arXiv subjects

Weidong Zhang

Publications and source records attributed to Weidong Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2023

Mastering Asymmetrical Multiplayer Game with Multi-Agent Asymmetric-Evolution Reinforcement Learning

Asymmetrical multiplayer (AMP) game is a popular game genre which involves multiple types of agents competing or collaborating with each other in the game. It is difficult to train powerful agents that can defeat top human players in AMP games by typical self-play training method because of unbalancing characteristics in their asymmetrical environments. We propose asymmetric-evolution training (AET), a novel multi-agent reinforcement learning framework that can train multiple kinds of agents simultaneously in AMP game. We designed adaptive data adjustment (ADA) and environment randomization (ER) to optimize the AET process. We tested our method in a complex AMP game named Tom \& Jerry, and our AIs trained without using any human data can achieve a win rate of 98.5% against top human players over 65 matches. The ablation experiments indicated that the proposed modules are beneficial to the framework.

cs.AI↗

arXiv2023

LookinGood^π: Real-time Person-independent Neural Re-rendering for High-quality Human Performance Capture

We propose LookinGood^π, a novel neural re-rendering approach that is aimed to (1) improve the rendering quality of the low-quality reconstructed results from human performance capture system in real-time; (2) improve the generalization ability of the neural rendering network on unseen people. Our key idea is to utilize the rendered image of reconstructed geometry as the guidance to assist the prediction of person-specific details from few reference images, thus enhancing the re-rendered result. In light of this, we design a two-branch network. A coarse branch is designed to fix some artifacts (i.e. holes, noise) and obtain a coarse version of the rendered input, while a detail branch is designed to predict "correct" details from the warped references. The guidance of the rendered image is realized by blending features from two branches effectively in the training of the detail branch, which improves both the warping accuracy and the details' fidelity. We demonstrate that our method outperforms state-of-the-art methods at producing high-fidelity images on unseen people.

cs.CV↗

arXiv2023

Directional emission of nanoscale chiral sources modified by gap plasmons

Efficient manipulation of the emission direction of a chiral nanoscale light source is significant for information transmission and on-chip information processing. Here, we propose a scheme to control the directionality of nanoscale chiral light sources based on gap plasmons. The gap plasmon mode formed by a gold nanorod and a silver nanowire realizes the highly directional emission of chiral light sources. Based on the optical spin-locked light propagation, the hybrid structure enables the directional coupling of chiral emission to achieve a contrast ratio of 99.5%. The emission direction can be manipulated by tailoring the configuration of the structure, such as the positions, aspect ratios, and orientation of the nanorod. Besides, a great local field enhancement exists for highly enhanced emission rates within the nanogap. This chiral nanoscale light source manipulation scheme provides a way for chiral valleytronics and integrated photonics.

physics.optics↗

arXiv2022

Underwater Ranker: Learn Which Is Better and How to Be Better

In this paper, we present a ranking-based underwater image quality assessment (UIQA) method, abbreviated as URanker. The URanker is built on the efficient conv-attentional image Transformer. In terms of underwater images, we specially devise (1) the histogram prior that embeds the color distribution of an underwater image as histogram token to attend global degradation and (2) the dynamic cross-scale correspondence to model local degradation. The final prediction depends on the class tokens from different scales, which comprehensively considers multi-scale dependencies. With the margin ranking loss, our URanker can accurately rank the order of underwater images of the same scene enhanced by different underwater image enhancement (UIE) algorithms according to their visual quality. To achieve that, we also contribute a dataset, URankerSet, containing sufficient results enhanced by different UIE algorithms and the corresponding perceptual rankings, to train our URanker. Apart from the good performance of URanker, we found that a simple U-shape UIE network can obtain promising performance when it is coupled with our pre-trained URanker as additional supervision. In addition, we also propose a normalization tail that can significantly improve the performance of UIE networks. Extensive experiments demonstrate the state-of-the-art performance of our method. The key designs of our method are discussed. We will release our dataset and code.

