arXiv subjects

Jianan Li

Publications and source records attributed to Jianan Li.

At least 109 records · Page 6Linked to original sources

arXiv2020

Temporal Graph Modeling for Skeleton-based Action Recognition

Graph Convolutional Networks (GCNs), which model skeleton data as graphs, have obtained remarkable performance for skeleton-based action recognition. Particularly, the temporal dynamic of skeleton sequence conveys significant information in the recognition task. For temporal dynamic modeling, GCN-based methods only stack multi-layer 1D local convolutions to extract temporal relations between adjacent time steps. With the repeat of a lot of local convolutions, the key temporal information with non-adjacent temporal distance may be ignored due to the information dilution. Therefore, these methods still remain unclear how to fully explore temporal dynamic of skeleton sequence. In this paper, we propose a Temporal Enhanced Graph Convolutional Network (TE-GCN) to tackle this limitation. The proposed TE-GCN constructs temporal relation graph to capture complex temporal dynamic. Specifically, the constructed temporal relation graph explicitly builds connections between semantically related temporal features to model temporal relations between both adjacent and non-adjacent time steps. Meanwhile, to further explore the sufficient temporal dynamic, multi-head mechanism is designed to investigate multi-kinds of temporal relations. Extensive experiments are performed on two widely used large-scale datasets, NTU-60 RGB+D and NTU-120 RGB+D. And experimental results show that the proposed model achieves the state-of-the-art performance by making contribution to temporal modeling for action recognition.

cs.CV↗

arXiv2020

Constraining the quasar radio-loud fraction at $z \sim 6$ with deep radio observations

We carry out a series of deep Karl G. Jansky Very Large Array (VLA) S-band observations of a sample of 21 quasars at $z\sim6$. The new observations expand the searches of radio continuum emission to the optically faint quasar population at the highest redshift with rest-frame $4400 \rm Å$ luminosities down to $3 \times10^{11} \ L_{\odot}$. We report the detections of two new radio-loud quasars: CFHQS J2242+0334 (hereafter J2242+0334) at $z=5.88$ and CFHQS J0227$-$0605 (hereafter J0227$-$0605) at $z=6.20$, detected with 3 GHz flux densities of $87.0 \pm 6.3 \ μ\rm Jy$ and $55.4 \pm 6.7 \ μ\rm Jy$, respectively. Their radio \replaced{loudness}{loudnesses} are estimated to be $54.9 \pm 4.7$ and $16.5 \pm 3.2$, respectively. To better constrain the radio-loud fraction (RLF), we combine the new measurements with the archival VLA L-band data as well as available data from the literature, considering the upper limits for non-detections and \deleted{and} possible selection effects. The final derived RLF is $9.4 \pm 5.7\%$ for the optically selected quasars at $z\sim6$. We also compare the RLF to that of the quasar samples at low redshift and check the RLF in different quasar luminosity bins. The RLF for the optically faint objects is still poorly constrained due to the limited sample size. Our \replaced{result}{results} show no evidence of significant quasar RLF evolution with redshift. There is also no clear trend of RLF evolution with quasar UV/optical luminosity due to the limited sample size of optically faint objects with deep radio observations.

astro-ph.GA↗

arXiv2020

Attribute-conditioned Layout GAN for Automatic Graphic Design

Modeling layout is an important first step for graphic design. Recently, methods for generating graphic layouts have progressed, particularly with Generative Adversarial Networks (GANs). However, the problem of specifying the locations and sizes of design elements usually involves constraints with respect to element attributes, such as area, aspect ratio and reading-order. Automating attribute conditional graphic layouts remains a complex and unsolved problem. In this paper, we introduce Attribute-conditioned Layout GAN to incorporate the attributes of design elements for graphic layout generation by forcing both the generator and the discriminator to meet attribute conditions. Due to the complexity of graphic designs, we further propose an element dropout method to make the discriminator look at partial lists of elements and learn their local patterns. In addition, we introduce various loss designs following different design principles for layout optimization. We demonstrate that the proposed method can synthesize graphic layouts conditioned on different element attributes. It can also adjust well-designed layouts to new sizes while retaining elements' original reading-orders. The effectiveness of our method is validated through a user study.

