arXiv subjects

Xin Lu

Publications and source records attributed to Xin Lu.

At least 181 records · Page 10Linked to original sources

arXiv2021

Fast Camera Image Denoising on Mobile GPUs with Deep Learning, Mobile AI 2021 Challenge: Report

Image denoising is one of the most critical problems in mobile photo processing. While many solutions have been proposed for this task, they are usually working with synthetic data and are too computationally expensive to run on mobile devices. To address this problem, we introduce the first Mobile AI challenge, where the target is to develop an end-to-end deep learning-based image denoising solution that can demonstrate high efficiency on smartphone GPUs. For this, the participants were provided with a novel large-scale dataset consisting of noisy-clean image pairs captured in the wild. The runtime of all models was evaluated on the Samsung Exynos 2100 chipset with a powerful Mali GPU capable of accelerating floating-point and quantized neural networks. The proposed solutions are fully compatible with any mobile GPU and are capable of processing 480p resolution images under 40-80 ms while achieving high fidelity results. A detailed description of all models developed in the challenge is provided in this paper.

eess.IV↗

arXiv2021

Surface plasmonics of Weyl semimetals

Smooth interfaces of topological systems are known to host massive surface states along with the topologically protected chiral one. We show that in Weyl semimetals these massive states, along with the chiral Fermi arc, strongly alter the form of the Fermi-arc plasmon, Most saliently, they yield further collective plasmonic modes that are absent in a conventional interfaces. The plasmon modes are completely anisotropic as a consequence of the underlying anisotropy in the surface model and expected to have a clear-cut experimental signature, e.g. in electron-energy loss spectroscopy.

cond-mat.mes-hall↗

arXiv2021

RefineMask: Towards High-Quality Instance Segmentation with Fine-Grained Features

The two-stage methods for instance segmentation, e.g. Mask R-CNN, have achieved excellent performance recently. However, the segmented masks are still very coarse due to the downsampling operations in both the feature pyramid and the instance-wise pooling process, especially for large objects. In this work, we propose a new method called RefineMask for high-quality instance segmentation of objects and scenes, which incorporates fine-grained features during the instance-wise segmenting process in a multi-stage manner. Through fusing more detailed information stage by stage, RefineMask is able to refine high-quality masks consistently. RefineMask succeeds in segmenting hard cases such as bent parts of objects that are over-smoothed by most previous methods and outputs accurate boundaries. Without bells and whistles, RefineMask yields significant gains of 2.6, 3.4, 3.8 AP over Mask R-CNN on COCO, LVIS, and Cityscapes benchmarks respectively at a small amount of additional computational cost. Furthermore, our single-model result outperforms the winner of the LVIS Challenge 2020 by 1.3 points on the LVIS test-dev set and establishes a new state-of-the-art. Code will be available at https://github.com/zhanggang001/RefineMask.

cs.CV↗

arXiv2021

Equalization Loss v2: A New Gradient Balance Approach for Long-tailed Object Detection

Recently proposed decoupled training methods emerge as a dominant paradigm for long-tailed object detection. But they require an extra fine-tuning stage, and the disjointed optimization of representation and classifier might lead to suboptimal results. However, end-to-end training methods, like equalization loss (EQL), still perform worse than decoupled training methods. In this paper, we reveal the main issue in long-tailed object detection is the imbalanced gradients between positives and negatives, and find that EQL does not solve it well. To address the problem of imbalanced gradients, we introduce a new version of equalization loss, called equalization loss v2 (EQL v2), a novel gradient guided reweighing mechanism that re-balances the training process for each category independently and equally. Extensive experiments are performed on the challenging LVIS benchmark. EQL v2 outperforms origin EQL by about 4 points overall AP with 14-18 points improvements on the rare categories. More importantly, it also surpasses decoupled training methods. Without further tuning for the Open Images dataset, EQL v2 improves EQL by 7.3 points AP, showing strong generalization ability. Codes have been released at https://github.com/tztztztztz/eqlv2

cs.CV↗

arXiv2021

Nodeless superconductivity in the kagome metal CsV$_3$Sb$_5$

The recently discovered kagome metal series $A$V$_3$Sb$_5$ ($A$=K, Rb, Cs) exhibits topologically nontrivial band structures, chiral charge order and superconductivity, presenting a unique platform for realizing exotic electronic states. The nature of the superconducting state and the corresponding pairing symmetry are key questions that demand experimental clarification. Here, using a technique based on the tunneling diode oscillator, the magnetic penetration depth $Δλ(T)$ of CsV$_3$Sb$_5$ was measured down to 0.07 K. A clear exponential behavior in $Δλ(T)$ with marked deviations from a $T$ or $T^2$ temperature dependence is observed at low temperatures, indicating a deficiency of nodal quasiparticles. Temperature dependence of the superfluid density and electronic specific heat can be described by two-gap $s$-wave superconductivity, consistent with the presence of multiple Fermi surfaces in CsV$_3$Sb$_5$. These results evidence nodeless superconductivity in CsV$_3$Sb$_5$ under ambient pressure, and constrain the allowed pairing symmetry.

