arXiv subjects

Yong Zhang

Publications and source records attributed to Yong Zhang.

At least 217 records · Page 12Linked to original sources

arXiv2025

Twist-induced near-field radiative thermal regulator assisted by cylindrical surface modes

Near-field radiative heat transfer (RHT) can surpass Planck's blackbody limit by several orders of magnitude due to the tunneling effect of thermal photons. The ability to understand and regulate RHT is of great significance in contactless energy transfer. In this work, we construct a rotating system with a hexagonal boron nitride (h-BN) cylinder for actively regulating the RHT between two nanoparticles (NPs). The results show that when the two NPs are located directly above the cylinder, energy can be directionally transmitted along the cylindrical channel in the form of low-loss surface waves, which can significantly enhance RHT. In addition, we find that the RHT can be regulated by actively manipulating the excitation of cylindrical surface modes. When the rotation point is located in the middle of the line connecting the two NPs, the modulation contrast approaches five orders of magnitude, higher than that of cylinders made of other materials under the same conditions. When its diameter is slightly less than the distance between NPs, the h-BN cylinder shows excellent tunability in the heat exchange. The present work may offer a theoretical possibility for actively regulating and controlling near-field RHT between arbitrary objects based on cylindrical waveguides.

physics.optics↗

arXiv2024

VideoMaker: Zero-shot Customized Video Generation with the Inherent Force of Video Diffusion Models

Zero-shot customized video generation has gained significant attention due to its substantial application potential. Existing methods rely on additional models to extract and inject reference subject features, assuming that the Video Diffusion Model (VDM) alone is insufficient for zero-shot customized video generation. However, these methods often struggle to maintain consistent subject appearance due to suboptimal feature extraction and injection techniques. In this paper, we reveal that VDM inherently possesses the force to extract and inject subject features. Departing from previous heuristic approaches, we introduce a novel framework that leverages VDM's inherent force to enable high-quality zero-shot customized video generation. Specifically, for feature extraction, we directly input reference images into VDM and use its intrinsic feature extraction process, which not only provides fine-grained features but also significantly aligns with VDM's pre-trained knowledge. For feature injection, we devise an innovative bidirectional interaction between subject features and generated content through spatial self-attention within VDM, ensuring that VDM has better subject fidelity while maintaining the diversity of the generated video. Experiments on both customized human and object video generation validate the effectiveness of our framework.

cs.CV↗

arXiv2024

Deep Learning for Detecting and Early Predicting Chronic Obstructive Pulmonary Disease from Spirogram Time Series

Chronic Obstructive Pulmonary Disease (COPD) is a chronic lung condition characterized by airflow obstruction. Current diagnostic methods primarily rely on identifying prominent features in spirometry (Volume-Flow time series) to detect COPD, but they are not adept at predicting future COPD risk based on subtle data patterns. In this study, we introduce a novel deep learning-based approach, DeepSpiro, aimed at the early prediction of future COPD risk. DeepSpiro consists of four key components: SpiroSmoother for stabilizing the Volume-Flow curve, SpiroEncoder for capturing volume variability-pattern through key patches of varying lengths, SpiroExplainer for integrating heterogeneous data and explaining predictions through volume attention, and SpiroPredictor for predicting the disease risk of undiagnosed high-risk patients based on key patch concavity, with prediction horizons of 1, 2, 3, 4, 5 years, or even longer. Evaluated on the UK Biobank dataset, DeepSpiro achieved an AUC of 0.8328 for COPD detection and demonstrated strong predictive performance for future COPD risk (p-value < 0.001). In summary, DeepSpiro can effectively predicts the long-term progression of the COPD disease.

cs.LG↗

arXiv2024

CustomCrafter: Customized Video Generation with Preserving Motion and Concept Composition Abilities

Customized video generation aims to generate high-quality videos guided by text prompts and subject's reference images. However, since it is only trained on static images, the fine-tuning process of subject learning disrupts abilities of video diffusion models (VDMs) to combine concepts and generate motions. To restore these abilities, some methods use additional video similar to the prompt to fine-tune or guide the model. This requires frequent changes of guiding videos and even re-tuning of the model when generating different motions, which is very inconvenient for users. In this paper, we propose CustomCrafter, a novel framework that preserves the model's motion generation and conceptual combination abilities without additional video and fine-tuning to recovery. For preserving conceptual combination ability, we design a plug-and-play module to update few parameters in VDMs, enhancing the model's ability to capture the appearance details and the ability of concept combinations for new subjects. For motion generation, we observed that VDMs tend to restore the motion of video in the early stage of denoising, while focusing on the recovery of subject details in the later stage. Therefore, we propose Dynamic Weighted Video Sampling Strategy. Using the pluggability of our subject learning modules, we reduce the impact of this module on motion generation in the early stage of denoising, preserving the ability to generate motion of VDMs. In the later stage of denoising, we restore this module to repair the appearance details of the specified subject, thereby ensuring the fidelity of the subject's appearance. Experimental results show that our method has a significant improvement compared to previous methods. Code is available at https://github.com/WuTao-CS/CustomCrafter

cs.CV↗

arXiv2024

CustomTTT: Motion and Appearance Customized Video Generation via Test-Time Training