cs.CV↗

arXiv2020

SCUBA2 High Redshift Bright Quasar Survey: Far-infrared Properties and Weak-line Features

We present a submillimetre continuum survey ('SCUBA2 High rEdshift bRight quasaR surveY', hereafter SHERRY) of 54 high redshift quasars at $5.6 3.5σ$). The new SHERRY detections indicate far-infrared (FIR) luminosities of $\rm 3.5\times10^{12}$ to $\rm 1.4\times10^{13}$ $L_{\odot}$, implying extreme star formation rates of 90 to 1060 $M_{\odot}$ yr$^{-1}$ in the quasar host galaxies. Compared with $z =$ 2$-$5 samples, the FIR luminous quasars ($L_{\rm FIR} > 10^{13}\,L_{\odot}$) are more rare at $z \sim 6$. The optical/near-infrared (NIR) spectra of these objects show 11% (6/54) of the sources have weak Ly$α$, emission line features, which may relate to different sub-phases of the central active galactic nuclei (AGNs). Our SCUBA2 survey confirms the trend reported in the literature that quasars with submillimeter detections tend to have weaker ultraviolet (UV) emission lines compared to quasars with nondetections. The connection between weak UV quasar line emission and bright dust continuum emission powered by massive star formation may suggest an early phase of AGN-galaxy evolution, in which the broad line region is starting to develop slowly or is shielded from the central ionization source, and has unusual properties such as weak line features or bright FIR emission.

astro-ph.GA↗

arXiv2020

Ionized and atomic interstellar medium in the z = 6.003 quasar SDSS J2310+1855

Observing the interstellar medium (ISM) in $z \gtrsim 6$ quasars host galaxies is essential for understanding the co-evolution between the supermassive black holes and their hosts. To probe the gas physical conditions and search for imprints of Active Galactic Nuclei (AGN) on the ISM, we report ALMA observations of the $\rm [N\ II]_{122 μm}$ and $\rm [O\ I]_{146 μm}$ lines and the underlying continuum from the $z=6.003$ quasar SDSS J231038.88+185519.7. Together with previous $\rm [C\ II]_{158 μm}$ and $\rm [O\ III]_{88 μm}$ observations, we use the ratios of these fine-structure lines to probe the ISM properties. Similar to other high-$z$ systems, this object exhibits a $\rm [C\ II]_{158 μm}$/$\rm [O\ I]_{146 μm}$ ratio comparable to the lowest values found in local (Ultra) luminous infrared galaxies, suggesting a "warmer" and "denser" gas component compared to typical local systems. The $\rm [O\ III]_{88 μm}$/$\rm [O\ I]_{146 μm}$ ratio is lower than that of other local and high-$z$ systems, indicating a smaller ionized gas fraction in this quasar. The $\rm [O\ III]_{88 μm}$/$\rm [N\ II]_{122 μm}$ ratio is comparable to that of local systems, and suggests a metallicity of $Z/Z_{\odot}$=1.5$-$2.1. Based on the $\rm [N\ II]_{122 μm}$ detection, we estimate that $17\%$ of the $\rm [C\ II]_{158 μm}$ emission is associated with ionized gas. The $\rm [N\ II]_{122 μm}$ line shows a "flux deficit" comparable to local systems. The $\rm [O\ I]_{146 μm}$ line, with a $\rm [O\ I]_{146 μm}$/FIR ratio $\ge 2\times$ than expected from the local relation, indicates no $\rm [O\ I]_{\rm 146 μm}$ deficit. The low $\rm [C\ II]_{158 μm}$/$\rm [O\ I]_{146 μm}$ ratio, together with the high $\rm [O\ I]_{146 μm}$/FIR ratio in J2310+1855, reveals that the warm and dense gas is likely a result of AGN heating to the ISM.

