Limin Wang - University of Pennsylvania

Limin Wang
Are you Limin Wang?

Claim your profile, edit publications, add additional information:

Contact Details

Limin Wang
University of Pennsylvania
United States

Pubs By Year

Pub Categories

Computer Science - Computer Vision and Pattern Recognition (18)
Astrophysics (12)
Physics - Materials Science (6)
Physics - Superconductivity (6)
Physics - Fluid Dynamics (5)
Physics - Strongly Correlated Electrons (5)
High Energy Physics - Phenomenology (4)
Physics - Computational Physics (4)
High Energy Physics - Theory (1)
General Relativity and Quantum Cosmology (1)
Physics - Chemical Physics (1)
Physics - Mesoscopic Systems and Quantum Hall Effect (1)
Physics - Soft Condensed Matter (1)
Quantum Physics (1)

Publications Authored By Limin Wang

Detecting activities in untrimmed videos is an important yet challenging task. In this paper, we tackle the difficulties of effectively locating the start and the end of a long complex action, which are often met by existing methods. Our key contribution is the structured segment network, a novel framework for temporal action detection, which models the temporal structure of each activity instance via a structured temporal pyramid. Read More

Deep ConvNets have been shown to be effective for the task of human pose estimation from single images. However, several challenging issues arise in the video-based case such as self-occlusion, motion blur, and uncommon poses with few or no examples in training data sets. Temporal information can provide additional cues about the location of body joints and help to alleviate these issues. Read More

We report observations of magnetoresistance, quantum oscillations and angle-resolved photoemission in RhSb$_3$, a unfilled skutterudite semimetal with low carrier density. The calculated electronic band structure of RhSb$_3$ entails a $Z_2$ quantum number $\nu_0=0,\nu_1=\nu_2=\nu_3=1$ in analogy to strong topological insulators, and inverted linear valence/conduction bands that touch at discrete points close to the Fermi level, in agreement with angle-resolved photoemission results. Transport experiments reveal an unsaturated linear magnetoresistance that approaches a factor of 200 at 60 T magnetic fields, and quantum oscillations observable up to 150~K that are consistent with a large Fermi velocity ($\sim 1. Read More

Current action recognition methods heavily rely on trimmed videos for model training. However, it is very expensive and time-consuming to acquire a large-scale trimmed video dataset. This paper presents a new weakly supervised architecture, called UntrimmedNet, which is able to directly learn from untrimmed videos without the need of temporal annotations of action instances. Read More

Detecting activities in untrimmed videos is an important but challenging task. The performance of existing methods remains unsatisfactory, e.g. Read More

Convolutional Neural Networks (CNNs) have made remarkable progress on scene recognition, partially due to these recent large-scale scene datasets, such as the Places and Places2. Scene categories are often defined by multi-level information, including local objects, global layout, and background environment, thus leading to large intra-class variations. In addition, with the increasing number of scene categories, label ambiguity has become another crucial issue in large-scale classification. Read More

Event recognition in still images is an intriguing problem and has potential for real applications. This paper addresses the problem of event recognition by proposing a convolutional neural network that exploits knowledge of objects and scenes for event classification (OS2E-CNN). Intuitively, it stands to reason that there exists a correlation among the concepts of objects, scenes, and events. Read More

This paper presents the method that underlies our submission to the untrimmed video classification task of ActivityNet Challenge 2016. We follow the basic pipeline of temporal segment networks and further raise the performance via a number of other techniques. Specifically, we use the latest deep model architecture, e. Read More

Deep convolutional networks have achieved great success for visual recognition in still images. However, for action recognition in videos, the advantage over traditional methods is not so evident. This paper aims to discover the principles to design effective ConvNet architectures for action recognition in videos and learn these models given limited training samples. Read More

The deep two-stream architecture exhibited excellent performance on video based action recognition. The most computationally expensive step in this approach comes from the calculation of optical flow which prevents it to be real-time. This paper accelerates this architecture by replacing optical flow with motion vector which can be obtained directly from compressed videos without extra calculation. Read More