cs.CV↗

arXiv2022

Nonlinear stability of the totally geodesic wave maps in non-isotropic manifolds

In this article we investigate a type of totally geodesic map which has its image being a geodesic in an anisotropic Riemannian manifold. We consider its nonlinear stability among the family of wave maps. We first establish the factorization property and then formulate the stability problem into a PDE system in a specially constructed chart of geodesic normal coordinates. With a generalization of the hyperboloidal foliation, we establish the global existence result associate to small initial data for this PDE system, which leads to the geometric stability.

math.AP↗

arXiv2021

SPatchGAN: A Statistical Feature Based Discriminator for Unsupervised Image-to-Image Translation

For unsupervised image-to-image translation, we propose a discriminator architecture which focuses on the statistical features instead of individual patches. The network is stabilized by distribution matching of key statistical features at multiple scales. Unlike the existing methods which impose more and more constraints on the generator, our method facilitates the shape deformation and enhances the fine details with a greatly simplified framework. We show that the proposed method outperforms the existing state-of-the-art models in various challenging applications including selfie-to-anime, male-to-female and glasses removal.

cs.CV↗

arXiv2021

Plasmon-Exciton Coupling Effect on Plasmon Damping

Plasmon decay via the surface or interface is a critical process for practical energy conversion and plasmonic catalysis. However, the relationship between plasmon damping and the coupling between the plasmon and 2D materials is still unclear. The spectral splitting due to plasmon-exciton interaction impedes the conventional single-particle method to evaluate the plasmon damping rate by the spectral linewidth directly. Here, we investigated the interaction between a single gold nanorod (GNR) and 2D materials using the single-particle spectroscopy method assisted with in situ nanomanipulation technique by comparing scattering intensity and linewidth together. Our approach allows us to indisputably identify that the plasmon-exciton coupling in the GNR-WSe2 hybrid would induce plasmon damping. We can also isolate the contribution between the charge transfer channel and resonant energy transfer channel for the plasmon decay in the GNR-graphene hybrid by comparing that with thin hBN layers as an intermediate medium to block the charge transfer. We find out that the contact layer between the GNR and 2D materials contributes most of the interfacial plasmon damping. These findings contribute to a deep understanding of interfacial excitonic effects on the plasmon and 2D materials hybrid.

physics.optics↗

arXiv2021

Controlled plasmon-enhanced fluorescence by spherical microcavity

A surrounding electromagnetic environment can engineer spontaneous emissions from quantum emitters through the Purcell effect. For instance, a plasmonic antenna can efficiently confine an electromagnetic field and enhance the fluorescent process. In this study, we demonstrate that a photonic microcavity can modulate plasmon-enhanced fluorescence by engineering the local electromagnetic environment. Consequently, we constructed a plasmon-enhanced emitter (PE-emitter), which comprised a nanorod and a nanodiamond, using the nanomanipulation technique. Furthermore, we controlled a polystyrene sphere approaching the PE-emitter and investigated in situ the associated fluorescent spectrum and lifetime. The emission of PE-emitter can be enhanced resonantly at the photonic modes as compared to that within the free spectral range. The spectral shape modulated by photonic modes is independent of the separation between the PS sphere and PE-emitter. The band integral of the fluorescence decay rate can be enhanced or suppressed after the PS sphere couples to the PE-emitters, depending on the coupling strength between the plasmonic antenna and the photonic cavity. These findings can be utilized in sensing and imaging applications.

physics.optics↗

arXiv2021

Local phase delay effect on the asymmetric spectroscopy of plasmon-exciton coupling systems

The phase delay of a local electric field, being well-known in plasmonic nanostructures, has seldom been investigated to modulate the plasmon-exciton interaction. Here, with the single-particle spectroscopy method, we experimentally investigate the phase effect in plasmon-exciton coupling systems consisting of monolayer WSe2 and an individual gold nanorod. The local plasmon phase delay is tuned by adopting various nanorods with different resonant energies respective to the exciton. We find that the local plasmon phase delay between the excitons and the plasmonic modes is as equally essential as the amplitude. The phase delay modulates the plasmon-exciton coupling considerably, resulting in an asymmetric spectral line-shape due to the interference behavior. There is an excellent agreement for the phase delay between the numerically calculated near-field phase distribution and the experimental results. The local phase delay can act as an effective way to modulate the properties of plexcitonic coupling at the nanoscale, which may have potential applications in nanoscale sensing, solar energy devices, and enhancing nonlinear processes.