astro-ph.GA↗

arXiv2020

Local Grid Rendering Networks for 3D Object Detection in Point Clouds

The performance of 3D object detection models over point clouds highly depends on their capability of modeling local geometric patterns. Conventional point-based models exploit local patterns through a symmetric function (e.g. max pooling) or based on graphs, which easily leads to loss of fine-grained geometric structures. Regarding capturing spatial patterns, CNNs are powerful but it would be computationally costly to directly apply convolutions on point data after voxelizing the entire point clouds to a dense regular 3D grid. In this work, we aim to improve performance of point-based models by enhancing their pattern learning ability through leveraging CNNs while preserving computational efficiency. We propose a novel and principled Local Grid Rendering (LGR) operation to render the small neighborhood of a subset of input points into a low-resolution 3D grid independently, which allows small-size CNNs to accurately model local patterns and avoids convolutions over a dense grid to save computation cost. With the LGR operation, we introduce a new generic backbone called LGR-Net for point cloud feature extraction with simple design and high efficiency. We validate LGR-Net for 3D object detection on the challenging ScanNet and SUN RGB-D datasets. It advances state-of-the-art results significantly by 5.5 and 4.5 mAP, respectively, with only slight increased computation overhead.

cs.CV↗

arXiv2019

Probing the full CO spectral line energy distribution (SLED) in the nuclear region of a quasar-starburst system at $z=6.003$

We report Atacama Large Millimeter/submillimeter Array (ALMA) observations of CO $(8-7)$, $(9-8)$, $\rm H_{2}O (2_{0,2}-1_{1,1})$ and $\rm OH^{+} (1_{1}-0_{1})$ and NOrthern Extended Millimeter Array (NOEMA) observations of CO $(5-4)$, $(6-5)$, $(12-11)$ and $(13-12)$ towards the $z = 6.003$ quasar SDSS J231038.88+185519.7, aiming to probe the physical conditions of the molecular gas content of this source. We present the best sampled CO spectral line energy distribution (SLED) at $z = 6.003$, and analyzed it with the radiative transfer code MOLPOP-CEP. Fitting the CO SLED to a one-component model indicates a kinetic temperature $T_{\rm kin} = 228 \ \rm K$, molecular gas density $log (n(\rm H_{2})/\rm cm^{-3}$ )=4.75, and CO column density $log(N(\rm CO)/\rm cm^{-2}) =17.5$, although a two-component model better fits the data. In either case, the CO SLED is dominated by a "warm" and "dense" component. Compared to samples of local (Ultra) Luminous Infrared Galaxies ((U)LIRGs), starburst galaxies and high redshift Submillimeter Galaxies (SMGs), J2310+1855 exhibits higher CO excitation at ($J \geq 8$), like other high redshift quasars. The high CO excitation, together with the enhanced $L_{\rm H_{2}O}/ L_{IR} $, $L_{\rm H_{2}O}/ L_{CO} $ and $L_{OH^{+}}/L_{\rm H_{2}O} $ ratios, suggests that besides the UV radiation from young massive stars, other mechanisms such as shocks, cosmic rays and X-rays might also be responsible for the heating and ionization of the molecular gas. In the nuclear region probed by the molecular emissions lines, any of these mechanisms might be present due to the powerful quasar and the starburst activity.

astro-ph.GA↗

arXiv2019

One-dimensional Kronig-Penney superlattices at the LaAlO$_3$/SrTiO$_3$ interface

The paradigm of electrons interacting with a periodic lattice potential is central to solid-state physics. Semiconductor heterostructures and ultracold neutral atomic lattices capture many of the essential properties of 1D electronic systems. However, fully one-dimensional superlattices are highly challenging to fabricate in the solid state due to the inherently small length scales involved. Conductive atomic-force microscope (c-AFM) lithography has recently been demonstrated to create ballistic few-mode electron waveguides with highly quantized conductance and strongly attractive electron-electron interactions. Here we show that artificial Kronig-Penney-like superlattice potentials can be imposed on such waveguides, introducing a new superlattice spacing that can be made comparable to the mean separation between electrons. The imposed superlattice potential "fractures" the electronic subbands into a manifold of new subbands with magnetically-tunable fractional conductance (in units of $e^2/h$). The lowest $G=2e^2/h$ plateau, associated with ballistic transport of spin-singlet electron pairs, is stable against de-pairing up to the highest magnetic fields explored ($|B|=16$ T). A 1D model of the system suggests that an engineered spin-orbit interaction in the superlattice contributes to the enhanced pairing observed in the devices. These findings represent an important advance in the ability to design new families of quantum materials with emergent properties, and mark a milestone in the development of a solid-state 1D quantum simulation platform.