Benefiting from large-scale pre-training of text-video pairs, current text-to-video (T2V) diffusion models can generate high-quality videos from the text description. Besides, given some reference images or videos, the parameter-efficient fine-tuning method, i.e. LoRA, can generate high-quality customized concepts, e.g., the specific subject or the motions from a reference video. However, combining the trained multiple concepts from different references into a single network shows obvious artifacts. To this end, we propose CustomTTT, where we can joint custom the appearance and the motion of the given video easily. In detail, we first analyze the prompt influence in the current video diffusion model and find the LoRAs are only needed for the specific layers for appearance and motion customization. Besides, since each LoRA is trained individually, we propose a novel test-time training technique to update parameters after combination utilizing the trained customized models. We conduct detailed experiments to verify the effectiveness of the proposed methods. Our method outperforms several state-of-the-art works in both qualitative and quantitative evaluations.

cs.CV↗

arXiv2024

Tunable beam splitting via photorefractive nonlinearity and its applications in chiral waveguide induction and vortex generation

We report experimental observation and theoretical explanation of novel propagation regimes for optical beams in an artificial nonlinear material with outstanding photorefractive properties. Nondiffractive beams, which keep their shapes invariant in the free space, feature self-splitting from the middle in two separating secondary beams, due to the light-matter interaction. The splitting degree is controlled by means of a phase-pre-modulation method. We propose applications of the self-splitting to the creation of an effectively chiral waveguide and the generation of even-order vortices.

physics.optics↗

arXiv2024

Task-Agnostic Language Model Watermarking via High Entropy Passthrough Layers

In the era of costly pre-training of large language models, ensuring the intellectual property rights of model owners, and insuring that said models are responsibly deployed, is becoming increasingly important. To this end, we propose model watermarking via passthrough layers, which are added to existing pre-trained networks and trained using a self-supervised loss such that the model produces high-entropy output when prompted with a unique private key, and acts normally otherwise. Unlike existing model watermarking methods, our method is fully task-agnostic, and can be applied to both classification and sequence-to-sequence tasks without requiring advanced access to downstream fine-tuning datasets. We evaluate the proposed passthrough layers on a wide range of downstream tasks, and show experimentally our watermarking method achieves a near-perfect watermark extraction accuracy and false-positive rate in most cases without damaging original model performance. Additionally, we show our method is robust to both downstream fine-tuning, fine-pruning, and layer removal attacks, and can be trained in a fraction of the time required to train the original model. Code is available in the paper.

cs.CL↗

arXiv2024

Multi-User Matching and Resource Allocation in Vision Aided Communications

Visual perception is an effective way to obtain the spatial characteristics of wireless channels and to reduce the overhead for communications system. A critical problem for the visual assistance is that the communications system needs to match the radio signal with the visual information of the corresponding user, i.e., to identify the visual user that corresponds to the target radio signal from all the environmental objects. In this paper, we propose a user matching method for environment with a variable number of objects. Specifically, we apply 3D detection to extract all the environmental objects from the images taken by multiple cameras. Then, we design a deep neural network (DNN) to estimate the location distribution of users by the images and beam pairs at multiple moments, and thereby identify the users from all the extracted environmental objects. Moreover, we present a resource allocation method based on the taken images to reduce the time and spectrum overhead compared to traditional resource allocation methods. Simulation results show that the proposed user matching method outperforms the existing methods, and the proposed resource allocation method can achieve $92\%$ transmission rate of the traditional resource allocation method but with the time and spectrum overhead significantly reduced.

eess.SP↗

arXiv2024

A Search for Asymmetric Kinematic Components in Circumstellar Envelopes Using OH Main Line Masers