cond-mat.supr-con↗

arXiv2021

A Transition-based Parser for Unscoped Episodic Logical Forms

"Episodic Logic:Unscoped Logical Form" (EL-ULF) is a semantic representation capturing predicate-argument structure as well as more challenging aspects of language within the Episodic Logic formalism. We present the first learned approach for parsing sentences into ULFs, using a growing set of annotated examples. The results provide a strong baseline for future improvement. Our method learns a sequence-to-sequence model for predicting the transition action sequence within a modified cache transition system. We evaluate the efficacy of type grammar-based constraints, a word-to-symbol lexicon, and transition system state features in this task. Our system is available at https://github.com/genelkim/ulf-transition-parser We also present the first official annotated ULF dataset at https://www.cs.rochester.edu/u/gkim21/ulf/resources/.

cs.CL↗

arXiv2021

Temperature evolution of quasiparticle dispersion dynamics in semimetallic 1T-TiTe2 via high-resolution angle-resolved photoemission spectroscopy and ultrafast optical pump-probe spectroscopy

High-resolution angle-resolved photoemission spectroscopy and ultrafast optical pump-probe spectroscopy were used to study semimetallic 1T - TiTe2 quasiparticle dispersion and dynamics. A kink and a flat band, having the same energy scale and temperature-dependent behaviors along the G-M direction, were detected. Both manifested at low temperatures but blurred as temperature increased. The kink was formed by an electron-phonon coupling. And the localized flat band might be closely related to an electron-phonon coupling. Ultrafast optical spectroscopy identified multiple distinct time scales in the 10-300 K range. Quantitative analysis of the fastest decay process evidenced a significant lifetime temperature dependence at high temperatures, while this starts to change slowly below ~ 100 K where an anomalous Hall coefficient occurred. At low temperature, a coherent A1g phonon mode with a frequency of ~ 4.36 THz was extracted. Frequency temperature dependence suggests that phonon hardening occurs as temperature falls and anharmonic effects can explain it. Frequency fluence dependence indicates that the phonons soften as fluence increases.

cond-mat.mtrl-sci↗

arXiv2020

Dirac quantum well engineering on the surface of topological insulator

We investigate a quantum well that consists of a thin topological insulator sandwiched between two trivial insulators. More specifically, we consider smooth interfaces between these different types of materials such that the interfaces host not only the chiral interface states, whose existence is dictated by the bulk-edge correspondence, but also massive Volkov-Pankratov states. We investigate possible hybridization between these interface states as a function of the width of the topological material and of the characteristic interface size. Most saliently, we find a strong qualitative difference between an extremely weak effect on the chiral interface states and a more common hybridization of the massive Volkov-Pankratov states that can be easily understood in terms of quantum tunneling in the framework of the model of a (Dirac) quantum well we introduce here.

cond-mat.mes-hall↗

arXiv2020

MimicDet: Bridging the Gap Between One-Stage and Two-Stage Object Detection

Modern object detection methods can be divided into one-stage approaches and two-stage ones. One-stage detectors are more efficient owing to straightforward architectures, but the two-stage detectors still take the lead in accuracy. Although recent work try to improve the one-stage detectors by imitating the structural design of the two-stage ones, the accuracy gap is still significant. In this paper, we propose MimicDet, a novel and efficient framework to train a one-stage detector by directly mimic the two-stage features, aiming to bridge the accuracy gap between one-stage and two-stage detectors. Unlike conventional mimic methods, MimicDet has a shared backbone for one-stage and two-stage detectors, then it branches into two heads which are well designed to have compatible features for mimicking. Thus MimicDet can be end-to-end trained without the pre-train of the teacher network. And the cost does not increase much, which makes it practical to adopt large networks as backbones. We also make several specialized designs such as dual-path mimicking and staggered feature pyramid to facilitate the mimicking process. Experiments on the challenging COCO detection benchmark demonstrate the effectiveness of MimicDet. It achieves 46.1 mAP with ResNeXt-101 backbone on the COCO test-dev set, which significantly surpasses current state-of-the-art methods.

cs.CV↗

arXiv2020

Second Harmonic Generation from a Single Plasmonic Nanorod Strongly Coupled to a WSe2 Monolayer