cond-mat.str-el↗

arXiv2019

Engineered spin-orbit interactions in LaAlO$_3$/SrTiO$_3$-based 1D serpentine electron waveguides

The quest to understand, design, and synthesize new forms of quantum matter guides much of contemporary research in condensed matter physics. One-dimensional (1D) electronic systems form the basis for some of the most interesting and exotic phases of quantum matter. The variety of experimentally-accessible ballistic 1D electronic systems is highly restricted, and furthermore these systems typically have few tuning parameters other than electric and magnetic fields. However, electron waveguides formed from two-dimensional (2D) LaAlO$_3$/SrTiO$_3$ heterointerfaces exhibit remarkable 1D properties, including ballistic multi-mode transport and strong attractive electron-electron interaction, but these systems conspicuously lack strong or tunable spin-orbit interactions. Here we describe a new class of quasi-1D nanostructures, based on LaAlO$_3$/SrTiO$_3$ electron waveguides, in which a sinusoidal transverse spatial modulation is imposed. Nanowires created with this "serpentine" modulation display unique dispersive features in the subband spectra, namely (1) a significant shift ($\sim$ 7 tesla) in the spin-dependent subband minima, and (2) fractional conductance plateaus, some of which are continuously tunable with a magnetic field. The first property can be understood as an engineered spin-orbit interaction associated with the periodic acceleration of electrons as they undulate through the nanowire (ballistically), while the second property signifies the presence of enhanced electron-electron scattering in this system due to the imposed periodic structure. The ability to engineer these interactions in quantum wires contributes to the tool set of a 1D solid-state quantum simulation platform.

cond-mat.str-el↗

arXiv2019

Gate-Tunable Optical Nonlinearities and Extinction in Graphene/LaAlO$_3$/SrTiO$_3$ Nanostructures

Pristine, undoped graphene has a constant absorption of 2.3 % across the visible to near-infrared (VIS-NIR) region of the electromagnetic spectrum. Under certain conditions, such as nanostructuring and intense gating, graphene can interact more robustly with VIS-NIR light and exhibit a large nonlinear optical response. Here, we explore the optical properties of graphene/LaAlO$_3$/SrTiO$_3$ nanostructures, where nanojunctions formed at the LaAlO$_3$/SrTiO$_3$ interface enable large (~10$^8$ V/m) electric fields to be applied to graphene over a scale of ~10 nm. Upon illumination with ultrafast VIS-NIR light, graphene/LaAlO$_3$/SrTiO$_3$ nanostructures produce broadband THz emission as well as a sum-frequency generated (SFG) response. Strong spectrally sharp, gate-tunable extinction features (>99.99%) are observed in both the VIS-NIR and SFG regions alongside significant intensification of the nonlinear response. The observed gate-tunable strong graphene-light interaction and nonlinear optical response are of fundamental interest and open the way for future exploitation in graphene-based optical devices.

cond-mat.mes-hall↗

arXiv2019

Minimal Non-orthogonal Gate Decomposition for Qubits with Limited Control

In quantum control theory, a question of fundamental and practical interest is how an arbitrary unitary transformation can be decomposed into minimum number of elementary rotations for implementation, subject to various physical constraints. Examples include the singlet-triplet (ST) and exchange-only (EO) qubits in quantum-dot systems, and gate construction in the Solovay-Kitaev algorithm. For two important scenarios, we present complete solutions to the problems of optimal decomposition of single-qubit unitary gates with non-orthogonal rotations. For each unitary gate, we give the criteria for determining the minimal number of pieces, the explicit gate construction procedure, as well as a computer code for practical uses. Our results include an analytic explanation to the four-gate decomposition of EO qubits, previously determined numerically by Divincenzo et al [Nature, 408, 339 (2000)]. Furthermore, compared with the approaches of Ramon sequence and its variant [Phys. Rev. Lett., 118, 216802 (2017)], our method can reduce about 50% of gate time for ST qubits. Finally, our approach can be extended to solve the problem of optimal control of topological qubits, where gate construction is achieved through the braiding operations.

cond-mat.mes-hall↗

arXiv2019

Reconfigurable edge-state engineering in graphene using LaAlO$_3$/SrTiO$_3$ nanostructures