Circumstellar OH maser lines are useful for studying the dynamics of the circumstellar envelope (CSE) around evolved stars. This study aims to identify CSEs around cold stars, which exhibit deviations from the spherical expansion, by comparing the velocity ranges of the OH main lines (1665/1667 MHz) with those of the satellite line (1612 MHz), using a database of circumstellar OH maser sources. We performed this comparison for 377 circumstellar OH maser sources. In addition, using infrared two-color diagrams, we examined the evolutionary stages and infrared properties of objects showing velocity excess (velocity excess means the detection of the main lines outside the velocity range of the satellite line). A periodicity analysis of the WISE light curves was also carried out. As a result of the velocity range comparison, eight circumstellar OH maser sources were found to exhibit velocity excess. The infrared colors of these objects match those of post-AGB stars. Periodic variations were observed in the WISE light curves of five of these eight objects. The results suggest that examining velocity excess of the main lines relative to the satellite line is scientifically significant because mainline masers probe the CSE dynamics over a broader range of evolutionary stages compared to the 22.235 GHz H${_2}$O maser line. Additionally, during the post-AGB phase, the emission regions of the mainline and 22.235 GHz H${_2}$O masers may overlap in a CSE, whereas they originate from different regions during the AGB phase.

astro-ph.SR↗

arXiv2024

DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos

Estimating video depth in open-world scenarios is challenging due to the diversity of videos in appearance, content motion, camera movement, and length. We present DepthCrafter, an innovative method for generating temporally consistent long depth sequences with intricate details for open-world videos, without requiring any supplementary information such as camera poses or optical flow. The generalization ability to open-world videos is achieved by training the video-to-depth model from a pre-trained image-to-video diffusion model, through our meticulously designed three-stage training strategy. Our training approach enables the model to generate depth sequences with variable lengths at one time, up to 110 frames, and harvest both precise depth details and rich content diversity from realistic and synthetic datasets. We also propose an inference strategy that can process extremely long videos through segment-wise estimation and seamless stitching. Comprehensive evaluations on multiple datasets reveal that DepthCrafter achieves state-of-the-art performance in open-world video depth estimation under zero-shot settings. Furthermore, DepthCrafter facilitates various downstream applications, including depth-based visual effects and conditional video generation.

cs.CV↗

arXiv2024

MAIN-VC: Lightweight Speech Representation Disentanglement for One-shot Voice Conversion

One-shot voice conversion aims to change the timbre of any source speech to match that of the unseen target speaker with only one speech sample. Existing methods face difficulties in satisfactory speech representation disentanglement and suffer from sizable networks as some of them leverage numerous complex modules for disentanglement. In this paper, we propose a model named MAIN-VC to effectively disentangle via a concise neural network. The proposed model utilizes Siamese encoders to learn clean representations, further enhanced by the designed mutual information estimator. The Siamese structure and the newly designed convolution module contribute to the lightweight of our model while ensuring performance in diverse voice conversion tasks. The experimental results show that the proposed model achieves comparable subjective scores and exhibits improvements in objective metrics compared to existing methods in a one-shot voice conversion scenario.

cs.SD↗

arXiv2024

Text-guided Zero-Shot Object Localization

Object localization is a hot issue in computer vision area, which aims to identify and determine the precise location of specific objects from image or video. Most existing object localization methods heavily rely on extensive labeled data, which are costly to annotate and constrain their applicability. Therefore, we propose a new Zero-Shot Object Localization (ZSOL) framework for addressing the aforementioned challenges. In the proposed framework, we introduce the Contrastive Language Image Pre-training (CLIP) module which could integrate visual and linguistic information effectively. Furthermore, we design a Text Self-Similarity Matching (TSSM) module, which could improve the localization accuracy by enhancing the representation of text features extracted by CLIP module. Hence, the proposed framework can be guided by prompt words to identify and locate specific objects in an image in the absence of labeled samples. The results of extensive experiments demonstrate that the proposed method could improve the localization performance significantly and establishes an effective benchmark for further research.

cs.CV↗

arXiv2024

Sign-changing solution for an overdetermined elliptic problem on unbounded domain

We prove the existence of two smooth families of unbounded domains in $\mathbb{R}^{N+1}$ with $N\geq1$ such that \begin{equation} -Δu=λu\,\, \text{in}\,\,Ω, \,\, u=0,\,\,\partial_νu=\text{const}\,\,\text{on}\,\,\partialΩ\nonumber \end{equation} admits a sign-changing solution. The domains bifurcate from the straight cylinder $B_1\times \mathbb{R}$, where $B_1$ is the unit ball in $\mathbb{R}^N$. These results can be regarded as counterexamples to the Berenstein conjecture on unbounded domain. Unlike most previous papers in this direction, a very delicate issue here is that there may be two-dimensional kernel space at some bifurcation point. Thus a Crandall-Rabinowitz type bifurcation theorem from high-dimensional kernel space is also established to achieve the goal.

math.AP↗

arXiv2024

Modeling and scaling spontaneous imbibition with generalized fractional flow theory and non-Boltzmann transformation

