arXiv subjects

Peng Wang

Publications and source records attributed to Peng Wang.

At least 865 records · Page 48Linked to original sources

arXiv2019

The ApolloScape Open Dataset for Autonomous Driving and its Application

Autonomous driving has attracted tremendous attention especially in the past few years. The key techniques for a self-driving car include solving tasks like 3D map construction, self-localization, parsing the driving road and understanding objects, which enable vehicles to reason and act. However, large scale data set for training and system evaluation is still a bottleneck for developing robust perception models. In this paper, we present the ApolloScape dataset [1] and its applications for autonomous driving. Compared with existing public datasets from real scenes, e.g. KITTI [2] or Cityscapes [3], ApolloScape contains much large and richer labelling including holistic semantic dense point cloud for each site, stereo, per-pixel semantic labelling, lanemark labelling, instance segmentation, 3D car instance, high accurate location for every frame in various driving videos from multiple sites, cities and daytimes. For each task, it contains at lease 15x larger amount of images than SOTA datasets. To label such a complete dataset, we develop various tools and algorithms specified for each task to accelerate the labelling process, such as 3D-2D segment labeling tools, active labelling in videos etc. Depend on ApolloScape, we are able to develop algorithms jointly consider the learning and inference of multiple tasks. In this paper, we provide a sensor fusion scheme integrating camera videos, consumer-grade motion sensors (GPS/IMU), and a 3D semantic map in order to achieve robust self-localization and semantic segmentation for autonomous driving. We show that practically, sensor fusion and joint learning of multiple tasks are beneficial to achieve a more robust and accurate system. We expect our dataset and proposed relevant algorithms can support and motivate researchers for further development of multi-sensor fusion and multi-task learning in the field of computer vision.

cs.CV↗

arXiv2019

Comparative analysis of layered structures in empirical investor networks and cellphone communication networks

Empirical investor networks (EIN) proposed by \cite{Ozsoylev-Walden-Yavuz-Bildik-2014-RFS} are assumed to capture the information spreading path among investors. Here, we perform a comparative analysis between the EIN and the cellphone communication networks (CN) to test whether EIN is an information exchanging network from the perspective of the layer structures of ego networks. We employ two clustering algorithms ($k$-means algorithm and $H/T$ break algorithm) to detect the layer structures for each node in both networks. We find that the nodes in both networks can be clustered into two groups, one that has a layer structure similar to the theoretical Dunbar Circle corresponding to that the alters in ego networks exhibit a four-layer hierarchical structure with the cumulative number of 5, 15, 50 and 150 from the inner layer to the outer layer, and the other one having an additional inner layer with about 2 alters compared with the Dunbar Circle. We also find that the scale ratios, which are estimated based on the unique parameters in the theoretical model of layer structures \citep{Tamarit-Cuesta-Dunbar-Sanchez-2018-PNAS}, conform to a log-normal distribution for both networks. Our results not only deepen our understanding on the topological structures of EIN, but also provide empirical evidence of the channels of information diffusion among investors.

q-fin.ST↗

arXiv2019

Evaluating Local Geometric Feature Representations for 3D Rigid Data Matching

Local geometric descriptors remain an essential component for 3D rigid data matching and fusion. The devise of a rotational invariant local geometric descriptor usually consists of two steps: local reference frame (LRF) construction and feature representation. Existing evaluation efforts have mainly been paid on the LRF or the overall descriptor, yet the quantitative comparison of feature representations remains unexplored. This paper fills this gap by comprehensively evaluating nine state-of-the-art local geometric feature representations. Our evaluation is on the ground that ground-truth LRFs are leveraged such that the ranking of tested feature representations are more convincing as opposed to existing studies. The experiments are deployed on six standard datasets with various application scenarios (shape retrieval, point cloud registration, and object recognition) and data modalities (LiDAR, Kinect, and Space Time) as well as perturbations including Gaussian noise, shot noise, data decimation, clutter, occlusion, and limited overlap. The evaluated terms cover the major concerns for a feature representation, e.g., distinctiveness, robustness, compactness, and efficiency. The outcomes present interesting findings that may shed new light on this community and provide complementary perspectives to existing evaluations on the topic of local geometric feature description. A summary of evaluated methods regarding their peculiarities is also presented to guide real-world applications and new descriptor crafting.

