arXiv subjects

Wen Liu

Publications and source records attributed to Wen Liu.

At least 55 records · Page 3Linked to original sources

arXiv2023

Michelangelo: Conditional 3D Shape Generation based on Shape-Image-Text Aligned Latent Representation

We present a novel alignment-before-generation approach to tackle the challenging task of generating general 3D shapes based on 2D images or texts. Directly learning a conditional generative model from images or texts to 3D shapes is prone to producing inconsistent results with the conditions because 3D shapes have an additional dimension whose distribution significantly differs from that of 2D images and texts. To bridge the domain gap among the three modalities and facilitate multi-modal-conditioned 3D shape generation, we explore representing 3D shapes in a shape-image-text-aligned space. Our framework comprises two models: a Shape-Image-Text-Aligned Variational Auto-Encoder (SITA-VAE) and a conditional Aligned Shape Latent Diffusion Model (ASLDM). The former model encodes the 3D shapes into the shape latent space aligned to the image and text and reconstructs the fine-grained 3D neural fields corresponding to given shape embeddings via the transformer-based decoder. The latter model learns a probabilistic mapping function from the image or text space to the latent shape space. Our extensive experiments demonstrate that our proposed approach can generate higher-quality and more diverse 3D shapes that better semantically conform to the visual or textural conditional inputs, validating the effectiveness of the shape-image-text-aligned space for cross-modality 3D shape generation.

cs.CV↗

arXiv2023

Executing your Commands via Motion Diffusion in Latent Space

We study a challenging task, conditional human motion generation, which produces plausible human motion sequences according to various conditional inputs, such as action classes or textual descriptors. Since human motions are highly diverse and have a property of quite different distribution from conditional modalities, such as textual descriptors in natural languages, it is hard to learn a probabilistic mapping from the desired conditional modality to the human motion sequences. Besides, the raw motion data from the motion capture system might be redundant in sequences and contain noises; directly modeling the joint distribution over the raw motion sequences and conditional modalities would need a heavy computational overhead and might result in artifacts introduced by the captured noises. To learn a better representation of the various human motion sequences, we first design a powerful Variational AutoEncoder (VAE) and arrive at a representative and low-dimensional latent code for a human motion sequence. Then, instead of using a diffusion model to establish the connections between the raw motion sequences and the conditional inputs, we perform a diffusion process on the motion latent space. Our proposed Motion Latent-based Diffusion model (MLD) could produce vivid motion sequences conforming to the given conditional inputs and substantially reduce the computational overhead in both the training and inference stages. Extensive experiments on various human motion generation tasks demonstrate that our MLD achieves significant improvements over the state-of-the-art methods among extensive human motion generation tasks, with two orders of magnitude faster than previous diffusion models on raw motion sequences.

cs.CV↗

arXiv2023

The Radial Distribution of Ion-scale Waves in the Inner Heliosphere

Determining the mechanism responsible for the plasma heating and particle acceleration is a fundamental problem in the study of the heliosphere. Due to efficient wave-particle interactions of ion-scale waves with charged particles, these waves are widely believed to be a major contributor to ion energization, and their contribution considerably depends on the wave occurrence rate. By analyzing the radial distribution of quasi-monochromatic ion-scale waves observed by the Parker Solar Probe, this work shows that the wave occurrence rate is significantly enhanced in the near-Sun solar wind, specifically 21%$-$29% below 0.3 au, in comparison to 6%$-$14% beyond 0.3 au. The radial decrease of the wave occurrence rate is not only induced by the sampling effect of a single spacecraft detection, but also by the physics relating to the wave excitation, such as the enhanced ion beam instability in the near-Sun solar wind. This work also shows that the wave normal angle $θ$, the absolute value of ellipticity $ε$, the wave frequency $f$ normalized by the proton cyclotron frequency $f_{\mathrm{cp}}$, and the wave amplitude $δB$ normalized by the local background magnetic field $B_0$ slightly vary with the radial distance. The median values of $θ$, $|ε|$, $f$, and $δB$ are about $9^\circ$, $0.73$, $3f_{\mathrm{cp}}$, and $0.01B_0$, respectively. Furthermore, this study proposes that the wave mode nature of the observed left-handed and right-handed polarized waves corresponds to the Alfvén ion cyclotron mode wave and the fast-magnetosonic whistler mode wave, respectively.

