions. Finally, based on the guidelines, a new bidirectional SH0 wave piezoelectric transducer is proposed with very small beam divergence.

physics.app-ph↗

arXiv2019

SteganoGAN: High Capacity Image Steganography with GANs

Image steganography is a procedure for hiding messages inside pictures. While other techniques such as cryptography aim to prevent adversaries from reading the secret message, steganography aims to hide the presence of the message itself. In this paper, we propose a novel technique for hiding arbitrary binary data in images using generative adversarial networks which allow us to optimize the perceptual quality of the images produced by our model. We show that our approach achieves state-of-the-art payloads of 4.4 bits per pixel, evades detection by steganalysis tools, and is effective on images from multiple datasets. To enable fair comparisons, we have released an open source library that is available online at https://github.com/DAI-Lab/SteganoGAN.

cs.CV↗

arXiv2018

A Finite Strain Constitutive Model Considering Transformation Induced Plasticity for Shape Memory Alloys under Cyclic Loading

Many engineering applications of Shape Memory Alloys (SMAs) involve passing back and forth through phase transformation many times. Repeated phase transformation develops permanent deformations originating from the significant distortion that phase transformation induces at the austenite-martensite interfaces and grain boundaries. This distortion drives dislocation activity resulting in an observable macroscopic Transformation-Induced Plastic (TRIP) deformation, which occurs at effective stress levels much lower than the plastic yield limit of the material. TRIP strains may accumulate up to 20% during the lifetime of an SMA component and thus a finite strain constitutive model is required for simulating their response. In this work, a 3-D finite strain model is developed based on logarithmic strain which is the only strain measure whose rate is equal to the stretching, that describes phase transformation and TRIP deformation in a thermodynamically consistent setting. The model is implemented in Abaqus finite element software through a user-defined material subroutine (UMAT). Boundary value problems such as strip and a torque tube under both pseudoelastic and actuation cyclic loadings are performed to test the capabilities of the newly proposed model.

cond-mat.mtrl-sci↗

arXiv2018

Generation of multiple plasmons in strontium niobates mediated by local field effects

Recently, an anomalous generation of multiple plasmons with large spectral weight transfer in the visible to ultraviolet range (energies below the band gap) has been experimentally observed in the insulating-like phase of oxygen-rich strontium niobium oxides (SrNbO$_{3+δ}$). Here, we investigate the ground state and dielectric properties of SrNbO$_{3+δ}$ as a function of $δ$ by means of extensive first principle calculations. We find that in the random phase approximation by taking into account the local field effects (LFEs), our calculations are able to reproduce both the unconventional multiple generations of plasmons and spectral weight transfers, consistent with experimental data. Interestingly, these unconventional plasmons can be tuned by oxygen stoichiometry as well as microscopic superstructure. This unusual predominance of LFEs in this class of materials is ascribed to the strong electronic inhomogeneity and high polarizability and paves a new path to induce multiple plasmons in the untapped visible to ultraviolet ranges of insulating-like oxides.

cond-mat.str-el↗

arXiv2018

A Three-Dimensional Constitutive Modeling for Shape Memory Alloys Considering Two-Way Shape Memory Effect and Transformation-Induced Plasticity

Shape memory alloys (SMAs) have been intensively investigated as actuators for the past several decades. Due to their high actuation energy density compared to other active materials, their current and potential applications in engineering fields are expanding rapidly. Prior to being used as actuators, SMAs are usually subjected to a training process to stabilize their behavior. During the training process, permanent changes are introduced in the microstructure of the material which results in the generation of internal stresses and a large amount of irrecoverable Transformation Induced Plastic strain (TRIP). The generated internal stresses along with a potential thermal loading provide the driving force to induce the oriented phase transformation so that the SMA-based actuators are able to exhibit the Two-Way Shape Memory Effect (TWSME) without applying external bias load. To predict this intrinsic phenomenon, a three-dimensional phenomenological constitutive model for untrained SMAs is presented. The proposed model utilizes the martensitic volume fraction, transformation strain, TRIP strain, and internal stress as internal state variables so that it is able to account for the evolution of TRIP strain and the TWSME for untrained SMAs under cyclic thermomechanical loading conditions. In the end, boundary value problems considering an untrained SMA material under isothermal/isobaric cyclic loading are solved and the predicted cyclic response is compared against available experimental data to demonstrate the proposed capabilities.