cs.CV↗

arXiv2019

Phase Structures and Transitions of Born-Infeld Black Holes in a Grand Canonical Ensemble

To make a Born-Infeld (BI) black hole thermally stable, we consider two types of boundary conditions, i.e., the asymptotically anti-de Sitter (AdS) space and a Dirichlet wall placed in the asymptotically flat space. The phase structures and transitions of these two types of BI black holes, namely BI-AdS black holes and BI black holes in a cavity, are investigated in a grand canonical ensemble, where the temperature and the potential are fixed. For BI-AdS black holes, the globally stable phases can be the thermal AdS space. For small values of the potential, there is a Hawking-Page-like first order phase transition between the BI-AdS black holes and the thermal-AdS space. However, the phase transition becomes zeroth order when the values of the potential are large enough. For BI black holes in a cavity, the globally stable phases can be a naked singularity or an extremal black hole with the horizon merging with the wall, which both are on the boundaries of the physical parameter region. The thermal flat space is never globally preferred. Besides a first order phase transition, there is a second order phase transition between the globally stable phases. Thus, it shows that the phase structures and transitions of BI black holes with these two different boundary conditions have several dissimilarities.

gr-qc↗

arXiv2019

Free-fall Rainbow BTZ Black Hole

Doubly special relativity (DSR) is an effective model for encoding quantum gravity in flat spacetime. To incorporate DSR into general relativity, one could use "Gravity's rainbow", where the spacetime background felt by a test particle would depend on its energy. For a black hole, there are two natural orthonormal frames, the stationary one hovering above it and freely falling one along geodesics. Since the rainbow metric is the metric that the radiated particles "see", a more natural orthonormal frame is the one anchored to the particles. And the cases with the stationary orthonormal frame have been extensively studied in the literature. In this paper, we investigate properties of rainbow BTZ black holes in the scenario with the free-fall orthonormal frame. We first review the thermodynamic properties of a BTZ black hole. Furthermore, we obtain the Free-fall (FF) rainbow BTZ black hole and then calculate its Hawking temperature via the Hamilton-Jacobi method. Finally, we discuss the thermodynamic properties of a FF stationary rainbow BTZ black hole .

gr-qc↗

arXiv2019

Piecewise classifier mappings: Learning fine-grained learners for novel categories with few examples

Humans are capable of learning a new fine-grained concept with very little supervision, \emph{e.g.}, few exemplary images for a species of bird, yet our best deep learning systems need hundreds or thousands of labeled examples. In this paper, we try to reduce this gap by studying the fine-grained image recognition problem in a challenging few-shot learning setting, termed few-shot fine-grained recognition (FSFG). The task of FSFG requires the learning systems to build classifiers for novel fine-grained categories from few examples (only one or less than five). To solve this problem, we propose an end-to-end trainable deep network which is inspired by the state-of-the-art fine-grained recognition model and is tailored for the FSFG task. Specifically, our network consists of a bilinear feature learning module and a classifier mapping module: while the former encodes the discriminative information of an exemplar image into a feature vector, the latter maps the intermediate feature into the decision boundary of the novel category. The key novelty of our model is a "piecewise mappings" function in the classifier mapping module, which generates the decision boundary via learning a set of more attainable sub-classifiers in a more parameter-economic way. We learn the exemplar-to-classifier mapping based on an auxiliary dataset in a meta-learning fashion, which is expected to be able to generalize to novel categories. By conducting comprehensive experiments on three fine-grained datasets, we demonstrate that the proposed method achieves superior performance over the competing baselines.

