arXiv subjects

Wei Li

Publications and source records attributed to Wei Li.

At least 1,081 records · Page 60Linked to original sources

arXiv2020

Can scaling analysis be used to interpret the anti-parity-time symmetry in heat transfer?

In a previous work (Li et al. Science 364, 170) [1], we proposed a heat transfer system that preserves the anti-parity-time (APT) symmetry, and observe the rest-to-motion phase transition during the symmetry breaking. Recently, it was suggested (Zhao et al. arXiv:1906.08431) [2] that the behaviours of the system can be understood using scaling analysis based on the Péclet and Nusselt numbers (Pe and Nu). It was further proposed that there exists a third regime in the phase diagram in addition to the symmetric and symmetry broken phases. Although we appreciate the proposal to characterize the contributions of coupling, diffusion, and advection with dimensionless numbers, here we show that they do not help to predict or interpret the behaviours of the APT system. The dimensionless numbers do not provide enough details about the system to conclude that there is a motionless phase, a phase transition, to find the critical point, or to give the correct phase diagram with only two regimes.

physics.flu-dyn↗

arXiv2020

Transformer based Multilingual document Embedding model

One of the current state-of-the-art multilingual document embedding model LASER is based on the bidirectional LSTM neural machine translation model. This paper presents a transformer-based sentence/document embedding model, T-LASER, which makes three significant improvements. Firstly, the BiLSTM layers is replaced by the attention-based transformer layers, which is more capable of learning sequential patterns in longer texts. Secondly, due to the absence of recurrence, T-LASER enables faster parallel computations in the encoder to generate the text embedding. Thirdly, we augment the NMT translation loss function with an additional novel distance constraint loss. This distance constraint loss would further bring the embeddings of parallel sentences close together in the vector space; we call the T-LASER model trained with distance constraint, cT-LASER. Our cT-LASER model significantly outperforms both BiLSTM-based LASER and the simpler transformer-based T-LASER.

cs.CL↗

arXiv2020

NMT-based Cross-lingual Document Embeddings

This paper investigates a cross-lingual document embedding method that improves the current Neural machine Translation framework based Document Vector (NTDV or simply NV). NV is developed with a self-attention mechanism under the neural machine translation (NMT) framework. In NV, each pair of parallel documents in different languages are projected to the same shared layer in the model. However, the pair of NV embeddings are not guaranteed to be similar. This paper further adds a distance constraint to the training objective function of NV so that the two embeddings of a parallel document are required to be as close as possible. The new method will be called constrained NV (cNV). In a cross-lingual document classification task, the new cNV performs as well as NV and outperforms other published studies that require forward-pass decoding. Compared with the previous NV, cNV does not need a translator during testing, and so the method is lighter and more flexible.

cs.CL↗

arXiv2020

Transforming heat transfer with thermal metamaterials and devices

The demand for sophisticated tools and approaches in heat management and control has triggered fast development of emerging fields including conductive thermal metamaterials, nanophononics, far-field and near-field radiative thermal management, etc. In this review, we cast a unified perspective on the control of heat transfer, based on which the related studies can be considered as complementary paradigms toward manipulating physical parameters and realizing unprecedented phenomena in heat transfer using artificial structures, such as thermal conductivity in heat conduction, thermal emissivity in radiation, and properties related to multi-physical effects. The review is divided into three parts that focus on the three main categories of heat flow control, respectively. Thermal conduction and radiation are emphasized in the first and second parts at both macro- and micro-scale. The third part discusses the efforts to actively introduce heat sources or tune the material parameters with multi-physical effects in both conduction and radiation, including works using thermal convection. We conclude the review with challenges in this research topic and new possibilities about topological thermal effects, heat waves, and quantum thermal effects.

physics.app-ph↗

arXiv2020

Structure-Aware Network for Lane Marker Extraction with Dynamic Vision Sensor

Lane marker extraction is a basic yet necessary task for autonomous driving. Although past years have witnessed major advances in lane marker extraction with deep learning models, they all aim at ordinary RGB images generated by frame-based cameras, which limits their performance in extreme cases, like huge illumination change. To tackle this problem, we introduce Dynamic Vision Sensor (DVS), a type of event-based sensor to lane marker extraction task and build a high-resolution DVS dataset for lane marker extraction. We collect the raw event data and generate 5,424 DVS images with a resolution of 1280$\times$800 pixels, the highest one among all DVS datasets available now. All images are annotated with multi-class semantic segmentation format. We then propose a structure-aware network for lane marker extraction in DVS images. It can capture directional information comprehensively with multidirectional slice convolution. We evaluate our proposed network with other state-of-the-art lane marker extraction models on this dataset. Experimental results demonstrate that our method outperforms other competitors. The dataset is made publicly available, including the raw event data, accumulated images and labels.

