arXiv subjects

Zhifeng Li

Publications and source records attributed to Zhifeng Li.

81 records · Page 5Linked to original sources

arXiv2017

Granular materials flow like complex fluids

Granular materials such as sand, powders, foams etc. are ubiquitous in our daily life, as well as in industrial and geotechnical applications. Although these disordered systems form stable structures if unperturbed, in practice they do relax because of the presence of unavoidable external influences such as tapping or shear. Often it is tacitly assumed that for granular systems this relaxation dynamics is similar to the one of thermal glass-formers, but in fact experimental difficulties have so far prevented to determine the dynamic properties of three dimensional granular systems on the particle level. This lack of experimental data, combined with the fact that in these systems the motion of the particles involves friction, makes it very challenging to come up with an accurate description of their relaxation dynamics. Here we use X-ray tomography to determine the microscopic relaxation dynamics of hard granular ellipsoids that are subject to an oscillatory shear. We find that the distribution function of the particle displacement can be described by a Gumbel law with a shape parameter that is independent of time and the strain amplitude $γ$. Despite this universality, the mean squared displacement of a tagged particle shows power-laws as a function of time with an exponent that depends on $γ$ and the time interval considered. We argue that these results are directly related to the existence of the microscopic relaxation mechanisms that involve friction and memory effects. These observations demonstrate that on the particle level the dynamical behavior of granular systems is qualitatively different from the one of thermal glass-formers and instead more similar to the one of complex fluids. Thus we conclude that granular materials can relax even when the driving is weak, an insight which impacts our understanding of the nature of granular solids.

cond-mat.soft↗

arXiv2017

Detecting Faces Using Region-based Fully Convolutional Networks

Face detection has achieved great success using the region-based methods. In this report, we propose a region-based face detector applying deep networks in a fully convolutional fashion, named Face R-FCN. Based on Region-based Fully Convolutional Networks (R-FCN), our face detector is more accurate and computational efficient compared with the previous R-CNN based face detectors. In our approach, we adopt the fully convolutional Residual Network (ResNet) as the backbone network. Particularly, We exploit several new techniques including position-sensitive average pooling, multi-scale training and testing and on-line hard example mining strategy to improve the detection accuracy. Over two most popular and challenging face detection benchmarks, FDDB and WIDER FACE, Face R-FCN achieves superior performance over state-of-the-arts.

cs.CV↗

arXiv2017

Face R-CNN

Faster R-CNN is one of the most representative and successful methods for object detection, and has been becoming increasingly popular in various objection detection applications. In this report, we propose a robust deep face detection approach based on Faster R-CNN. In our approach, we exploit several new techniques including new multi-task loss function design, online hard example mining, and multi-scale training strategy to improve Faster R-CNN in multiple aspects. The proposed approach is well suited for face detection, so we call it Face R-CNN. Extensive experiments are conducted on two most popular and challenging face detection benchmarks, FDDB and WIDER FACE, to demonstrate the superiority of the proposed approach over state-of-the-arts.

cs.CV↗

arXiv2017

Origin of Non-cubic Scaling Law in Disordered Granular Packing

Recent diffraction experiments on metallic glasses have unveiled an unexpected non-cubic scaling law between density and average interatomic distance, which lead to the speculations on the presence of fractal glass order. Using X-ray tomography we identify here a similar non-cubic scaling law in disordered granular packing of spherical particles. We find that the scaling law is directly related to the contact neighbors within first nearest neighbor shell, and therefore is closely connected to the phenomenon of jamming. The seemingly universal scaling exponent around 2.5 arises due to the isostatic condition with contact number around 6, and we argue that the exponent should not be universal.

