arXiv subjects

Hua Zhou

Publications and source records attributed to Hua Zhou.

At least 127 records · Page 7Linked to original sources

arXiv2019

2-bit Model Compression of Deep Convolutional Neural Network on ASIC Engine for Image Retrieval

Image retrieval utilizes image descriptors to retrieve the most similar images to a given query image. Convolutional neural network (CNN) is becoming the dominant approach to extract image descriptors for image retrieval. For low-power hardware implementation of image retrieval, the drawback of CNN-based feature descriptor is that it requires hundreds of megabytes of storage. To address this problem, this paper applies deep model quantization and compression to CNN in ASIC chip for image retrieval. It is demonstrated that the CNN-based features descriptor can be extracted using as few as 2-bit weights quantization to deliver a similar performance as floating-point model for image retrieval. In addition, to implement CNN in ASIC, especially for large scale images, the limited buffer size of chips should be considered. To retrieve large scale images, we propose an improved pooling strategy, region nested invariance pooling (RNIP), which uses cropped sub-images for CNN. Testing results on chip show that integrating RNIP with the proposed 2-bit CNN model compression approach is capable of retrieving large scale images.

cs.AI↗

arXiv2019

Matrix Linear Discriminant Analysis

We propose a novel linear discriminant analysis approach for the classification of high-dimensional matrix-valued data that commonly arises from imaging studies. Motivated by the equivalence of the conventional linear discriminant analysis and the ordinary least squares, we consider an efficient nuclear norm penalized regression that encourages a low-rank structure. Theoretical properties including a non-asymptotic risk bound and a rank consistency result are established. Simulation studies and an application to electroencephalography data show the superior performance of the proposed method over the existing approaches.

stat.ME↗

arXiv2019

OPENMENDEL: A Cooperative Programming Project for Statistical Genetics

Statistical methods for genomewide association studies (GWAS) continue to improve. However, the increasing volume and variety of genetic and genomic data make computational speed and ease of data manipulation mandatory in future software. In our view, a collaborative effort of statistical geneticists is required to develop open source software targeted to genetic epidemiology. Our attempt to meet this need is called the OPENMENDELproject (https://openmendel.github.io). It aims to (1) enable interactive and reproducible analyses with informative intermediate results, (2) scale to big data analytics, (3) embrace parallel and distributed computing, (4) adapt to rapid hardware evolution, (5) allow cloud computing, (6) allow integration of varied genetic data types, and (7) foster easy communication between clinicians, geneticists, statisticians, and computer scientists. This article reviews and makes recommendations to the genetic epidemiology community in the context of the OPENMENDEL project.

stat.AP↗

arXiv2018

Anisotropic structural dynamics of monolayer crystals revealed by femtosecond surface x-ray scattering

X-ray scattering is one of the primary tools to determine crystallographic configuration with atomic accuracy. However, the measurement of ultrafast structural dynamics in monolayer crystals remains a long-standing challenge due to a significant reduction of diffraction volume and complexity of data analysis, prohibiting the application of ultrafast x-ray scattering to study nonequilibrium structural properties at the two-dimensional limit. Here, we demonstrate femtosecond surface x-ray diffraction in combination with crystallographic model-refinement calculations to quantify the ultrafast structural dynamics of monolayer WSe$_2$ crystals supported on a substrate. We found the absorbed optical photon energy is preferably coupled to the in-plane lattice vibrations within 2 picoseconds while the out-of-plane lattice vibration amplitude remains unchanged during the first 10 picoseconds. The model-assisted fitting suggests an asymmetric intralayer spacing change upon excitation. The observed nonequilibrium anisotropic structural dynamics in two-dimensional materials agrees with first-principles nonadiabatic modeling in both real and momentum space, marking the distinct structural dynamics of monolayer crystals from their bulk counterparts. The demonstrated methods unlock the benefit of surface sensitive x-ray scattering to quantitatively measure ultrafast structural dynamics in atomically thin materials and across interfaces.

cond-mat.mtrl-sci↗

arXiv2018

Electric-field Control of Magnetism with Emergent Topological Hall Effect in SrRuO3 through Proton Evolution