cs.CV↗

arXiv2020

Gluing affine Yangians with bi-fundamentals

The affine Yangian of $\mathfrak{gl}_1$ is isomorphic to the universal enveloping algebra of $\mathcal{W}_{1+\infty}$ and can serve as a building block in the construction of new vertex operator algebras. In [1], a two-parameter family generalization of $\mathcal{N}=2$ supersymmetric $\mathcal{W}_{\infty}$ algebra was constructed by "gluing" two affine Yangians of $\mathfrak{gl}_1$ using operators that transform as $(\square, \overline{\square})$ and $(\overline{\square}, \square)$ w.r.t. the two affine Yangians. In this paper we realize a similar (but non-isomorphic) two-parameter gluing construction where the gluing operators transform as $(\square, \square)$ and $(\overline{\square}, \overline{\square})$ w.r.t. the two affine Yangians. The corresponding representation space consists of pairs of plane partitions connected by a common leg whose cross-section takes the shape of Young diagrams, offering a more transparent geometric picture than the previous construction.

hep-th↗

arXiv2020

TEAM: We Need More Powerful Adversarial Examples for DNNs

Although deep neural networks (DNNs) have achieved success in many application fields, it is still vulnerable to imperceptible adversarial examples that can lead to misclassification of DNNs easily. To overcome this challenge, many defensive methods are proposed. Indeed, a powerful adversarial example is a key benchmark to measure these defensive mechanisms. In this paper, we propose a novel method (TEAM, Taylor Expansion-Based Adversarial Methods) to generate more powerful adversarial examples than previous methods. The main idea is to craft adversarial examples by minimizing the confidence of the ground-truth class under untargeted attacks or maximizing the confidence of the target class under targeted attacks. Specifically, we define the new objective functions that approximate DNNs by using the second-order Taylor expansion within a tiny neighborhood of the input. Then the Lagrangian multiplier method is used to obtain the optimize perturbations for these objective functions. To decrease the amount of computation, we further introduce the Gauss-Newton (GN) method to speed it up. Finally, the experimental result shows that our method can reliably produce adversarial examples with 100% attack success rate (ASR) while only by smaller perturbations. In addition, the adversarial example generated with our method can defeat defensive distillation based on gradient masking.

cs.LG↗

arXiv2020

Maximal nighttime electrical power generation via optimal radiative cooling

We present a systematic optimization of nighttime thermoelectric power generation system utilizing radiative cooling. We show that an electrical power density over 2 W/m2, two orders of magnitude higher than the previously reported experimental result, is achievable using existing technologies. This system combines radiative cooling and thermoelectric power generation and operates at night when solar energy harvesting is unavailable. The thermoelectric power generator (TEG) itself covers less than 1 percent of the system footprint area when achieving this optimal power generation, showing economic feasibility. We study the influence of emissivity spectra, thermal convection, thermoelectric figure of merit and the area ratio between the TEG and the radiative cooler on the power generation performance. We optimize the thermal radiation emitter attached to the cold side and propose practical material implementation. The importance of the optimal emitter is elucidated by the gain of 153% in power density compared to regular blackbody emitters.

physics.app-ph↗

arXiv2020

Key Generation for Internet of Things: A Contemporary Survey

Key generation is a promising technique to bootstrap secure communications for the Internet of Things (IoT) devices that have no prior knowledge between each other. In the past few years, a variety of key generation protocols and systems have been proposed. In this survey, we review and categorise recent key generation systems based on a novel taxonomy. Then, we provide both quantitative and qualitative comparisons of existing approaches. We also discuss the security vulnerabilities of key generation schemes and possible countermeasures. Finally, we discuss the current challenges and point out several potential research directions.

cs.CR↗

arXiv2020

The Cluster-in-Molecule Local Correlation Method with an Accurate Distant Pair Correction for Large Systems

The cluster-in-molecule (CIM) local correlation approach with an accurate distant pair correlation energy correction is presented. For large systems, the inclusion of distant pair correlation energies is essential for the accurate predictions of absolute correlation energies and relative energies. Here we propose a simple and efficient scheme for evaluating the distant pair correlation energy correction. The corrections can be readily extracted from electron correlation calculations of clusters with almost no additional effort. Benchmark calculations show that the improved CIM approach can recover more than 99.97% of the conventional correlation energy. By combining the CIM approach with the domain based local pair natural orbital (DLPNO) local correlation approach, we have provided accurate binding energies at the CIM-DLPNO-CCSD(T) level for a test set consisting of eight weakly bound complexes ranging in size from 200 to 1027 atoms. With these results as the reference data, the accuracy and applicability of other electron correlation methods and a few density functional methods for large systems have been assessed.