Monolayer transition metal dichalcogenides, coupled to metal plasmonic nanocavities, have recently emerged as new platforms for strong light-matter interactions. These systems are expected to have nonlinear optical properties that will enable them to be used as entangled photon sources, compact wave-mixing devices, and other elements for classical and quantum photonic technologies. Here we report the first experimental investigation of the nonlinear properties of these strongly coupled systems, by observing second harmonic generation from a WSe2 monolayer strongly coupled to a single gold nanorod. The pump frequency dependence of the second harmonic signal displays a pronounced splitting that can be explained by a coupled oscillator model with second-order nonlinearities. Rigorous numerical simulations utilizing a nonperturbative nonlinear hydrodynamic model of conduction electrons support this interpretation and reproduce experimental results. Our study thus lays the groundwork for understanding the nonlinear properties of strongly coupled nanoscale systems.

physics.optics↗

arXiv2020

Information Mandala: Statistical Distance Matrix with Clustering

In machine learning, observation features are measured in a metric space to obtain their distance function for optimization. Given similar features that are statistically sufficient as a population, a statistical distance between two probability distributions can be calculated for more precise learning. Provided the observed features are multi-valued, the statistical distance function is still efficient. However, due to its scalar output, it cannot be applied to represent detailed distances between feature elements. To resolve this problem, this paper extends the traditional statistical distance to a matrix form, called a statistical distance matrix. In experiments, the proposed approach performs well in object recognition tasks and clearly and intuitively represents the dissimilarities between cat and dog images in the CIFAR dataset, even when directly calculated using the image pixels. By using the hierarchical clustering of the statistical distance matrix, the image pixels can be separated into several clusters that are geometrically arranged around a center like a Mandala pattern. The statistical distance matrix with clustering, called the Information Mandala, is beyond ordinary saliency maps and can help to understand the basic principles of the convolution neural network.

stat.ML↗

arXiv2020

Optical control of valley Zeeman effect through many-exciton interactions

Charge carriers in two-dimensional transition metal dichalcogenides (TMDs), such as WSe$_2$, have their spin and valley-pseudospin locked into an optically-addressable index that is proposed as a basis for future information processing. The manipulation of this spin-valley index requires tuning its energy, typically through external magnetic field (B), which is cumbersome. Thus, other efficient routes like all-optical control of spin-valley index are desirable. Here, we show that many-body interactions amongst interlayer excitons in WSe$_2$/MoSe$_2$ heterobilayer induce a steady-state valley Zeeman splitting corresponding to B $\sim$ 6 Tesla. This anomalous splitting, present at incident powers as low as $μ$Ws, increases with power and enhances, suppresses or even flips the sign of a B-induced splitting. Moreover, the $\it{g}$-factor of valley Zeeman splitting can be tuned by $\sim$ 30 $\%$ with incident power. In addition to valleytronics, our results are relevant for achieving optical non-reciprocity using two-dimensional materials.

cond-mat.mes-hall↗

arXiv2019

Strange metal behavior in a pure ferromagnetic Kondo lattice

The strange metal phases found to develop in a wide range of materials near a quantum critical point (QCP), have posed a long-standing mystery. The frequent association of strange metals with unconventional superconductivity and antiferromagnetic QCPs has led to a belief that they are highly entangled quantum states. Ferromagnets, by contrast are regarded as an unlikely setting for strange metals, for they are weakly entangled and their QCPs are often interrupted by competing phases or first order phase transitions. Here, we provide compelling evidence that the stoichiometric heavy fermion ferromagnet CeRh$_6$Ge$_4$ becomes a strange metal at a pressure-induced QCP: specific heat and resistivity measurements demonstrate that the FM transition is continuously suppressed to zero temperature revealing a strange metal phase. We argue that strong magnetic anisotropy plays a key role in this process,injecting entanglement, in the form of triplet resonating valence bonds (tRVBs) into the ordered ferromagnet. We show that the singular transformation from tRVBs into Kondo singlets that occurs at the QCP causes a jump in the Fermi surface volume: a key driver of strange metallic behavior. Our results open up a new direction for research into FM quantum criticality, while also establishing an important new setting for the strange metal problem. Most importantly, strange metallic behavior at a FM quantum critical point suggests that it is quantum entanglement rather than the destruction of antiferromagnetism that is the common driver of the many varied examples of strange metallic behavior.