astro-ph.SR↗

arXiv2023

A New Paradigm for Device-free Indoor Localization: Deep Learning with Error Vector Spectrum in Wi-Fi Systems

The demand for device-free indoor localization using commercial Wi-Fi devices has rapidly increased in various fields due to its convenience and versatile applications. However, random frequency offset (RFO) in wireless channels poses challenges to the accuracy of indoor localization when using fluctuating channel state information (CSI). To mitigate the RFO problem, an error vector spectrum (EVS) is conceived thanks to its higher resolution of signal and robustness to RFO. To address these challenges, this paper proposed a novel error vector assisted learning (EVAL) for device-free indoor localization. The proposed EVAL scheme employs deep neural networks to classify the location of a person in the indoor environment by extracting ample channel features from the physical layer signals. We conducted realistic experiments based on OpenWiFi project to extract both EVS and CSI to examine the performance of different device-free localization techniques. Experimental results show that our proposed EVAL scheme outperforms conventional machine learning methods and benchmarks utilizing either CSI amplitude or phase information. Compared to most existing CSI-based localization schemes, a new paradigm with higher positioning accuracy by adopting EVS is revealed by our proposed EVAL system.

eess.SP↗

arXiv2023

Magnetic detection under high pressures using designed silicon vacancy centers in silicon carbide

Pressure-induced magnetic phase transition is attracting interest due to its ability to detect superconducting behaviour at high pressures in diamond anvil cells. However, detection of the local sample magnetic properties is a great challenge due to the small sample chamber volume. Recently, optically detected magnetic resonance (ODMR) of nitrogen vacancy (NV) centers in diamond have been used for in-situ pressure-induced phase transition detection. However, owing to their four orientation axes and temperature-dependent zero-field-splitting, interpreting the observed ODMR spectra of NV centers remain challenging. Here, we study the optical and spin properties of implanted silicon vacancy defects in 4H-SiC, which is single-axis and temperature-independent zero-field-splitting. Using this technique, we observe the magnetic phase transition of Nd2Fe14B at about 7 GPa and map the critical temperature-pressure phase diagram of the superconductor YBa2Cu3O6.6. These results highlight the potential of silicon vacancy-based quantum sensors for in-situ magnetic detection at high pressures.

physics.app-ph↗

arXiv2022

Coordinates Are NOT Lonely -- Codebook Prior Helps Implicit Neural 3D Representations

Implicit neural 3D representation has achieved impressive results in surface or scene reconstruction and novel view synthesis, which typically uses the coordinate-based multi-layer perceptrons (MLPs) to learn a continuous scene representation. However, existing approaches, such as Neural Radiance Field (NeRF) and its variants, usually require dense input views (i.e. 50-150) to obtain decent results. To relive the over-dependence on massive calibrated images and enrich the coordinate-based feature representation, we explore injecting the prior information into the coordinate-based network and introduce a novel coordinate-based model, CoCo-INR, for implicit neural 3D representation. The cores of our method are two attention modules: codebook attention and coordinate attention. The former extracts the useful prototypes containing rich geometry and appearance information from the prior codebook, and the latter propagates such prior information into each coordinate and enriches its feature representation for a scene or object surface. With the help of the prior information, our method can render 3D views with more photo-realistic appearance and geometries than the current methods using fewer calibrated images available. Experiments on various scene reconstruction datasets, including DTU and BlendedMVS, and the full 3D head reconstruction dataset, H3DS, demonstrate the robustness under fewer input views and fine detail-preserving capability of our proposed method.