cs.CV↗

arXiv2019

The Origin of Lopsided Satellite Galaxy Distribution in Galaxy Pairs

It is well known that satellite galaxies are not isotropically distributed among their host galaxies as suggested by most interpretations of the $Λ$CDM model. One type of anisotropy recently detected in the SDSS (and seen when examining the distribution of satellites in the Local Group and in the Centaurus group) is a tendency to be so-called "lopsided". Namely, in pairs of galaxies (like Andromeda and the Milky Way) the satellites are more likely to inhabit the region in between the pair, rather than on opposing sides. Although recent studies found a similar set up when comparing pairs of galaxies in $Λ$CDM simulations indicating that such a set up is not inconsistent with $Λ$CDM, the origin has yet to be explained. Here we examine the origin of such lopsided setups by first identifying such distributions in pairs of galaxies in numerical cosmological simulations, and then tracking back the orbital trajectories of satellites (which at $z=0$ display the effect). We report two main results: first, the lopsided distribution was stronger in the past and weakens towards $z=0$. Second, the weakening of the signal is due to the interaction of satellite galaxies with the pair. Finally, we show that the $z=0$ signal is driven primarily by satellites that are on first approach, who have yet to experience a "flyby". This suggests that the signal seen in the observations is also dominated by dynamically young accretion events.

astro-ph.GA↗

arXiv2019

Contact transparency in mechanically assembled 2D material devices

Two-dimensional atomic crystals (2DACs) can be mechanically assembled with precision for the fabrication of heterostructures, allowing for the combination of material building blocks with great flexibility. In addition, while conventional nanolithography can be detrimental to most of the 2DACs which are not sufficiently inert, mechanical assembly potentially minimizes the nanofabrication processing and preserves the intrinsic physical properties of the 2DACs. In this work we study the interfacial charge transport between various 2DACs and electrical contacts, by fabricating and characterizing 2DAC-superconductor junctions through mechanical transfer. Compared to devices fabricated with conventional nanolithography, mechanically assembled devices show comparable or better interface transparency. Surface roughness at the electrical contacts is identified to be a major limitation to the interface quality.

cond-mat.mes-hall↗

arXiv2019

Flash: Efficient Dynamic Routing for Offchain Networks

Offchain networks emerge as a promising solution to address the scalability challenge of blockchain. Participants directly make payments through a network of payment channels without the overhead of committing onchain transactions. Routing is critical to the performance of offchain networks. Existing solutions use either static routing with poor performance or dynamic routing with high overhead to obtain the dynamic channel balance information. In this paper, we propose Flash, a new dynamic routing solution that leverages the unique characteristics of transactions in offchain networks to strike a better tradeoff between path optimality and probing overhead. By studying the traces of real offchain networks, we find that the payment sizes are heavy-tailed, and most payments are highly recurrent. Flash thus differentiates the treatment of elephant payments from mice payments. It uses a modified max-flow algorithm for elephant payments to find paths with sufficient capacity, and strategically routes the payment across paths to minimize the transaction fees. Mice payments are directly sent by looking up a routing table with a few precomputed paths to reduce probing overhead. Testbed experiments and data-driven simulations show that Flash improves the success volume of payments by up to 2.3x compared to the state-of-the-art routing algorithm.

cs.NI↗

arXiv2019

Can we find steady-state solutions to multiscale rarefied gas flows within dozens of iterations?

One of the central problems in the study of rarefied gas dynamics is to find the steady-state solution of the Boltzmann equation quickly. When the Knudsen number is large, i.e. the system is highly rarefied, the conventional iteration scheme can lead to convergence within a few iterations. However, when the Knudsen number is small, i.e. the flow falls in the near-continuum regime, hundreds of thousands iterations are needed, and yet the "converged" solutions are prone to be contaminated by accumulated error and large numerical dissipation. Recently, based on the gas kinetic models, the implicit unified gas kinetic scheme (UGKS) and its variants have significantly reduced the iterations in the near-continuum flow regime, but still much higher than that of the highly rarefied gas flows. In this paper, we put forward a general synthetic iteration scheme (GSIS) to find the steady-state solutions of general rarefied gas flows within dozens of iterations at any Knudsen number. As the GSIS does not rely on the specific kinetic model/collision operator, it can be naturally extended to quickly find converged solutions for mixture flows and even flows involving chemical reactions. These two superior advantages are also expected to accelerate the slow convergence in simulation of near-continuum flows via the direct simulation Monte Carlo method and its low-variance version.