Actionness was introduced to quantify the likelihood of containing a generic action instance at a specific location. Accurate and efficient estimation of actionness is important in video analysis and may benefit other relevant tasks such as action recognition and action detection. This paper presents a new deep architecture for actionness estimation, called hybrid fully convolutional network (H-FCN), which is composed of appearance FCN (A-FCN) and motion FCN (M-FCN). Read More

We report the electronic structure, electric and thermal transport properties of Ru$_{1-x}$Ir$_{x}$Se$_2$ ($x \leq 0.2$). RuSe$_2$ is a semiconductor that crystallizes in a cubic pyrite unit cell. Read More

We report two-dimensional quantum transport and Dirac fermions in BaMnBi2 single crystals. BaMnBi2 is a layered bad metal with highly anisotropic conductivity and magnetic order below 290 K. Magnetotransport properties, nonzero Berry phase, small cyclotronmass, and the first-principles band structure calculations indicate the presence of Dirac fermions in Bi square nets. Read More

We theoretically consider the superconductivity of the topological half-Heusler semimetals YPtBi and LuPtBi. We show that pairing occurs between j=3/2 fermion states, which leads to qualitative differences from the conventional theory of pairing between j=1/2 states. In particular, this permits Cooper pairs with quintet or septet total angular momentum, in addition to the usual singlet and triplet states. Read More

In all known fermionic superfluids, Cooper pairs are composed of spin-1/2 quasi-particles that pair to form either spin-singlet or spin-triplet bound states. The "spin" of a Bloch electron, however, is fixed by the symmetries of the crystal and the atomic orbitals from which it is derived, and in some cases can behave as if it were a spin-3/2 particle. The superconducting state of such a system allows pairing beyond spin-triplet, with higher spin quasi-particles combining to form quintet or even septet pairs. Read More

Convolutional neural networks (CNN) have recently achieved remarkable successes in various image classification and understanding tasks. The deep features obtained at the top fully-connected layer of the CNN (FC-features) exhibit rich global semantic information and are extremely effective in image classification. On the other hand, the convolutional features in the middle layers of the CNN also contain meaningful local information, but are not fully explored for image representation. Read More

Event recognition from still images is one of the most important problems for image understanding. However, compared with object recognition and scene recognition, event recognition has received much less research attention in computer vision community. This paper addresses the problem of cultural event recognition in still images and focuses on applying deep learning methods on this problem. Read More

VGGNets have turned out to be effective for object recognition in still images. However, it is unable to yield good performance by directly adapting the VGGNet models trained on the ImageNet dataset for scene recognition. This report describes our implementation of training the VGGNets on the large-scale Places205 dataset. Read More

Deep convolutional networks have achieved great success for object recognition in still images. However, for action recognition in videos, the improvement of deep convolutional networks is not so evident. We argue that there are two reasons that could probably explain this result. Read More

Visual features are of vital importance for human action understanding in videos. This paper presents a new video representation, called trajectory-pooled deep-convolutional descriptor (TDD), which shares the merits of both hand-crafted features and deep-learned features. Specifically, we utilize deep architectures to learn discriminative convolutional feature maps, and conduct trajectory-constrained pooling to aggregate these convolutional features into effective descriptors. Read More

Event recognition from still images is of great importance for image understanding. However, compared with event recognition in videos, there are much fewer research works on event recognition in images. This paper addresses the issue of event recognition from images and proposes an effective method with deep neural networks. Read More

We report a high-pressure study of simultaneous low-temperature electrical resistivity and Hall effect measurements on high quality single-crystalline KFe2As2 using designer diamond anvil cell techniques with applied pressures up to 33 GPa. In the low pressure regime, we show that the superconducting transition temperature T_c finds a maximum onset value of 7 K near 2 GPa, in contrast to previous reports that find a minimum T_c and reversal of pressure dependence at this pressure. Upon applying higher pressures, this T_c is diminished until a sudden drastic enhancement occurs coincident with a first-order structural phase transition into a collapsed tetragonal phase. Read More