cs.CV↗

arXiv2022

Multi-state Swap Test Algorithm

Estimating the overlap between two states is an important task with several applications in quantum information. However, the typical swap test circuit can only measure a sole pair of quantum states at a time. In this study we designed a recursive quantum circuit to measure overlaps of multiple quantum states $|ϕ_1...ϕ_n\rangle$ concurrently with $O(n\log n)$ controlled-swap (CSWAP) gates and $O(\log n)$ ancillary qubits. This circuit enables us to get all pairwise overlaps among input quantum states $|\langleϕ_i|ϕ_j\rangle|^2$. Compared with existing schemes for measuring the overlap of multiple quantum states, our scheme provides higher precision and less consumption of ancillary qubits. In addition, we performed simulation experiments on IBM quantum cloud platform to verify the superiority of the scheme.

quant-ph↗

arXiv2022

Study the contributions of S, P, and D-wave resonances to the quasi-two-body decays $B^{0}_{s} \rightarrow ψ(3686,3770)Kπ$ in perturbative QCD approach

Based on the perturbative quantum chromodynamics (pQCD) approach and the quasi-two-body approximation, we have studied the three-body decays B^{0}_{s}\rightarrow ψ(3686,3770)\emph{K}π, which include the contributions of the intermediate resonances \bar{\emph{K}}^{*}_{0}(1430)^{0}, \bar{\emph{K}}^{*}(892)^{0}, \bar{\emph{K}}^{*}(1410)^{0}, \bar{\emph{K}}^{*}(1680)^{0}, and \bar{\emph{K}}^{*}_{2}(1430)^{0}. The time-like form factors corresponding to the distribution amplitudes of the S, P, and D-wave of the kaon-pion pair have been adopted in the parameterized form, which describe the interactions between \emph{K} and πin the resonance region. First, the decays B^{0}_{s}\rightarrow ψ(2S,1D)\emph{K}^{-}π^{+} have been calculated followed by the calculation of the branching ratios of the decays B^{0}_{s}\rightarrow ψ(3686,3770)\emph{K}^{-}π^{+} using the 2S-1D mixing scheme. In addition, the pQCD predictions for the decays B^{0}_{s}\rightarrow ψ(2S,1D)\emph{K}πand B^{0}_{s}\rightarrow ψ(3686,3770)\emph{K}πhave been obtained using the narrow-width approximation relation given by the Clebsch-Gorden coefficients. Our work shows that the \bar{\emph{K}}^{*}(892)^{0} resonance is the main contributor to the total decay, and the branching ratio and the longitudinal polarization fraction of the ψ(2S)\bar{\emph{K}}^{*}(892)^{0} decay mode agree well with the currently available data within errors. Furthermore, the theoretical predictions of the ψ(2S) and ψ(3686) decay modes are very close, indicating that they can be regarded as the same meson state. Finally, the pQCD predictions for branching ratios of decays B^{0}_{s}\rightarrow ψ(3686,3770)\emph{K}πare of the order of 10^{-5} and 10^{-6}, respectively, which can be verified using the ongoing LHCb and Belle II experiments.

hep-ph↗

arXiv2022

Learning-based multiplexed transmission of scattered twisted light through a kilometer-scale standard multimode fiber

Multiplexing multiple orbital angular momentum (OAM) modes of light has the potential to increase data capacity in optical communication. However, the distribution of such modes over long distances remains challenging. Free-space transmission is strongly influenced by atmospheric turbulence and light scattering, while the wave distortion induced by the mode dispersion in fibers disables OAM demultiplexing in fiber-optic communications. Here, a deep-learning-based approach is developed to recover the data from scattered OAM channels without measuring any phase information. Over a 1-km-long standard multimode fiber, the method is able to identify different OAM modes with an accuracy of more than 99.9% in parallel demultiplexing of 24 scattered OAM channels. To demonstrate the transmission quality, color images are encoded in multiplexed twisted light and our method achieves decoding the transmitted data with an error rate of 0.13%. Our work shows the artificial intelligence algorithm could benefit the use of OAM multiplexing in commercial fiber networks and high-performance optical communication in turbulent environments.