Ionic substitution forms an essential pathway to manipulate the carrier density and crystalline symmetry of materials via ion-lattice-electron coupling, leading to a rich spectrum of electronic states in strongly correlated systems. Using the ferromagnetic metal SrRuO3 as a model system, we demonstrate an efficient and reversible control of both carrier density and crystalline symmetry through the ionic liquid gating induced protonation. The insertion of protons electron-dopes SrRuO3, leading to an exotic ferromagnetic to paramagnetic phase transition along with the increase of proton concentration. Intriguingly, we observe an emergent topological Hall effect at the boundary of the phase transition as the consequence of the newly-established Dzyaloshinskii-Moriya interaction owing to the breaking of inversion symmetry in protonated SrRuO3 with the proton compositional film-depth gradient. We envision that electric-field controlled protonation opens a novel strategy to design material functionalities.

cond-mat.str-el↗

arXiv2018

Thickness-dependent self-polarization-induced intrinsic magnetoelectric effects in La0.67Sr0.33MnO3/PbZr0.52Ti0.48O3 heterostructures

The self-polarization of PbZr0.52Ti0.48O3 thin film is switched by changing film thickness through the competition between the strain relaxation-induced flexoelectric fields and the interfacial effects. Without an applied electric field, this reversal of self-polarization is exploited to control the magnetic properties of La0.67Sr0.33MnO3 by the competition/cooperation between the charge-mediated and the strain-mediated effects. Scanning transmission electron microscopy, polarized near edge x-ray absorption spectroscopy, and half-integer diffraction measurements are employed to decode this intrinsic magnetoelectric effects in La0.67Sr0.33MnO3/PbZr0.52Ti0.48O3 heterostructures. With PbZr0.52Ti0.48O3 films < 48 nm, the self-polarization-driven carrier density modulation around La0.67Sr0.33MnO3/PbZr0.52Ti0.48O3 interface and the strain-mediated Mn 3d orbital occupancy work together to enhance magnetism of 14 unit cells La0.67Sr0.33MnO3 film; with PbZr0.52Ti0.48O3 layers > 48 nm, the strain-induced the change of bond length/angle of MnO6 accompanied with a modified spin configuration are responsible for the decrease in Curie temperature and magnetization of 14 unit cells La0.67Sr0.33MnO3 film.

cond-mat.mtrl-sci↗

arXiv2018

Proximal Distance Algorithms: Theory and Examples

Proximal distance algorithms combine the classical penalty method of constrained minimization with distance majorization. If $f(\boldsymbol{x})$ is the loss function, and $C$ is the constraint set in a constrained minimization problem, then the proximal distance principle mandates minimizing the penalized loss $f(\boldsymbol{x})+\fracρ{2}\mathop{dist}(x,C)^2$ and following the solution $\boldsymbol{x}_ρ$ to its limit as $ρ$ tends to $\infty$. At each iteration the squared Euclidean distance $\mathop{dist}(\boldsymbol{x},C)^2$ is majorized by the spherical quadratic $\| \boldsymbol{x}-P_C(\boldsymbol{x}_k)\|^2$, where $P_C(\boldsymbol{x}_k)$ denotes the projection of the current iterate $\boldsymbol{x}_k$ onto $C$. The minimum of the surrogate function $f(\boldsymbol{x})+\fracρ{2}\|\boldsymbol{x}-P_C(\boldsymbol{x}_k)\|^2$ is given by the proximal map $\mathop{prox}_{ρ^{-1}f}[P_C(\boldsymbol{x}_k)]$. The next iterate $\boldsymbol{x}_{k+1}$ automatically decreases the original penalized loss for fixed $ρ$. Since many explicit projections and proximal maps are known, it is straightforward to derive and implement novel optimization algorithms in this setting. These algorithms can take hundreds if not thousands of iterations to converge, but the stereotyped nature of each iteration makes proximal distance algorithms competitive with traditional algorithms. For convex problems, we prove global convergence. Our numerical examples include a) linear programming, b) nonnegative quadratic programming, c) projection to the closest kinship matrix, d) projection onto a second-order cone constraint, e) calculation of Horn's copositive matrix index, f) linear complementarity programming, and g) sparse principal components analysis. The proximal distance algorithm in each case is competitive or superior in speed to traditional methods.