physics.comp-ph↗

arXiv2019

On the Hopf Algebra of Rooted Trees

We find a formula to compute the number of the generators, which generate the $n$-filtered space of Hopf algebra of rooted trees, i.e. the number of equivalent classes of rooted trees with weight $n$. Applying Hopf algebra of rooted trees, we show that the analogue of Andruskiewitsch and Schneider's Conjecture is not true. The Hopf algebra of rooted trees and the enveloping algebra of the Lie algebra of rooted trees are two important examples of Hopf algebras. We give their representation and show that they have not any nonzero integrals. We structure their graded Drinfeld doubles and show that they are local quasitriangular Hopf algebras.

math-ph↗

arXiv2019

Training a Binary Weight Object Detector by Knowledge Transfer for Autonomous Driving

Autonomous driving has harsh requirements of small model size and energy efficiency, in order to enable the embedded system to achieve real-time on-board object detection. Recent deep convolutional neural network based object detectors have achieved state-of-the-art accuracy. However, such models are trained with numerous parameters and their high computational costs and large storage prohibit the deployment to memory and computation resource limited systems. Low-precision neural networks are popular techniques for reducing the computation requirements and memory footprint. Among them, binary weight neural network (BWN) is the extreme case which quantizes the float-point into just $1$ bit. BWNs are difficult to train and suffer from accuracy deprecation due to the extreme low-bit representation. To address this problem, we propose a knowledge transfer (KT) method to aid the training of BWN using a full-precision teacher network. We built DarkNet- and MobileNet-based binary weight YOLO-v2 detectors and conduct experiments on KITTI benchmark for car, pedestrian and cyclist detection. The experimental results show that the proposed method maintains high detection accuracy while reducing the model size of DarkNet-YOLO from 257 MB to 8.8 MB and MobileNet-YOLO from 193 MB to 7.9 MB.

cs.CV↗

arXiv2019

Thermodynamics and Weak Cosmic Censorship Conjecture in Nonlinear Electrodynamics Black Holes via Charged Particle Absorption

We first obtain the $D$-dimensional asymptotically AdS charged black hole solution in general nonlinear electrodynamics (NLED). We then use the Hamilton-Jacobi method to describe the motion in curved spacetime of a scalar particle and a fermion, which, as we show, satisfy the same Hamilton-Jacobi equation. With absorbing a charged particle, the variation of the generic charged NLED black hole is calculated in the normal and extended phase spaces. In the normal phase space, where the cosmological constant and dimensionful parameters in NLED are fixed, the first and second laws of thermodynamics are satisfied. In the extended phase space, where the cosmological constant and dimensionful parameters in NLED are treated as thermodynamic variables, the first law of thermodynamics is also satisfied. However, the black hole entropy can either increase or decrease depending on the changes in the dimensionful parameters. Furthermore, we find that the weak cosmic censorship conjecture is valid for the extremal and near-extremal black holes in the both phase spaces.

gr-qc↗

arXiv2019

Evaluation of the mechanical behaviour of tufted preforms

The weak properties in the thickness of the laminates in terms of rigidity, fatigue and impact resistance involve delamination sensitivity for thick structures. This issue has led to the development of three-dimensional (3D) preforms with reinforcement through-the-thickness, making it possible to combine good mechanical performance in-plane and out-of-plane. In this study, the tufting technology is used to strength through-the-thickness of stacked layers. This reinforcement technique has extensively gained attention in the composite industry due to its ability to reinforce thick pieces. The present study focuses on the mechanical behaviour of tufted preforms before resin impregnation due to their application in the forming steps during the manufacturing process. Therefore, an experimental investigation of the mechanical behaviour (tensile, bending and in-plane shear) was carried out in order to outline the influence of tufting on the mechanical properties of the final structure.

physics.app-ph↗

arXiv2019

Multi-Label Image Recognition with Graph Convolutional Networks

The task of multi-label image recognition is to predict a set of object labels that present in an image. As objects normally co-occur in an image, it is desirable to model the label dependencies to improve the recognition performance. To capture and explore such important dependencies, we propose a multi-label classification model based on Graph Convolutional Network (GCN). The model builds a directed graph over the object labels, where each node (label) is represented by word embeddings of a label, and GCN is learned to map this label graph into a set of inter-dependent object classifiers. These classifiers are applied to the image descriptors extracted by another sub-net, enabling the whole network to be end-to-end trainable. Furthermore, we propose a novel re-weighted scheme to create an effective label correlation matrix to guide information propagation among the nodes in GCN. Experiments on two multi-label image recognition datasets show that our approach obviously outperforms other existing state-of-the-art methods. In addition, visualization analyses reveal that the classifiers learned by our model maintain meaningful semantic topology.