physics.optics↗

arXiv2021

Speech Drives Templates: Co-Speech Gesture Synthesis with Learned Templates

Co-speech gesture generation is to synthesize a gesture sequence that not only looks real but also matches with the input speech audio. Our method generates the movements of a complete upper body, including arms, hands, and the head. Although recent data-driven methods achieve great success, challenges still exist, such as limited variety, poor fidelity, and lack of objective metrics. Motivated by the fact that the speech cannot fully determine the gesture, we design a method that learns a set of gesture template vectors to model the latent conditions, which relieve the ambiguity. For our method, the template vector determines the general appearance of a generated gesture sequence, while the speech audio drives subtle movements of the body, both indispensable for synthesizing a realistic gesture sequence. Due to the intractability of an objective metric for gesture-speech synchronization, we adopt the lip-sync error as a proxy metric to tune and evaluate the synchronization ability of our model. Extensive experiments show the superiority of our method in both objective and subjective evaluations on fidelity and synchronization.

cs.CV↗

arXiv2021

Construction of Attenuation Relationship of Peak Ground Velocity Using Machine Learning and Examination of Station Correction Factor

This study tries to develop new attenuation relationships of peak ground velocity using machine learning methods; random forest and neural network. In order to compare with the predictors obtained by machine learning, we have also constructed a new attenuation relationship of peak ground velocity using three-stage regression procedure proposed by Molas and Yamazaki (1995). In this study, 6,944 ground motion records at 1,184 seismic observation stations which were observed during the 32 earthquakes are employed to construct the attenuation relationships. Ground motion records from the 4 recent earthquakes are used as the test dataset. The test results show that when the shortest distance from the fault is small, the predictions by machine learning techniques are more accurate than the traditional equation. However, there is still a problem of overestimation in the predictors of machine learning, even if weights are added to the training dataset. In addition, the station correction factors based on machine learning were derived and proved to be correlated with the average shear wave velocity.

physics.geo-ph↗

arXiv2021

Electromagnetic Proton Beam Instabilities in the Inner Heliosphere: Energy Transfer Rate, Radial Distribution, and Effective Excitation

Differential flows among different ion species are often observed in the solar wind, and such ion differential flows can provide the free energy to drive Alfvén/ion-cyclotron and fast-magnetosonic/whistler instabilities. Previous works mainly focused on ion beam instability under the parameters representative of the solar wind nearby 1 au. In this paper we further study proton beam instability using the radial models of the magnetic field and plasma parameters in the inner heliosphere. We explore a comprehensive distribution of proton beam instability as functions of the heliocentric distance and the beam speed. We also perform a detailed analysis of the energy transfer between unstable waves and particles and quantify how much the free energy of the proton beam flows into unstable waves and other kinds of particle species (i.e., proton core, alpha particle, and electron). This work clarifies that both parallel and perpendicular electric fields are responsible for the excitation of oblique Alfvén/ion-cyclotron and oblique fast-magnetosonic/whistler instabilities. Moreover, this work proposes an effective growth length to estimate whether the instability is efficiently excited or not. It shows that oblique Alfvén/ion-cyclotron instability, oblique fast-magnetosonic/whistler instability, and oblique Alfvén/ion-beam instability can be efficiently driven by proton beams drifting at the speed $\sim$ 600-1300 km s$^{-1}$ in the solar atmosphere. In particular, oblique Alfvén/ion-cyclotron waves driven in the solar atmosphere can be significantly damped therein, leading to the solar corona heating. These results are helpful for understanding proton beam dynamics in the inner heliosphere and can be verified through in situ satellite measurements.

astro-ph.SR↗

arXiv2021

Planar Integrated Magneto Optical Trap

Abstract The magneto-optical trap (MOT) is an essential tool for collecting and preparing cold atoms with a wide range of applications. We demonstrate a planar-integrated MOT by combining an optical grating chip with a magnetic coil chip. The flat grating chip simplifies the conventional six-beam configuration down to a single laser beam; the flat coil chip replaces the conventional anti-Helmholtz coils of a cylindrical geometry. We trap 10^{4} cold ^{87}\text{Rb} atoms in the planar-integrated MOT, at a point 3-9 mm above the chip surface. This novel configuration effectively reduces the volume, weight, and complexity of the MOT, bringing benefits to applications including gravimeter, clock and quantum memory devices.