physics.optics↗

arXiv2020

Tips and Tricks for Webly-Supervised Fine-Grained Recognition: Learning from the WebFG 2020 Challenge

WebFG 2020 is an international challenge hosted by Nanjing University of Science and Technology, University of Edinburgh, Nanjing University, The University of Adelaide, Waseda University, etc. This challenge mainly pays attention to the webly-supervised fine-grained recognition problem. In the literature, existing deep learning methods highly rely on large-scale and high-quality labeled training data, which poses a limitation to their practicability and scalability in real world applications. In particular, for fine-grained recognition, a visual task that requires professional knowledge for labeling, the cost of acquiring labeled training data is quite high. It causes extreme difficulties to obtain a large amount of high-quality training data. Therefore, utilizing free web data to train fine-grained recognition models has attracted increasing attentions from researchers in the fine-grained community. This challenge expects participants to develop webly-supervised fine-grained recognition methods, which leverages web images in training fine-grained recognition models to ease the extreme dependence of deep learning methods on large-scale manually labeled datasets and to enhance their practicability and scalability. In this technical report, we have pulled together the top WebFG 2020 solutions of total 54 competing teams, and discuss what methods worked best across the set of winning teams, and what surprisingly did not help.

cs.CV↗

arXiv2020

SARG: A Novel Semi Autoregressive Generator for Multi-turn Incomplete Utterance Restoration

Dialogue systems in open domain have achieved great success due to the easily obtained single-turn corpus and the development of deep learning, but the multi-turn scenario is still a challenge because of the frequent coreference and information omission. In this paper, we investigate the incomplete utterance restoration which has brought general improvement over multi-turn dialogue systems in recent studies. Meanwhile, jointly inspired by the autoregression for text generation and the sequence labeling for text editing, we propose a novel semi autoregressive generator (SARG) with the high efficiency and flexibility. Moreover, experiments on two benchmarks show that our proposed model significantly outperforms the state-of-the-art models in terms of quality and inference speed.

cs.CL↗

arXiv2020

Soft policy optimization using dual-track advantage estimator

In reinforcement learning (RL), we always expect the agent to explore as many states as possible in the initial stage of training and exploit the explored information in the subsequent stage to discover the most returnable trajectory. Based on this principle, in this paper, we soften the proximal policy optimization by introducing the entropy and dynamically setting the temperature coefficient to balance the opportunity of exploration and exploitation. While maximizing the expected reward, the agent will also seek other trajectories to avoid the local optimal policy. Nevertheless, the increase of randomness induced by entropy will reduce the train speed in the early stage. Integrating the temporal-difference (TD) method and the general advantage estimator (GAE), we propose the dual-track advantage estimator (DTAE) to accelerate the convergence of value functions and further enhance the performance of the algorithm. Compared with other on-policy RL algorithms on the Mujoco environment, the proposed method not only significantly speeds up the training but also achieves the most advanced results in cumulative return.

cs.LG↗

arXiv2020

GeoLayout: Geometry Driven Room Layout Estimation Based on Depth Maps of Planes

The task of room layout estimation is to locate the wall-floor, wall-ceiling, and wall-wall boundaries. Most recent methods solve this problem based on edge/keypoint detection or semantic segmentation. However, these approaches have shown limited attention on the geometry of the dominant planes and the intersection between them, which has significant impact on room layout. In this work, we propose to incorporate geometric reasoning to deep learning for layout estimation. Our approach learns to infer the depth maps of the dominant planes in the scene by predicting the pixel-level surface parameters, and the layout can be generated by the intersection of the depth maps. Moreover, we present a new dataset with pixel-level depth annotation of dominant planes. It is larger than the existing datasets and contains both cuboid and non-cuboid rooms. Experimental results show that our approach produces considerable performance gains on both 2D and 3D datasets.

cs.CV↗

arXiv2020

Smooth, Time-invariant Regulation of Nonholonomic Systems via Energy Pumping-and-Damping