math.OC↗

arXiv2018

A Novel Co-design Peta-scale Heterogeneous Cluster for Deep Learning Training

Large scale deep Convolution Neural Networks (CNNs) increasingly demands the computing power. It is key for researchers to own a great powerful computing platform to leverage deep learning (DL) advancing.On the other hand, as the commonly-used accelerator, the commodity GPUs cards of new generations are more and more expensive. Consequently, it is of importance to design an affordable distributed heterogeneous system that provides powerful computational capacity and develop a well-suited software that efficiently utilizes its computational capacity. In this paper, we present our co-design distributed system including a peta-scale GPU cluster, called "Manoa". Based on properties and topology of Manoa, we first propose job server framework and implement it, named "MiMatrix". The central node of MiMatrix, referred to as the job server, undertakes all of controlling, scheduling and monitoring, and I/O tasks without weight data transfer for AllReduce processing in each iteration. Therefore, MiMatrix intrinsically solves the bandwidth bottleneck of central node in parameter server framework that is widely used in distributed DL tasks. Meanwhile, we also propose a new AllReduce algorithm, GPUDirect RDMA-Aware AllReduce~(GDRAA), in which both computation and handshake message are O(1) and the number of synchronization is two in each iteration that is a theoretical minimum number. Owe to the dedicated co-design distributed system, MiMatrix efficiently makes use of the Manoa's computational capacity and bandwidth. We benchmark Manoa Resnet50 and Resenet101 on Imagenet-1K dataset. Some of results have demonstrated state-of-the-art.

cs.CV↗

arXiv2018

Persistence of Island Arrangements During Layer-by-Layer Growth Revealed Using Coherent X-rays

Understanding surface dynamics during epitaxial film growth is key to growing high quality materials with controllable properties. X-ray photon correlation spectroscopy (XPCS) using coherent x-rays opens new opportunities for in situ observation of atomic-scale fluctuation dynamics during crystal growth. Here, we present the first XPCS measurements of 2D island dynamics during homoepitaxial growth in the layer-by-layer mode. Analysis of the results using two-time correlations reveals a new phenomenon - a memory effect in island nucleation sites on successive crystal layers. Simulations indicate that this persistence in the island arrangements arises from communication between islands on different layers via adatoms. With the worldwide advent of new coherent x-ray sources, the XPCS methods pioneered here will be widely applicable to atomic-scale processes on surfaces.

cond-mat.mtrl-sci↗

arXiv2018

Deep Neural Network Capacity

In recent years, deep neural network exhibits its powerful superiority on information discrimination in many computer vision applications. However, the capacity of deep neural network architecture is still a mystery to the researchers. Intuitively, larger capacity of neural network can always deposit more information to improve the discrimination ability of the model. But, the learnable parameter scale is not feasible to estimate the capacity of deep neural network. Due to the overfitting, directly increasing hidden nodes number and hidden layer number are already demonstrated not necessary to effectively increase the network discrimination ability. In this paper, we propose a novel measurement, named "total valid bits", to evaluate the capacity of deep neural networks for exploring how to quantitatively understand the deep learning and the insights behind its super performance. Specifically, our scheme to retrieve the total valid bits incorporates the skilled techniques in both training phase and inference phase. In the network training, we design decimal weight regularization and 8-bit forward quantization to obtain the integer-oriented network representations. Moreover, we develop adaptive-bitwidth and non-uniform quantization strategy in the inference phase to find the neural network capacity, total valid bits. By allowing zero bitwidth, our adaptive-bitwidth quantization can execute the model reduction and valid bits finding simultaneously. In our extensive experiments, we first demonstrate that our total valid bits is a good indicator of neural network capacity. We also analyze the impact on network capacity from the network architecture and advanced training skills, such as dropout and batch normalization.