physics.atom-ph↗

arXiv2021

BO 2.0: Plasma Wave and Instability Analysis with Enhanced Polarization Calculations

Besides the relation between the wave vector $\bm k$ and the complex frequency $ω$, wave polarization is useful for characterizing the properties of a plasma wave. The polarization of the electromagnetic fields, $δ\bm E$ and $δ\bm B$, have been widely used in plasma physics research. Here, we derive equations for the density and velocity perturbations, $δn_s$ and $δ{\bm v}_s$, respectively, of each species in the electromagnetic kinetic plasma dispersion relation by using their relation to the species current density perturbation $δ{\bm J}_s$. Then we compare results with those of another commonly used plasma dispersion code (WHAMP) and with those of a multi-fluid plasma dispersion relation. We also summarize a number of useful polarization quantities, such as magnetic ellipticity, orientation of the major axis of the magnetic ellipse, various ratios of field energies and kinetic energies, species compressibility, parallel phase ratio, Alfvén-ratio, etc., which are useful for plasma physics research, especially for space plasma studies. This work represents an extension of the BO electromagnetic dispersion code [H.S. Xie, Comput. Phys. Comm. 244 (2019) 343-371] to enhance its calculation of polarization and to include the capability of solving the electromagnetic magnetized multi-fluid plasma dispersion relation.

physics.plasm-ph↗

arXiv2020

Liquid Warping GAN with Attention: A Unified Framework for Human Image Synthesis

We tackle human image synthesis, including human motion imitation, appearance transfer, and novel view synthesis, within a unified framework. It means that the model, once being trained, can be used to handle all these tasks. The existing task-specific methods mainly use 2D keypoints to estimate the human body structure. However, they only express the position information with no abilities to characterize the personalized shape of the person and model the limb rotations. In this paper, we propose to use a 3D body mesh recovery module to disentangle the pose and shape. It can not only model the joint location and rotation but also characterize the personalized body shape. To preserve the source information, such as texture, style, color, and face identity, we propose an Attentional Liquid Warping GAN with Attentional Liquid Warping Block (AttLWB) that propagates the source information in both image and feature spaces to the synthesized reference. Specifically, the source features are extracted by a denoising convolutional auto-encoder for characterizing the source identity well. Furthermore, our proposed method can support a more flexible warping from multiple sources. To further improve the generalization ability of the unseen source images, a one/few-shot adversarial learning is applied. In detail, it firstly trains a model in an extensive training set. Then, it finetunes the model by one/few-shot unseen image(s) in a self-supervised way to generate high-resolution (512 x 512 and 1024 x 1024) results. Also, we build a new dataset, namely iPER dataset, for the evaluation of human motion imitation, appearance transfer, and novel view synthesis. Extensive experiments demonstrate the effectiveness of our methods in terms of preserving face identity, shape consistency, and clothes details. All codes and dataset are available on https://impersonator.org/work/impersonator-plus-plus.html.

cs.CV↗

arXiv2020

Electron Temperature Anisotropy and Electron Beam Constraints From Electron Kinetic Instabilities in the Solar Wind