The properties of graphene depend sensitively on doping with respect to the charge-neutrality point (CNP). Tuning the CNP usually requires electrical gating or chemical doping. Here, we describe a technique to reversibly control the CNP in graphene with nanoscale precision, utilizing LaAlO$_3$/SrTiO$_3$ (LAO/STO) heterostructures and conductive atomic force microscope (c-AFM) lithography. The local electron density and resulting conductivity of the LAO/STO interface can be patterned with a conductive AFM tip, and placed within two nanometers of an active graphene device. The proximal LAO/STO nanostructures shift the position of graphene CNP by ~ $10^{12}$ cm$^{-2}$, and are also gateable. Here we use this effect to create reconfigurable edge states in graphene, which are probed using the quantum Hall effect. Quantized resistance plateaus at $h/e^2$ and $h/3e^2$ are observed in a split Hall device, demonstrating edge transport along the c-AFM written edge that depends on the polarity of both the magnetic field and direction of currents. This technique can be readily extended to other device geometries.

cond-mat.mes-hall↗

arXiv2019

Star Formation and ISM Properties in Host Galaxies of Three Far-infrared Luminous Quasars at $z\sim6$

We present Karl G. Jansky Very Large Array (VLA) observations of the CO (2$-$1) line emission towards three far-infrared luminous quasars at $z\sim6$: SDSS J231038.88$+$185519.7 and SDSS J012958.51$-$003539.7 with $\sim0\farcs6$ resolution and SDSS J205406.42$-$000514.8 with $\sim2\farcs1$ resolution. All three sources are detected in the CO (2$-$1) line emission -- one source is marginally resolved, and the other two appear as point sources. Measurements of the CO (2$-$1) line emission allow us to calculate the molecular gas mass even without a CO excitation model. The inferred molecular gas masses are (0.8$-$4.3) $\times$ 10$^{10}$ $M_{\odot}$. The widths and redshifts derived from the CO (2$-$1) line are consistent with previous CO (6$-$5) and [\ion{C}{2}] measurements. We also report continuum measurements using the Herschel for SDSS J231038.88$+$185519.7 and SDSS J012958.51$-$003539.7, and for SDSS J231038.88+185519.7, data obtained at $\sim140$ and $\sim300$ GHz using the Atacama Large Millimeter/submillimeter Array (ALMA). In the case of SDSS J231038.88+185519.7, we present a detailed analysis of the spectral energy distribution and derive the dust temperature ($\sim40$ K), the dust mass ($\sim10^{9}$ $M_{\odot}$), the far-infrared luminosity (8$-$1000 $μ$m; $\sim10^{13}$ $ L_{\odot}$) and the star formation rate (2400$-$2700 $M_{\odot}$ yr$^{-1}$). Finally, an analysis of the photo-dissociation regions associated with the three high redshift quasars indicates that the interstellar medium in these sources has similar properties to local starburst galaxies.

astro-ph.GA↗

arXiv2019

LayoutGAN: Generating Graphic Layouts with Wireframe Discriminators

Layout is important for graphic design and scene generation. We propose a novel Generative Adversarial Network, called LayoutGAN, that synthesizes layouts by modeling geometric relations of different types of 2D elements. The generator of LayoutGAN takes as input a set of randomly-placed 2D graphic elements and uses self-attention modules to refine their labels and geometric parameters jointly to produce a realistic layout. Accurate alignment is critical for good layouts. We thus propose a novel differentiable wireframe rendering layer that maps the generated layout to a wireframe image, upon which a CNN-based discriminator is used to optimize the layouts in image space. We validate the effectiveness of LayoutGAN in various experiments including MNIST digit generation, document layout generation, clipart abstract scene generation and tangram graphic design.

cs.CV↗

arXiv2017

Direct Imaging of Sketched Conductive Nanostructures at the LaAlO3/SrTiO3 Interface