In this paper we propose an energy pumping-and-damping technique to regulate nonholonomic systems described by kinematic models. The controller design follows the widely popular interconnection and damping assignment passivity-based methodology, with the free matrices partially structured. Two asymptotic regulation objectives are considered: drive to zero the state or drive the systems total energy to a desired constant value. In both cases, the control laws are smooth, time-invariant, state-feedbacks. For the nonholonomic integrator we give an almost global solution for both problems, with the objectives ensured for all system initial conditions starting outside a set that has zero Lebesgue measure and is nowhere dense. For the general case of higher-order nonholonomic systems in chained form, a local stability result is given. Simulation results comparing the performance of the proposed controller with other existing designs are also provided.

eess.SY↗

arXiv2020

Comprehensive Investigation and Isolation have Effectively Suppressed the Spread of COVID-19

The outbreak of COVID-19 since Dec. 2019 has caused severe life and economic damage worldwide, many countries are trapped by medical resource constraints or absence of targeted therapeutics, and therefore the implement of systemic policies to block this pandemic should be prioritized. Based on the transmission mechanisms and physicochemical properties of betacoronaviruses, we construct a fine-grained transmission dynamics model (ICRD) to forecast the crucial information of public concern, therein using dynamical coefficients to quantify the impact of the implement time and intensity of containment policies on the spread of epidemic. We find that the comprehensive investigation policy for susceptible population and the quarantine for suspected cases eminently contribute to reduce casualties during the phase of the dramatic increase of diagnosed cases. Statistic evidences strongly suggest that society should take such forceful public health interventions to cut the infection channels in the initial stage until the pandemic is interrupted.

q-bio.PE↗

arXiv2020

A New Signal Injection-based Method for Estimation of Position in Interior Permanent Magnet Synchronous Motors

Several heuristic procedures to estimate the rotor position of permanent magnet synchronous motors (PMSM) via signal injection have been reported in the literature. Using averaging theory, a framework to analyse such schemes has been recently proposed. However, to the best of our knowledge, no theoretical analysis of the performance of the conventional linear time invariant filtering methods, which are widely used as standard industrial practice, has been reported in the literature. The objective of this note is to propose a new method that, on one hand, is amenable to a rigorous theoretical analysis and, on the other hand, ensures an improved accuracy in the position estimation. An additional advantage of the new method is that it relies on the use of linear operators, implementable with simple computations. The effectiveness of the proposed scheme is assessed by experiments on an interior PMSM platform driven by a 521 V DC bus with 5-kHz PWM.

eess.SY↗

arXiv2019

The Expressivity and Training of Deep Neural Networks: toward the Edge of Chaos?

Expressivity is one of the most significant issues in assessing neural networks. In this paper, we provide a quantitative analysis of the expressivity for the deep neural network (DNN) from its dynamic model, where the Hilbert space is employed to analyze the convergence and criticality. We study the feature mapping of several widely used activation functions obtained by Hermite polynomials, and find sharp declines or even saddle points in the feature space, which stagnate the information transfer in DNNs. We then present a new activation function design based on the Hermite polynomials for better utilization of spatial representation. Moreover, we analyze the information transfer of DNNs, emphasizing the convergence problem caused by the mismatch between input and topological structure. We also study the effects of input perturbations and regularization operators on critical expressivity. Our theoretical analysis reveals that DNNs use spatial domains for information representation and evolve to the edge of chaos as depth increases. In actual training, whether a particular network can ultimately arrive the edge of chaos depends on its ability to overcome convergence and pass information to the required network depth. Finally, we demonstrate the empirical performance of the proposed hypothesis via multivariate time series prediction and image classification examples.

cs.LG↗

arXiv2019

On Generation of Virtual Outputs via Signal Injection: Application to Observer Design for Electromechanical Systems

Probing signal injection is a well-established technique to extract additional information from a weakly (or non) observable dynamical system. Using averaging theory, a framework to analyse such schemes for general nonlinear systems has been recently proposed in [Combes et. al., 2016], where it is shown that the signal injection may be used to generate a new high frequency component of the systems output that can be used for state observation or controller design. A key step for the success of this technique is the implementation of a filter to reconstruct this virtual output from the measurement of the overall systems output. The main contribution of this paper is to propose a new filter with guaranteed convergence properties that outperforms the classical designs. The method is applied to a general class of electromechanical systems, and its performance is assessed via simulations and experiments on the benchmark example of a 1-dof magnetic levitation system.

eess.SY↗