cs.CV↗

arXiv2018

Characterization of the X-ray Coherence Properties of an Undulator Beamline at the Advanced Photon Source

In anticipation of the increased use of coherent x-ray methods and the need to upgrade beamlines to match improved source quality, we have characterized the coherence properties of the x-rays delivered by beamline 12ID-D at the Advanced Photon Source. We compare the measured x-ray divergence, beam size, brightness, and coherent flux at energies up to 26 keV to the calculated values from the undulator source, and evaluate the effects of beamline optics such as a mirror, monochromator, and compound refractive lenses. Diffraction patterns from slits as a function of slit width are analyzed using wave propagation theory to obtain the beam divergence and thus coherence length. Imaging of the source using a compound refractive lens was found to be the most accurate method for determining the vertical divergence. While the brightness and coherent flux obtained without a monochromator ("pink beam") agree well with those calculated for the source, those measured with the monochromator were a factor of 3 to 6 lower than the source, primarily because of vertical divergence introduced by the monochromator. The methods we describe should be widely applicable for measuring the x-ray coherence properties of synchrotron beamlines.

physics.acc-ph↗

arXiv2017

MM Algorithms for Variance Component Estimation and Selection in Logistic Linear Mixed Model

Logistic linear mixed model is widely used in experimental designs and genetic analysis with binary traits. Motivated by modern applications, we consider the case with many groups of random effects and each group corresponds to a variance component. When the number of variance components is large, fitting the logistic linear mixed model is challenging. We develop two efficient and stable minorization-maximization (MM) algorithms for the estimation of variance components based on the Laplace approximation of the logistic model. One of them leads to a simple iterative soft-thresholding algorithm for variance component selection using maximum penalized approximated likelihood. We demonstrate the variance component estimation and selection performance of our algorithms by simulation studies and a real data analysis.

stat.CO↗

arXiv2017

ChineseFoodNet: A large-scale Image Dataset for Chinese Food Recognition

In this paper, we introduce a new and challenging large-scale food image dataset called "ChineseFoodNet", which aims to automatically recognizing pictured Chinese dishes. Most of the existing food image datasets collected food images either from recipe pictures or selfie. In our dataset, images of each food category of our dataset consists of not only web recipe and menu pictures but photos taken from real dishes, recipe and menu as well. ChineseFoodNet contains over 180,000 food photos of 208 categories, with each category covering a large variations in presentations of same Chinese food. We present our efforts to build this large-scale image dataset, including food category selection, data collection, and data clean and label, in particular how to use machine learning methods to reduce manual labeling work that is an expensive process. We share a detailed benchmark of several state-of-the-art deep convolutional neural networks (CNNs) on ChineseFoodNet. We further propose a novel two-step data fusion approach referred as "TastyNet", which combines prediction results from different CNNs with voting method. Our proposed approach achieves top-1 accuracies of 81.43% on the validation set and 81.55% on the test set, respectively. The latest dataset is public available for research and can be achieved at https://sites.google.com/view/chinesefoodnet.

cs.CV↗

arXiv2017

An instrument for advanced in situ x-ray studies of metal-organic vapor phase epitaxy of III-nitrides

We describe an instrument that exploits the ongoing revolution in synchrotron sources, optics, and detectors to enable in situ studies of metal-organic vapor phase epitaxy (MOVPE) growth of III-nitride materials using coherent x-ray methods. The system includes high-resolution positioning of the sample and detector including full rotations, an x-ray transparent chamber wall for incident and diffracted beam access over a wide angular range, and minimal thermal sample motion, giving the sub-micron positional stability and reproducibility needed for coherent x-ray studies. The instrument enables surface x-ray photon correlation spectroscopy, microbeam diffraction, and coherent diffraction imaging of atomic-scale surface and film structure and dynamics during growth, to provide fundamental understanding of MOVPE processes.

cond-mat.mtrl-sci↗

arXiv2017

Tuning Interfacial Ferromagnetism in LaNiO$_3$/CaMnO$_3$ Superlattices by Stabilizing Non-Equilibrium Crystal Symmetry