Nanoscale control of the quasi-two-dimensional electron gas at the LaAlO3/SrTiO3 (LAO/STO) interface by a conductive probe tip has triggered the development of a number of electronic devices. While the spatial distribution of the conductance is crucial for such devices, it is challenging to directly visualize the local electrical properties at the buried interface. Here we demonstrate conductivity imaging of sketched nanostructures at the LAO/STO interface by microwave impedance microscopy (MIM) with a lateral resolution on the order of 100 nm. The sheet conductance extracted from the MIM data agrees with the transport measurement. The tip-induced insulator-to-metal transition is observed above a threshold voltage of +4 V. Our work paves the way to study emergent phenomena at oxide interfaces by probing nanoscale conductance distribution.

cond-mat.mes-hall↗

arXiv2017

Integrated Face Analytics Networks through Cross-Dataset Hybrid Training

Face analytics benefits many multimedia applications. It consists of a number of tasks, such as facial emotion recognition and face parsing, and most existing approaches generally treat these tasks independently, which limits their deployment in real scenarios. In this paper we propose an integrated Face Analytics Network (iFAN), which is able to perform multiple tasks jointly for face analytics with a novel carefully designed network architecture to fully facilitate the informative interaction among different tasks. The proposed integrated network explicitly models the interactions between tasks so that the correlations between tasks can be fully exploited for performance boost. In addition, to solve the bottleneck of the absence of datasets with comprehensive training data for various tasks, we propose a novel cross-dataset hybrid training strategy. It allows "plug-in and play" of multiple datasets annotated for different tasks without the requirement of a fully labeled common dataset for all the tasks. We experimentally show that the proposed iFAN achieves state-of-the-art performance on multiple face analytics tasks using a single integrated model. Specifically, iFAN achieves an overall F-score of 91.15% on the Helen dataset for face parsing, a normalized mean error of 5.81% on the MTFL dataset for facial landmark localization and an accuracy of 45.73% on the BNU dataset for emotion recognition with a single model.

cs.CV↗

arXiv2017

Dual Path Networks

In this work, we present a simple, highly efficient and modularized Dual Path Network (DPN) for image classification which presents a new topology of connection paths internally. By revealing the equivalence of the state-of-the-art Residual Network (ResNet) and Densely Convolutional Network (DenseNet) within the HORNN framework, we find that ResNet enables feature re-usage while DenseNet enables new features exploration which are both important for learning good representations. To enjoy the benefits from both path topologies, our proposed Dual Path Network shares common features while maintaining the flexibility to explore new features through dual path architectures. Extensive experiments on three benchmark datasets, ImagNet-1k, Places365 and PASCAL VOC, clearly demonstrate superior performance of the proposed DPN over state-of-the-arts. In particular, on the ImagNet-1k dataset, a shallow DPN surpasses the best ResNeXt-101(64x4d) with 26% smaller model size, 25% less computational cost and 8% lower memory consumption, and a deeper DPN (DPN-131) further pushes the state-of-the-art single model performance with about 2 times faster training speed. Experiments on the Places365 large-scale scene dataset, PASCAL VOC detection dataset, and PASCAL VOC segmentation dataset also demonstrate its consistently better performance than DenseNet, ResNet and the latest ResNeXt model over various applications.

cs.CV↗

arXiv2017

Perceptual Generative Adversarial Networks for Small Object Detection

Detecting small objects is notoriously challenging due to their low resolution and noisy representation. Existing object detection pipelines usually detect small objects through learning representations of all the objects at multiple scales. However, the performance gain of such ad hoc architectures is usually limited to pay off the computational cost. In this work, we address the small object detection problem by developing a single architecture that internally lifts representations of small objects to "super-resolved" ones, achieving similar characteristics as large objects and thus more discriminative for detection. For this purpose, we propose a new Perceptual Generative Adversarial Network (Perceptual GAN) model that improves small object detection through narrowing representation difference of small objects from the large ones. Specifically, its generator learns to transfer perceived poor representations of the small objects to super-resolved ones that are similar enough to real large objects to fool a competing discriminator. Meanwhile its discriminator competes with the generator to identify the generated representation and imposes an additional perceptual requirement - generated representations of small objects must be beneficial for detection purpose - on the generator. Extensive evaluations on the challenging Tsinghua-Tencent 100K and the Caltech benchmark well demonstrate the superiority of Perceptual GAN in detecting small objects, including traffic signs and pedestrians, over well-established state-of-the-arts.

cs.CV↗