Electron temperature anisotropies and electron beams are nonthermal features of the observed nonequilibrium electron velocity distributions in the solar wind. In collision-poor plasmas these nonequilibrium distributions are expected to be regulated by kinetic instabilities through wave-particle interactions. This study considers electron instabilities driven by the interplay of core electron temperature anisotropies and the electron beam, and firstly gives a comprehensive analysis of instabilities in arbitrary directions to the background magnetic field. It clarifies the dominant parameter regime (e.g., parallel core electron plasma beta $β_{\mathrm{ec\parallel}}$, core electron temperature anisotropy $A_{\mathrm{ec}}\equiv T_{\mathrm{ec\perp}}/T_{\mathrm{ec\parallel}}$, and electron beam velocity $V_{\mathrm{eb}}$) for each kind of electron instability (e.g., the electron beam-driven electron acoustic/magnetoacoustic instability, the electron beam-driven whistler instability, the electromagnetic electron cyclotron instability, the electron mirror instability, the electron firehose instability, and the ordinary-mode instability). It finds that the electron beam can destabilize electron acoustic/magnetoacoustic waves in the low-$β_{\mathrm{ec\parallel}}$ regime, and whistler waves in the medium- and large-$β_{\mathrm{ec\parallel}}$ regime. It also finds that a new oblique fast-magnetosonic/whistler instability is driven by the electron beam with $V_{\mathrm{eb}}\gtrsim7V_{\mathrm{A}}$ in a regime where $β_{\mathrm{ec\parallel}}\sim0.1-2$ and $A_{\mathrm{ec}}<1$. Moreover, this study presents electromagnetic responses of each kind of electron instability. These results provide a comprehensive overview for electron instability constraints on core electron temperature anisotropies and electron beams in the solar wind.

physics.space-ph↗

arXiv2020

Learning from Multimodal and Multitemporal Earth Observation Data for Building Damage Mapping

Earth observation technologies, such as optical imaging and synthetic aperture radar (SAR), provide excellent means to monitor ever-growing urban environments continuously. Notably, in the case of large-scale disasters (e.g., tsunamis and earthquakes), in which a response is highly time-critical, images from both data modalities can complement each other to accurately convey the full damage condition in the disaster's aftermath. However, due to several factors, such as weather and satellite coverage, it is often uncertain which data modality will be the first available for rapid disaster response efforts. Hence, novel methodologies that can utilize all accessible EO datasets are essential for disaster management. In this study, we have developed a global multisensor and multitemporal dataset for building damage mapping. We included building damage characteristics from three disaster types, namely, earthquakes, tsunamis, and typhoons, and considered three building damage categories. The global dataset contains high-resolution optical imagery and high-to-moderate-resolution multiband SAR data acquired before and after each disaster. Using this comprehensive dataset, we analyzed five data modality scenarios for damage mapping: single-mode (optical and SAR datasets), cross-modal (pre-disaster optical and post-disaster SAR datasets), and mode fusion scenarios. We defined a damage mapping framework for the semantic segmentation of damaged buildings based on a deep convolutional neural network algorithm. We compare our approach to another state-of-the-art baseline model for damage mapping. The results indicated that our dataset, together with a deep learning network, enabled acceptable predictions for all the data modality scenarios.

cs.CV↗

arXiv2020

Encoding Structure-Texture Relation with P-Net for Anomaly Detection in Retinal Images

Anomaly detection in retinal image refers to the identification of abnormality caused by various retinal diseases/lesions, by only leveraging normal images in training phase. Normal images from healthy subjects often have regular structures (e.g., the structured blood vessels in the fundus image, or structured anatomy in optical coherence tomography image). On the contrary, the diseases and lesions often destroy these structures. Motivated by this, we propose to leverage the relation between the image texture and structure to design a deep neural network for anomaly detection. Specifically, we first extract the structure of the retinal images, then we combine both the structure features and the last layer features extracted from original health image to reconstruct the original input healthy image. The image feature provides the texture information and guarantees the uniqueness of the image recovered from the structure. In the end, we further utilize the reconstructed image to extract the structure and measure the difference between structure extracted from original and the reconstructed image. On the one hand, minimizing the reconstruction difference behaves like a regularizer to guarantee that the image is corrected reconstructed. On the other hand, such structure difference can also be used as a metric for normality measurement. The whole network is termed as P-Net because it has a ``P'' shape. Extensive experiments on RESC dataset and iSee dataset validate the effectiveness of our approach for anomaly detection in retinal images. Further, our method also generalizes well to novel class discovery in retinal images and anomaly detection in real-world images.

eess.IV↗