cs.CV↗

arXiv2019

Vehicle Re-identification in Aerial Imagery: Dataset and Approach

In this work, we construct a large-scale dataset for vehicle re-identification (ReID), which contains 137k images of 13k vehicle instances captured by UAV-mounted cameras. To our knowledge, it is the largest UAV-based vehicle ReID dataset. To increase intra-class variation, each vehicle is captured by at least two UAVs at different locations, with diverse view-angles and flight-altitudes. We manually label a variety of vehicle attributes, including vehicle type, color, skylight, bumper, spare tire and luggage rack. Furthermore, for each vehicle image, the annotator is also required to mark the discriminative parts that helps them to distinguish this particular vehicle from others. Besides the dataset, we also design a specific vehicle ReID algorithm to make full use of the rich annotation information. It is capable of explicitly detecting discriminative parts for each specific vehicle and significantly outperforms the evaluated baselines and state-of-the-art vehicle ReID approaches.

cs.CV↗

arXiv2019

Pixel-aware Deep Function-mixture Network for Spectral Super-Resolution

Spectral super-resolution (SSR) aims at generating a hyperspectral image (HSI) from a given RGB image. Recently, a promising direction for SSR is to learn a complicated mapping function from the RGB image to the HSI counterpart using a deep convolutional neural network. This essentially involves mapping the RGB context within a size-specific receptive field centered at each pixel to its spectrum in the HSI. The focus thereon is to appropriately determine the receptive field size and establish the mapping function from RGB context to the corresponding spectrum. Due to their differences in category or spatial position, pixels in HSIs often require different-sized receptive fields and distinct mapping functions. However, few efforts have been invested to explicitly exploit this prior. To address this problem, we propose a pixel-aware deep function-mixture network for SSR, which is composed of a new class of modules, termed function-mixture (FM) blocks. Each FM block is equipped with some basis functions, i.e., parallel subnets of different-sized receptive fields. Besides, it incorporates an extra subnet as a mixing function to generate pixel-wise weights, and then linearly mixes the outputs of all basis functions with those generated weights. This enables us to pixel-wisely determine the receptive field size and the mapping function. Moreover, we stack several such FM blocks to further increase the flexibility of the network in learning the pixel-wise mapping. To encourage feature reuse, intermediate features generated by the FM blocks are fused in late stage, which proves to be effective for boosting the SSR performance. Experimental results on three benchmark HSI datasets demonstrate the superiority of the proposed method.

cs.CV↗

arXiv2019

The shape alignment of satellite galaxies in galaxy pairs in SDSS

It has been shown, both in simulations and observationally, that the tidal field of a large galaxy can torque its satellites such that the major axis of satellite galaxies points towards their hosts. This so-called `shape alignment' has been observed in isolated Milky Way-like galaxies but not in `Local Group'-like pairs. In this study, we investigate the shape alignment of satellite galaxies in galaxy pairs similar to the Local Group identified in the Sloan Digital Sky Survey Data Release 13 (SDSS DR13). By stacking tens of thousands of satellite galaxies around primary galaxy pairs, we find two statistically strong alignment signals. (1) The major axes of satellite galaxies located in the (projected) area between two primaries (the {\it facing} region) tend to be perpendicular to the line connecting the satellite to its host (tangential alignment), while (2) the major axes of satellite galaxies located in regions away from the other host (the {\it away} region) tend to be aligned with the line connecting the satellite to its host (radial alignment). These alignments are confirmed at $\sim5σ$ levels. The alignment signal increases with increasing primary brightness, decreasing pair separation, and decreasing satellite distance. The alignment signal is also found to be stronger in filamentary environments. These findings will shed light on understanding the mechanisms of how satellite galaxies are affected by the tidal field in galaxy pairs and will be useful for investigating galaxy intrinsic alignment in the analyses of weak gravitational lensing.

astro-ph.GA↗