We investigate continuous-time quantum walks of two indistinguishable anyons in one-dimensional lattices with both on-site and nearest-neighbor interactions based on the fractional Jordan-Wigner transformation. It is shown that the two-body correlations in position space are symmetric about the initial sites of two quantum walkers in the Bose limit ($\chi=0$ ) and Fermi limit ( $\chi=1$), while in momentum space this happens only in the Bose limit. An interesting asymmetry arises in the correlation for most cases with the statistical parameter $\chi$ varying in between. Read More

A numerical method, based on the discrete lattice Boltzmann equation, is presented for solving the volume-averaged Navier-Stokes equations. With a modified equilibrium distribution and an additional forcing term, the volume-averaged Navier-Stokes equations can be recovered from the lattice Boltzmann equation in the limit of small Mach number by the Chapman-Enskog analysis and Taylor expansion. Due to its advantages such as explicit solver and inherent parallelism, the method appears to be more competitive with traditional numerical techniques. Read More

Video based action recognition is one of the important and challenging problems in computer vision research. Bag of Visual Words model (BoVW) with local features has become the most popular method and obtained the state-of-the-art performance on several realistic datasets, such as the HMDB51, UCF50, and UCF101. BoVW is a general pipeline to construct a global representation from a set of local features, which is mainly composed of five steps: (i) feature extraction, (ii) feature pre-processing, (iii) codebook generation, (iv) feature encoding, and (v) pooling and normalization. Read More

The closure problem of turbulence is still a challenging issue in turbulence modeling. In this work, a stability condition is used to close turbulence. Specifically, we regard single-phase flow as a mixture of turbulent and non-turbulent fluids, separating the structure of turbulence. Read More

Discrete particle simulation, a combined approach of computational fluid dynamics and discrete methods such as DEM (Discrete Element Method), DSMC (Direct Simulation Monte Carlo), SPH (Smoothed Particle Hydrodynamics), PIC (Particle-In-Cell), etc., is becoming a practical tool for exploring lab-scale gas-solid systems owing to the fast development of parallel computation. However, gas-solid coupling and the corresponding fluid flow solver remain immature. Read More

This paper presents the shape optimization of a flat-type arborescent fluid distributor for the purpose of process intensification. A shape optimization algorithm based on the lattice Boltzmann method (LBM) is proposed with the objective of decreasing the flow resistance of such distributor at the constraint of constant fluid volume. Prototypes of the initial distributor as well as the optimized one are designed. Read More

An unexpected insensitivity of the Fermi surface to impurity scattering is found in Ru substituted BaFe$_2$As$_2$ from first-principles theory, offering a natural explanation of the unusual resilience of transport and superconductivity to a high level of disordered substitution in this material. This robustness is shown to originate from a coherent interference of correlated on-site and inter-site impurity scattering, similar in spirit to the microscopic mechanism of superdiffusion in one dimension. Our result also demonstrates a strong substitution dependence of the Fermi surface and carrier concentration, and provides a natural resolution to current discrepancies in recent photoelectron spectroscopy. Read More

Recent neutron scattering experiments addressing the magnetic state of the two-leg ladder selenide compound BaFe$_2$Se$_3$ have unveiled a dominant spin arrangement involving ferromagnetically ordered 2$\times$2 iron-superblocks, that are antiferromagnetically coupled among them (the "block-AFM" state). Using the electronic five-orbital Hubbard model, first principles techniques to calculate the electronic hopping amplitudes between irons, and the real-space Hartree-Fock approximation to handle the many-body effects, here it is shown that the exotic block-AFM state is indeed stable at realistic electronic densities close to $n \sim 6.0$. Read More