physics.chem-ph↗

arXiv2020

Quantum Many-Body Simulations of the 2D Fermi-Hubbard Model in Ultracold Optical Lattices

Understanding quantum many-body states of correlated electrons is one main theme in modern condensed matter physics. Given that the Fermi-Hubbard model, the prototype of correlated electrons, has been recently realized in ultracold optical lattices, it is highly desirable to have controlled numerical methodology to provide precise finite-temperature results upon doping, to directly compare with experiments. Here, we demonstrate the exponential tensor renormalization group (XTRG) algorithm [Phys. Rev. X 8, 031082 (2018)], complemented with independent determinant quantum Monte Carlo (DQMC) offer a powerful combination of tools for this purpose. XTRG provides full and accurate access to the density matrix and thus various spin and charge correlations, down to unprecedented low temperature of few percents of the fermion tunneling energy scale. We observe excellent agreement with ultracold fermion measurements at both half-filling and finite-doping, including the sign-reversal behavior in spin correlations due to formation of magnetic polarons, and the attractive hole-doublon and repulsive hole-hole pairs that are responsible for the peculiar bunching and antibunching behavior of the antimoments.

cond-mat.str-el↗

arXiv2020

Emergent risks in the Mt. Everest region in the time of anthropogenic climate change

In April and May 2019, as a part of the National Geographic and Roxel Perpetual Planet Everest Expedition, the most interdisciplinary scientific ever was launched. This research identified changing dynamics, including emergent risks resulting from natural and anthropogenic change to the natural system. We have identified compounded risks to ecosystem and human health, geologic hazards, and changing climate conditions that impact the local community, climbers, and trekkeers in the future. This review brings together perspectives from across the biological, geological, and health sciences to better understand emergent risks on Mt. Everest and in the Khumbu region. Understanding and mitigating these risks is critical for the ~10,000 people living in the Khumbu region, as well as the thousands of visiting trekkers and the hundreds of climbers who attempt to summit each year.

physics.geo-ph↗

arXiv2020

DVI: Depth Guided Video Inpainting for Autonomous Driving

To get clear street-view and photo-realistic simulation in autonomous driving, we present an automatic video inpainting algorithm that can remove traffic agents from videos and synthesize missing regions with the guidance of depth/point cloud. By building a dense 3D map from stitched point clouds, frames within a video are geometrically correlated via this common 3D map. In order to fill a target inpainting area in a frame, it is straightforward to transform pixels from other frames into the current one with correct occlusion. Furthermore, we are able to fuse multiple videos through 3D point cloud registration, making it possible to inpaint a target video with multiple source videos. The motivation is to solve the long-time occlusion problem where an occluded area has never been visible in the entire video. To our knowledge, we are the first to fuse multiple videos for video inpainting. To verify the effectiveness of our approach, we build a large inpainting dataset in the real urban road environment with synchronized images and Lidar data including many challenge scenes, e.g., long time occlusion. The experimental results show that the proposed approach outperforms the state-of-the-art approaches for all the criteria, especially the RMSE (Root Mean Squared Error) has been reduced by about 13%.

cs.CV↗

arXiv2020

PerMO: Perceiving More at Once from a Single Image for Autonomous Driving

We present a novel approach to detect, segment, and reconstruct complete textured 3D models of vehicles from a single image for autonomous driving. Our approach combines the strengths of deep learning and the elegance of traditional techniques from part-based deformable model representation to produce high-quality 3D models in the presence of severe occlusions. We present a new part-based deformable vehicle model that is used for instance segmentation and automatically generate a dataset that contains dense correspondences between 2D images and 3D models. We also present a novel end-to-end deep neural network to predict dense 2D/3D mapping and highlight its benefits. Based on the dense mapping, we are able to compute precise 6-DoF poses and 3D reconstruction results at almost interactive rates on a commodity GPU. We have integrated these algorithms with an autonomous driving system. In practice, our method outperforms the state-of-the-art methods for all major vehicle parsing tasks: 2D instance segmentation by 4.4 points (mAP), 6-DoF pose estimation by 9.11 points, and 3D detection by 1.37. Moreover, we have released all of the source code, dataset, and the trained model on Github.

cs.CV↗

arXiv2020

Vehicle Detection of Multi-source Remote Sensing Data Using Active Fine-tuning Network