Perovskite oxide heterostructures offer an important path forward for stabilizing and controlling low-dimensional magnetism. One of the guiding design principles for these materials systems is octahedral connectivity. In superlattices composed of perovskites with different crystal symmetries, variation of the relative ratio of the constituent layers as well as the individual layer thicknesses gives rise to non-equilibrium crystal symmetries that, in turn, lead to unprecedented control of interfacial ferromagnetism. We have found that in superlattices of CaMnO$_3$ (CMO) and LaNiO$_3$ (LNO), interfacial ferromagnetism can be modulated by a factor of three depending on LNO and CMO layer thicknesses as well as their relative ratio. Such an effect is only possible due to the non-equilibrium crystal symmetries at the interfaces and can be understood in terms of the anisotropy of the exchange interactions and modifications in the interfacial Ni-O-Mn and Mn-O-Mn bond angles and lengths with increasing LNO layer thickness. These results demonstrate the potential of engineering non-equilibrium crystal symmetries in designing ferromagnetism.

cond-mat.mtrl-sci↗

arXiv2017

Perovskite Quantum Organismoids

A central characteristic of living beings is the ability to learn from and respond to their environment leading to habit formation and decision making1-3. This behavior, known as habituation, is universal among forms of life with a central nervous system, and interestingly observed even in single cellular organisms that do not possess a brain4-5. Here, we report the discovery of habituation based plasticity utilizing a perovskite quantum system by dynamical modulation of electron localization via reversible dopant incorporation. Microscopic mechanisms and pathways that enable this organismic collective charge-lattice interaction are elucidated by a combination of first-principles theory, synchrotron investigations, ab-initio dynamical simulations and in-situ environmental breathing studies. We implement a new learning algorithm inspired from the conductance relaxation behavior of perovskites that naturally incorporates habituation and demonstrate "learning to forget": a key feature of animal and human brains6. Most surprisingly, our results show that incorporating this elementary skill in learning dramatically boosts the capability of artificial cognitive systems.

cond-mat.mtrl-sci↗

arXiv2017

Interface control of ferroelectricity in a SrRuO3/BaTiO3/SrRuO3 capacitor and its critical thickness

The atomic-scale synthesis of artificial oxide heterostructures offers new opportunities to create novel states that do not occur in nature. The main challenge related to synthesizing these structures is obtaining atomically sharp interfaces with designed termination sequences. Here, we demonstrate that the oxygen pressure (PO2) during growth plays an important role in controlling the interfacial terminations of SrRuO3/BaTiO3/SrRuO3 (SRO/BTO/SRO) ferroelectric capacitors. The SRO/BTO/SRO heterostructures were grown by the pulsed laser deposition (PLD) method. The top SRO/BTO interface grown at high PO2 (around 150 mTorr) usually exhibited a mixture of RuO2-BaO and SrO-TiO2 terminations. By reducing PO2, we obtained atomically sharp SRO/BTO top interfaces with uniform SrO-TiO2 termination. Using capacitor devices with symmetric and uniform interfacial termination, we were able to demonstrate for the first time that the ferroelectric (FE) critical thickness can reach the theoretical limit of 3.5 unit cells (u.c.).

cond-mat.mtrl-sci↗

arXiv2016

Algorithms for Fitting the Constrained Lasso

We compare alternative computing strategies for solving the constrained lasso problem. As its name suggests, the constrained lasso extends the widely-used lasso to handle linear constraints, which allow the user to incorporate prior information into the model. In addition to quadratic programming, we employ the alternating direction method of multipliers (ADMM) and also derive an efficient solution path algorithm. Through both simulations and real data examples, we compare the different algorithms and provide practical recommendations in terms of efficiency and accuracy for various sizes of data. We also show that, for an arbitrary penalty matrix, the generalized lasso can be transformed to a constrained lasso, while the converse is not true. Thus, our methods can also be used for estimating a generalized lasso, which has wide-ranging applications. Code for implementing the algorithms is freely available in the Matlab toolbox SparseReg.

stat.ML↗