We report temperature and magnetic field dependence of the thermal transport properties in single crystals of (Sr/Ca)MnBi$_2$ with linear energy dispersion. In SrMnBi$_2$ thermopower is positive, indicating hole-type carriers and the magnetic field enhances the thermopower significantly. The maximum change of thermopower is about 1600% in 9 T field and at 10 K. Read More

We report two dimensional Dirac fermions and quantum magnetoresistance in single crystals of CaMnBi$_2$. The non-zero Berry's phase, small cyclotron resonant mass and first-principle band structure suggest the existence of the Dirac fermions in the Bi square nets. The in-plane transverse magnetoresistance exhibits a crossover at a critical field $B^*$ from semiclassical weak-field $B^2$ dependence to the high-field unsaturated linear magnetoresistance ($\sim 120%$ in 9 T at 2 K) due to the quantum limit of the Dirac fermions. Read More

Quasielastic neutron scattering and molecular dynamics simulation data from PEO/PMMA blends found that for short times the self-dynamics of PEO chain follows the Rouse model, but at longer times past tc=1 to 2 ns it becomes slower and departs from the Rouse model in dependences on time, momentum transfer, and temperature. To explain the anomalies, others had proposed the random Rouse model (RRM) in which each monomer has different mobility taken from a broad log-normal distribution. Despite the success of the RRM, Diddens, Brodeck and Heuer [EPL, 95, 56003 (2011)] extracted the distribution of friction coefficients from the MD simulations of a PEO/PMMA blend and found the distribution is much narrower than expected from the RRM. Read More

We investigate the physical effects of translational symmetry breaking in Fe-based high-temperature superconductors due to alternating anion positions. In the representative parent compounds, including the newly discovered Fe-vacancy-ordered $\mathrm{K_{0.8}Fe_{1. Read More

Direct numerical simulation (DNS) for gas-solid flow is implemented on a multi-scale supercomputing system, Mole-8.5, featuring massive parallel GPU-CPU hybrid computing, for which the lattice Boltzmann method (LBM) is deployed together with the immersed moving boundary (IMB) method and discrete element method (DEM). A two-dimensional suspension with about 1,166,400 75-micron solid particles distributed in an area of 11. Read More

We propose a new class of quintessence models in which late times oscillations of a scalar field give rise to an effective equation of state which can be negative and hence drive the observed acceleration of the universe. Our ansatz provides a unified picture of quintessence and a new form of dark matter we call "Frustrated Cold Dark Matter" (FCDM). FCDM inhibits gravitational clustering on small scales and could provide a natural resolution to the core density problem for disc galaxy halos. Read More

We calculate the detectability of the polarization of the cosmic microwave background (CMB) as a function of the sky coverage, angular resolution, and instrumental sensitivity for a hypothetical experiment. We consider the gradient component of the polarization from density perturbations (scalar modes) and the curl component from gravitational waves (tensor modes). We show that the amplitude (and thus the detectability) of the polarization from density perturbations is roughly the same in any model as long as the model fits the big-bang-nucleosynthesis (BBN) baryon density and degree-scale anisotropy measurements. Read More

We derive an expression for the non-Gaussian cosmic-microwave-background (CMB) statistic $I_l^3$ defined recently by Ferreira, Magueijo, and G\'orski in terms of the slow-roll-inflation parameters $\epsilon$ and $\eta$. This result shows that a nonzero value of $I_l^3$ in COBE would rule out single-field slow-roll inflation. A sharp change in the slope of the inflaton potential could increase the predicted value of $I_l^3$, but not significantly. Read More

Affiliations: 1IfA, Edinburgh, 2Columbia, NY, 3IfA, Edinburgh, 4Columbia, NY
Category: Astrophysics