cond-mat.mtrl-sci↗

arXiv2018

A Finite Strain Constitutive Model for Martensitic Transformation in Shape Memory Alloys Based on Logarithmic Strain

Shape Memory Alloys (SMAs) are materials with the ability to recover apparently permanent deformation under specific thermomechanical loading. The majority of constitutive models for SMAs are developed based on the infinitesimal strain theory. However, such assumption may not be proper in the presence of geometric discontinuities, such as cracks, and repeated cycling loading that has been reported to induce irrecoverable strains up to 20% due to transformation induced plasticity. In addition to finite strains, SMA-based devices may also undergo large rotations. Thus, it is indispensable to develop a constitutive model based on the finite strain to provide accurate predictions of these actuators response. A three-dimensional phenomenological constitutive model for SMAs considering finite strains and finite rotations is proposed in this work. This model utilizes the logarithmic strain as the strain measure that is the strain measure whose logarithmic rate in a corotating material frame is equal to the rate of deformation tensor. In the proposed model, the martensitic volume fraction and the second-order logarithmic transformation strain tensor are chosen as the internal state variables associated with the inelastic transformation process. Numerical simulations considering basic SMAs component geometries such as a bar, a beam, and a torque tube are performed to test the capabilities of the proposed model under both mechanically and thermally induced phase transformation. The presented model formulation will be extended in future work for the incorporation of transformation-induced plasticity.

cond-mat.mtrl-sci↗

arXiv2018

Synthesizing Tabular Data using Generative Adversarial Networks

Generative adversarial networks (GANs) implicitly learn the probability distribution of a dataset and can draw samples from the distribution. This paper presents, Tabular GAN (TGAN), a generative adversarial network which can generate tabular data like medical or educational records. Using the power of deep neural networks, TGAN generates high-quality and fully synthetic tables while simultaneously generating discrete and continuous variables. When we evaluate our model on three datasets, we find that TGAN outperforms conventional statistical generative models in both capturing the correlation between columns and scaling up for large datasets.

cs.LG↗

arXiv2018

Computational Decomposition of Style for Controllable and Enhanced Style Transfer

Neural style transfer has been demonstrated to be powerful in creating artistic image with help of Convolutional Neural Networks (CNN). However, there is still lack of computational analysis of perceptual components of the artistic style. Different from some early attempts which studied the style by some pre-processing or post-processing techniques, we investigate the characteristics of the style systematically based on feature map produced by CNN. First, we computationally decompose the style into basic elements using not only spectrum based methods including Fast Fourier Transform (FFT), Discrete Cosine Transform (DCT) but also latent variable models such Principal Component Analysis (PCA), Independent Component Analysis (ICA). Then, the decomposition of style induces various ways of controlling the style elements which could be embedded as modules in state-of-the-art style transfer algorithms. Such decomposition of style brings several advantages. It enables the computational coding of different artistic styles by our style basis with similar styles clustering together, and thus it facilitates the mixing or intervention of styles based on the style basis from more than one styles so that compound style or new style could be generated to produce styled images. Experiments demonstrate the effectiveness of our method on not only painting style transfer but also sketch style transfer which indicates possible applications on picture-to-sketch problems.

cs.CV↗

arXiv2018

Lightweight (Reverse) Fuzzy Extractor with Multiple Referenced PUF Responses

A Physical unclonable functions (PUF), alike a fingerprint, exploits manufacturing randomness to endow each physical item with a unique identifier. One primary PUF application is the secure derivation of volatile cryptographic keys using a fuzzy extractor comprising of two procedures: i) secure sketch; and ii) entropy extraction. Although the entropy extractor can be lightweight, the overhead of the secure sketch responsible correcting naturally noisy PUF responses is usually costly. We observe that, in general, response unreliability with respect to a enrolled reference measurement increases with increasing differences between the in-the-field PUF operating condition and the operating condition used in evaluating the enrolled reference response. For the first time, we exploit such an important but inadvertent observation. In contrast to the conventional single reference response enrollment, we propose enrolling multiple reference responses (MRR) subject to the same challenge but under multiple distinct operating conditions. The critical observation here is that one of the reference operating conditions is likely to be closer to the operating condition of the field deployed PUF, thus, resulting in minimizing the expected unreliability when compared to the single reference under the nominal condition. Overall, MRR greatly reduces the demand for the expected number of erroneous bits for correction and, subsequently, achieve a significant reduction in the error correction overhead. The significant implementation efficiency gains from the proposed MRR method is demonstrated from software implementations of fuzzy extractors on batteryless resource constraint computational radio frequency identification devices, where realistic PUF data is collected from the embedded intrinsic SRAM PUFs.