Vehicle detection in remote sensing images has attracted increasing interest in recent years. However, its detection ability is limited due to lack of well-annotated samples, especially in densely crowded scenes. Furthermore, since a list of remotely sensed data sources is available, efficient exploitation of useful information from multi-source data for better vehicle detection is challenging. To solve the above issues, a multi-source active fine-tuning vehicle detection (Ms-AFt) framework is proposed, which integrates transfer learning, segmentation, and active classification into a unified framework for auto-labeling and detection. The proposed Ms-AFt employs a fine-tuning network to firstly generate a vehicle training set from an unlabeled dataset. To cope with the diversity of vehicle categories, a multi-source based segmentation branch is then designed to construct additional candidate object sets. The separation of high quality vehicles is realized by a designed attentive classifications network. Finally, all three branches are combined to achieve vehicle detection. Extensive experimental results conducted on two open ISPRS benchmark datasets, namely the Vaihingen village and Potsdam city datasets, demonstrate the superiority and effectiveness of the proposed Ms-AFt for vehicle detection. In addition, the generalization ability of Ms-AFt in dense remote sensing scenes is further verified on stereo aerial imagery of a large camping site.

cs.CV↗

arXiv2020

The influence of the Insight-HXMT/LE time response on timing analysis

LE is the low energy telescope of Insight-HXMT. It uses swept charge devices (SCDs) to detect soft X-ray photons. The time response of LE is caused by the structure of SCDs. With theoretical analysis and Monte Carlo simulations we discuss the influence of LE time response (LTR) on the timing analysis from three aspects: the power spectral density, the pulse profile and the time lag. After the LTR, the value of power spectral density monotonously decreases with the increasing frequency. The power spectral density of a sinusoidal signal reduces by a half at frequency 536 Hz. The corresponding frequency for QPO signals is 458 Hz. The Root mean square (RMS) of QPOs holds the similar behaviour. After the LTR, the centroid frequency and full width at half maxima (FWHM) of QPOs signals do not change. The LTR reduces the RMS of pulse profiles and shifts the pulse phase. In the time domain, the LTR only reduces the peak value of the crosscorrelation function while it does not change the peak position. Thus it will not affect the result of the time lag. When considering the time lag obtained from two instruments and one among them is LE, a 1.18 ms lag is expected caused by the LTR. The time lag calculated in the frequency domain is the same as that in the time domain.

astro-ph.IM↗

arXiv2020

T-matrix evaluation of three-dimensional acoustic radiation forces on nonspherical objects in Bessel beams with arbitrary order and location

Acoustic radiation forces (ARFs) induced by a single Bessel beam with arbitrary order and location on a nonspherical shape are studied using the T-matrix method (TMM) in three dimensions. Based on the radiation stress tensor approach and the multipole expansion method for the arbitrary Bessel beam, the ARF expressions are derived in terms of the incident and scattered beam shape coefficients independently with the corresponding homemade code packages. Several numerical experiments are conducted to verify the versatility of the TMM. The axial acoustic radiation forces (ARFs) of several typical shapes are considered in the analysis with the emphasis on the axial ARF reversal and the corresponding physical mechanism. This study may guide the experimental set-up to find negative axial ARFs quickly and effectively based on the predicted parameters with TMM. Relatively elongated shapes may be helpful for pulling forces in Bessel beams. Furthermore, the lateral ARFs for both convex and concave nonspherical shapes are also investigated with different topological charges, cone angles and offsets of the particle centroid to the beam axis in a broadband frequency regime. A brief theoretical derivation of the incident beam shape coefficients for the standing Bessel beams is also given. The present work could help to design the acoustic tweezers numerical toolbox which provides an acoustical alternative to the optical tweezers toolbox.

physics.app-ph↗

arXiv2020

Automatic Differentiation for Second Renormalization of Tensor Networks

Tensor renormalization group (TRG) constitutes an important methodology for accurate simulations of strongly correlated lattice models. Facilitated by the automatic differentiation technique widely used in deep learning, we propose a uniform framework of differentiable TRG ($\partial$TRG) that can be applied to improve various TRG methods, in an automatic fashion. Essentially, $\partial$TRG systematically extends the concept of second renormalization [PRL 103, 160601 (2009)] where the tensor environment is computed recursively in the backward iteration, in the sense that given the forward process of TRG, $\partial$TRG automatically finds the gradient through backpropagation, with which one can deeply "train" the tensor networks. We benchmark $\partial$TRG in solving the square-lattice Ising model, and demonstrate its power by simulating one- and two-dimensional quantum systems at finite temperature. The deep optimization as well as GPU acceleration renders $\partial$TRG manybody simulations with high efficiency and accuracy.

cond-mat.str-el↗