Since cosmic-microwave-background (CMB) and large-scale-structure (LSS) data will shortly improve dramatically with the Microwave Anisotropy Probe (MAP) and Planck Surveyor, and the Anglo-Australian 2-Degree Field (2dF) and Sloan Digital Sky Survey (SDSS), respectively, it is timely to ask which of the CMB or LSS will provide a better probe of primordial non-gaussianity. In this paper we consider this question, using the bispectrum as a discriminating statistic. We consider several non-gaussian models and find that in each case the CMB will provide a better probe of primordial non-gaussianity. Read More


We present simple analytic approximations for the linear and fully evolved nonlinear mass power spectrum for spatially flat cold dark matter (CDM) cosmological models with quintessence (Q). Quintessence is a time evolving, spatially inhomogeneous energy component with negative pressure and an equation of state w_Q < 0. It clusters gravitationally on large length scales but remains smooth like the cosmological constant on small length scales. Read More

Affiliations: 1Columbia, 2Princeton, 3Princeton, 4Princeton
Category: Astrophysics

We present a comprehensive study of the observational constraints on spatially flat cosmological models containing a mixture of matter and quintessence --- a time varying, spatially inhomogeneous component of the energy density of the universe with negative pressure. Our study also includes the limiting case of a cosmological constant. Low red shift constraints include the Hubble parameter, baryon fraction, cluster abundance, age of the universe, bulk velocity and shape of the mass power spectrum; intermediate red shift constraints are due to type 1a supernovae, gravitational lensing, the Ly-a forest, and the evolution of large scale structure; high red shift constraints are based on cosmic microwave background temperature anisotropy. Read More

A substantial fraction of the energy density of the universe may consist of quintessence in the form of a slowly-rolling scalar field. Since the energy density of the scalar field generally decreases more slowly than the matter energy density, it appears that the ratio of the two densities must be set to a special, infinitesimal value in the early universe in order to have the two densities nearly coincide today. Recently, we introduced the notion of tracker fields to avoid this initial conditions problem. Read More

Does Nature yield any manifestations of parity violation other than those observed in weak interactions? A map of the cosmic microwave background (CMB) temperature and polarization will provide a new signature of P violation. We give two examples of new P violating interactions, which may have something to do with Planck-scale physics, inflation, and/or quintessence, that would give rise to such a signature. Although these effects would most likely elude detection by MAP and the Planck Surveyor, they may be detectable with a future dedicated CMB polarization experiment. Read More

Affiliations: 1University of Pennsylvania, 2University of Pennsylvania, 3University of Pennsylvania
Category: Astrophysics

Recent observations suggest that a large fraction of the energy density of the universe has negative pressure. One explanation is vacuum energy density; another is quintessence in the form of a scalar field slowly evolving down a potential. In either case, a key problem is to explain why the energy density nearly coincides with the matter density today. Read More

Some form of missing energy may account for the difference between the observed cosmic matter density and the critical density. Two leading candidates are a cosmological constant and quintessence (a time-varying, inhomogenous component with negative pressure). We show that an ideal, full-sky cosmic background anisotropy experiment may not be able to distinguish the two, even when non-linear effects due to gravitational lensing are included. Read More

Affiliations: 1University of Pennsylvania, 2University of Pennsylvania
Category: Astrophysics

The abundance of rich clusters is a strong constraint on the mass power spectrum. The current constraint can be expressed in the form $\sigma_8 \Omega_m^{\gamma} = 0.5 \pm 0. Read More

Affiliations: 1University of Pennsylvania, 2Institut fur Theoretische Physik, ETH, 3University of Pennsylvania

We examine the theoretical foundations of standard methods for computing density perturbations in inflationary models. We find that: (1) the time-delay formalism (introduced by Guth and Pi, 1982) is only valid when inflation is well-described by the de Sitter solution and the equation-of-state is nearly unchanging; and, (2) the horizon-crossing/Bessel approximation extends to non-exponential inflation, but only if the equation-of-state is changing slowly. Integration of the gauge-invariant perturbation equations mode-by-mode is the only method reliable for general models. Read More