cond-mat.soft↗

arXiv2016

Range Loss for Deep Face Recognition with Long-tail

Convolutional neural networks have achieved great improvement on face recognition in recent years because of its extraordinary ability in learning discriminative features of people with different identities. To train such a well-designed deep network, tremendous amounts of data is indispensable. Long tail distribution specifically refers to the fact that a small number of generic entities appear frequently while other objects far less existing. Considering the existence of long tail distribution of the real world data, large but uniform distributed data are usually hard to retrieve. Empirical experiences and analysis show that classes with more samples will pose greater impact on the feature learning process and inversely cripple the whole models feature extracting ability on tail part data. Contrary to most of the existing works that alleviate this problem by simply cutting the tailed data for uniform distributions across the classes, this paper proposes a new loss function called range loss to effectively utilize the whole long tailed data in training process. More specifically, range loss is designed to reduce overall intra-personal variations while enlarging inter-personal differences within one mini-batch simultaneously when facing even extremely unbalanced data. The optimization objective of range loss is the $k$ greatest range's harmonic mean values in one class and the shortest inter-class distance within one batch. Extensive experiments on two famous and challenging face recognition benchmarks (Labeled Faces in the Wild (LFW) and YouTube Faces (YTF) not only demonstrate the effectiveness of the proposed approach in overcoming the long tail effect but also show the good generalization ability of the proposed approach.

cs.CV↗

arXiv2016

Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks

Face detection and alignment in unconstrained environment are challenging due to various poses, illuminations and occlusions. Recent studies show that deep learning approaches can achieve impressive performance on these two tasks. In this paper, we propose a deep cascaded multi-task framework which exploits the inherent correlation between them to boost up their performance. In particular, our framework adopts a cascaded structure with three stages of carefully designed deep convolutional networks that predict face and landmark location in a coarse-to-fine manner. In addition, in the learning process, we propose a new online hard sample mining strategy that can improve the performance automatically without manual sample selection. Our method achieves superior accuracy over the state-of-the-art techniques on the challenging FDDB and WIDER FACE benchmark for face detection, and AFLW benchmark for face alignment, while keeps real time performance.

cs.CV↗

arXiv2015

Efficient Face Alignment via Locality-constrained Representation for Robust Recognition

Practical face recognition has been studied in the past decades, but still remains an open challenge. Current prevailing approaches have already achieved substantial breakthroughs in recognition accuracy. However, their performance usually drops dramatically if face samples are severely misaligned. To address this problem, we propose a highly efficient misalignment-robust locality-constrained representation (MRLR) algorithm for practical real-time face recognition. Specifically, the locality constraint that activates the most correlated atoms and suppresses the uncorrelated ones, is applied to construct the dictionary for face alignment. Then we simultaneously align the warped face and update the locality-constrained dictionary, eventually obtaining the final alignment. Moreover, we make use of the block structure to accelerate the derived analytical solution. Experimental results on public data sets show that MRLR significantly outperforms several state-of-the-art approaches in terms of efficiency and scalability with even better performance.

cs.CV↗

arXiv2011

Nonmagnetic invisibility cloak with minimized scattering cross section

We propose one kind of transformation functions for nonmagnetic invisibility cloak with minimized scattering on the basis of generalized transformation. By matching the impedance at the outer surface of the cloak, the transformations with two parameters are determined. To confirm the performance of the cloak, full wave simulation based on the finite element method is carried out. Furthermore, total scattering cross section is computed to better illustrate the scattering characteristics of cloak with different parameters. In addition, based on the effective media theory, alternating layered system composed of two isotropic materials is employed to realize the cloak practically.

physics.optics↗

arXiv1997

Poisson Structures for Dispersionless Integrable Systems and Associated W-Algebras

In analogy to the KP theory, the second Poisson structure for the dispersionless KP hierarchy can be defined on the space of commutative pseudodifferential operators $L=p^n+\sum_{j=-\infty}^{n-1}u_j p^j$. The reduction of the Poisson structure to the symplectic submanifold $u_{n -1}=0$ gives rise to the w-algebras. In this paper, we discuss properties of this Poisson structure, its Miura transformation and reductions. We are particularly interested in the following two cases: a) L is pure polynomial in p with multiple roots and b) L has multiple poles at finite distance. The w-algebra corresponding to the case a) is defined as $w_ {[m_1,m_2, ... ,m_r]}$, where m_i means the multiplicity of roots and to the case b) is defined by $w(n,[m_1,m_2, ... ,m_r])$ where m_i is the multiplicity of poles. We prove that w(n,[m_1, m_2, ... , m_r])$-algebra is isomorphic via a transformation to $w_{[m_1,m_2, ... ,m_r]} \bigoplus w_{n+m} \bigoplus U(1) with $m=\sum m_i$. We also give the explicit free fields representations for these w-algebras.

hep-th↗