cs.CR↗

arXiv2018

Efficient Public Blockchain Client for Lightweight Users

Public blockchains provide a decentralized method for storing transaction data and have many applications in different sectors. In order for users to track transactions, a simple method is to let them keep a local copy of the entire public ledger. Since the size of the ledger keeps growing, this method becomes increasingly less practical, especially for lightweight users such as IoT devices and smartphones. In order to cope with the problem, several solutions have been proposed to reduce the storage burden. However, existing solutions either achieve a limited storage reduction (e.g., simple payment verification), or rely on some strong security assumption (e.g., the use of trusted server). In this paper, we propose a new approach to solving the problem. Specifically, we propose an \underline{e}fficient verification protocol for \underline{p}ublic \underline{b}lock\underline{c}hains, or EPBC for short. EPBC is particularly suitable for lightweight users, who only need to store a small amount of data that is {\it independent of} the size of the blockchain. We analyze EPBC's performance and security, and discuss its integration with existing public ledger systems. Experimental results confirm that EPBC is practical for lightweight users.

cs.CR↗

arXiv2018

Election with Bribed Voter Uncertainty: Hardness and Approximation Algorithm

Bribery in election (or computational social choice in general) is an important problem that has received a considerable amount of attention. In the classic bribery problem, the briber (or attacker) bribes some voters in attempting to make the briber's designated candidate win an election. In this paper, we introduce a novel variant of the bribery problem, "Election with Bribed Voter Uncertainty" or BVU for short, accommodating the uncertainty that the vote of a bribed voter may or may not be counted. This uncertainty occurs either because a bribed voter may not cast its vote in fear of being caught, or because a bribed voter is indeed caught and therefore its vote is discarded. As a first step towards ultimately understanding and addressing this important problem, we show that it does not admit any multiplicative $O(1)$-approximation algorithm modulo standard complexity assumptions. We further show that there is an approximation algorithm that returns a solution with an additive-$ε$ error in FPT time for any fixed $ε$.

cs.DS↗

arXiv2018

Quantum metasurface for multi-photon interference and state reconstruction

Metasurfaces based on resonant nanophotonic structures have enabled novel types of flat-optics devices often outperforming the capabilities of bulk components, yet these advances remain largely unexplored for quantum applications. We show that non-classical multi-photon interferences can be achieved at the subwavelength scale in all-dielectric metasurfaces. We simultaneously image multiple projections of quantum states with a single metasurface, enabling a robust reconstruction of amplitude, phase, coherence, and entanglement of multi-photon polarization-encoded states. One- and two-photon states are reconstructed through nonlocal photon correlation measurements with polarization-insensitive click-detectors positioned after the metasurface, and the scalability to higher photon numbers is established theoretically. Our work illustrates the feasibility of ultra-thin quantum metadevices for the manipulation and measurement of multi-photon quantum states with applications in free-space quantum imaging and communications.

quant-ph↗

arXiv2018

Weighted Linear Discriminant Analysis based on Class Saliency Information

In this paper, we propose a new variant of Linear Discriminant Analysis to overcome underlying drawbacks of traditional LDA and other LDA variants targeting problems involving imbalanced classes. Traditional LDA sets assumptions related to Gaussian class distribution and neglects influence of outlier classes, that might hurt in performance. We exploit intuitions coming from a probabilistic interpretation of visual saliency estimation in order to define saliency of a class in multi-class setting. Such information is then used to redefine the between-class and within-class scatters in a more robust manner. Compared to traditional LDA and other weight-based LDA variants, the proposed method has shown certain improvements on facial image classification problems in publicly available datasets.

cs.CV↗