cond-mat.str-el↗

arXiv2019

Steering of high energy electron beam in laser plasma accelerators

By using Dazzler system and tilting compressor grating, we provide an effective way of using the laser group delay dispersion (GDD) to continuously steer the high energy electron beam which is accelerated by asymmetric laser-wakefield. The deviation angle of electrons is as the same as the angular chirped laser pulse from its initial optical axis, which is determined by the laser pulse-front-tilt (PFT). This unique method can be continuously used to control over the pointing direction of electron-pulses to the requisite trajectories, especially for the alignment sensitive devices such as electron-positron collider or undulator. Besides, the effect of PFT on the qualities of electron beam has been investigated.

physics.plasm-ph↗

arXiv2019

Free-Form Image Inpainting with Gated Convolution

We present a generative image inpainting system to complete images with free-form mask and guidance. The system is based on gated convolutions learned from millions of images without additional labelling efforts. The proposed gated convolution solves the issue of vanilla convolution that treats all input pixels as valid ones, generalizes partial convolution by providing a learnable dynamic feature selection mechanism for each channel at each spatial location across all layers. Moreover, as free-form masks may appear anywhere in images with any shape, global and local GANs designed for a single rectangular mask are not applicable. Thus, we also present a patch-based GAN loss, named SN-PatchGAN, by applying spectral-normalized discriminator on dense image patches. SN-PatchGAN is simple in formulation, fast and stable in training. Results on automatic image inpainting and user-guided extension demonstrate that our system generates higher-quality and more flexible results than previous methods. Our system helps user quickly remove distracting objects, modify image layouts, clear watermarks and edit faces. Code, demo and models are available at: https://github.com/JiahuiYu/generative_inpainting

cs.CV↗

arXiv2019

Dipolar interactions between field-tuneable, localized emitters in van der Waals heterostructures

While photons in free space barely interact, matter can mediate interactions between them resulting in optical nonlinearities. Such interactions at the single-quantum level result in an on-site photon repulsion, crucial for photon-based quantum information processing and for realizing strongly interacting many-body states of light. Here, we report repulsive dipole-dipole interactions between electric field tuneable, localized interlayer excitons in MoSe$_2$/WSe$_2$ heterobilayer. The presence of a single, localized exciton with an out-of-plane, non-oscillating dipole moment increases the energy of the second excitation by $\sim$ 2 meV -- an order of magnitude larger than the emission linewidth and corresponding to an inter-dipole distance of $\sim$ 5 nm. At higher excitation power, multi-exciton complexes appear at systematically higher energies. The magnetic field dependence of the emission polarization is consistent with spin-valley singlet nature of the dipolar molecular state. Our finding is an important step towards the creation of excitonic few- and many-body states such as dipolar crystals with spin-valley spinor in van der Waals (vdW) heterostructures.

cond-mat.mes-hall↗

arXiv2019

Beyond the Coverage of Information Spreading: Analytical and Empirical Evidence of Re-exposure in Large-scale Online Social Networks

Peer influence and social contagion are key denominators in the adoption and participation of information spreading, such as news propagation, word-of-mouth or viral marketing. In this study, we argue that it is biased to only focus on the scale and coverage of information spreading, and propose that the level of influence reinforcement, quantified by the re-exposure rate, i.e., the rate of individuals who are repeatedly exposed to the same information, should be considered together to measure the effectiveness of spreading. We show that local network structural characteristics significantly affects the probability of being exposed or re-exposed to the same information. After analyzing trending news on the super large-scale online network of Sina Weibo (China's Twitter) with 430 million connected users, we find a class of users with extremely low exposure rate, even they are following tens of thousands of others; and the re-exposure rate is substantially higher for news with more transmission waves and stronger secondary forwarding. While exposure and re-exposure rate typically grow together with the scale of spreading, we find exceptional cases where it is possible to achieve a high exposure rate while maintaining low re-exposure rate, or vice versa.

physics.soc-ph↗

arXiv2019

MMDetection: Open MMLab Detection Toolbox and Benchmark

We present MMDetection, an object detection toolbox that contains a rich set of object detection and instance segmentation methods as well as related components and modules. The toolbox started from a codebase of MMDet team who won the detection track of COCO Challenge 2018. It gradually evolves into a unified platform that covers many popular detection methods and contemporary modules. It not only includes training and inference codes, but also provides weights for more than 200 network models. We believe this toolbox is by far the most complete detection toolbox. In this paper, we introduce the various features of this toolbox. In addition, we also conduct a benchmarking study on different methods, components, and their hyper-parameters. We wish that the toolbox and benchmark could serve the growing research community by providing a flexible toolkit to reimplement existing methods and develop their own new detectors. Code and models are available at https://github.com/open-mmlab/mmdetection. The project is under active development and we will keep this document updated.

cs.CV↗