Spontaneous imbibition (SI) is a process by which liquid is drawn into partially saturated porous media by capillary forces, relevant for subsurface processes like underground fluid storage and withdrawal. Accurate modeling and scaling of counter-current SI have long been challenging. In this study, we proposed a generalized fractional flow theory (GFFT) using the Hausdorff fractal derivative, combined with non-Boltzmann scaling. The model links imbibition distance to time through the power law exponent alpha/2, where alpha is the fractal index (0 < alpha < 2 in this study). We applied the GFFT to various experimental and stimulated datasets of both porous and fractured media, finding that alpha varied with factors such as contact angle (of the imbibing fluid), dynamic viscosity, pore structure, and fracture properties. By analyzing SI data from sandstones, diatomite, carbonate, and synthetic porous media, we demonstrated that the non-Boltzmann scaling provided a better collapse of the SI data than the traditional Boltzmann approach alpha = 1), with alpha values ranging from 0.88 to 1.54. These deviations illustrate the model's adaptability to different porous materials. Using the GFFT, we expect to better predict fluid imbibition rates when properties like porosity, permeability, initial and maximum saturations, viscosity, and wettability are known, offering a more accurate alternative to traditional models.

physics.flu-dyn↗

arXiv2024

User Activity Detection with Delay-Calibration for Asynchronous Massive Random Access

This work considers an uplink asynchronous massive random access scenario in which a large number of users asynchronously access a base station equipped with multiple receive antennas. The objective is to alleviate the problem of massive collision due to the limited number of orthogonal preambles of an access scheme in which user activity detection is performed. We propose a user activity detection with delay-calibration (UAD-DC) algorithm and investigate the benefits of oversampling for the estimation of continuous time delays at the receiver. The proposed algorithm iteratively estimates time delays and detects active users by noting that the collided users can be identified through accurate estimation of time delays. Due to the sporadic user activity patterns, the user activity detection problem can be formulated as a compressive sensing (CS) problem, which can be solved by a modified Turbo-CS algorithm under the consideration of correlated noise samples resulting from oversampling. A sliding-window technique is applied in the proposed algorithm to reduce the overall computational complexity. Moreover, we propose a new design of the pulse shaping filter by minimizing the Bayesian Cramér-Rao bound of the detection problem under the constraint of limited spectral bandwidth. Numerical results demonstrate the efficacy of the proposed algorithm in terms of the normalized mean squared error of the estimated channel, the probability of misdetection and the successful detection ratio.

eess.SP↗

arXiv2024

Functional central limit theorems for spatial averages of the parabolic Anderson model with delta initial condition in dimension $d\geq 1$

Let $\{u(t,x)\}_{t>0,x\in{{\mathbb R}^{d}}}$ denote the solution to a $d$-dimensional parabolic Anderson model with delta initial condition and driven by a multiplicative noise that is white in time and has a spatially homogeneous covariance given by a nonnegative-definite measure $f$. Let $S_{N,t}:=N^{-d}\int_{{[0,N]}^d}{[U(t,x)-1]}{\rm d}x$ denote the spatial average on ${{\mathbb R}^{d}}$. We obtain various functional central limit theorems (CLTs) for spatial averages based on the quantitative analysis of $f$ and spatial dimension $d$. In particular, when $f$ is given by Riesz kernel, that is, $f({\rm x})={\Vert x \Vert}^{-β}{\rm d}x$, $β\in(0,2\wedge d)$, the functional CLT is also based on the index $β$.

math.PR↗

arXiv2024

Accelerating Fock build via hybrid analytical-numerical integration

A very robust and efficient hybrid analytic-numerical Fock build, aMECP+aCOSx, has been developed for accelerating HF/DFT calculations. The essential idea is to extract those portions of the Fock matrix that can readily be evaluated analytically, so as to minimize numerical noises arising from the semi-numerical and numerical integrations. As a result, the combination of aMECP with a medium grid and aCOSx with a coarse grid is already sufficient to achieve an accuracy of less than 1μEh/atom in total energies. The acceleration of aMECP+aCOSx over the analytic Fock build is already seen in calculations of small molecular systems and is more enhanced in calculations of large molecules with extended basis sets.

physics.chem-ph↗

arXiv2024

Training Fair Models in Federated Learning without Data Privacy Infringement

Training fair machine learning models becomes more and more important. As many powerful models are trained by collaboration among multiple parties, each holding some sensitive data, it is natural to explore the feasibility of training fair models in federated learning so that the fairness of trained models, the data privacy of clients, and the collaboration between clients can be fully respected simultaneously. However, the task of training fair models in federated learning is challenging, since it is far from trivial to estimate the fairness of a model without knowing the private data of the participating parties, which is often constrained by privacy requirements in federated learning. In this paper, we first propose a federated estimation method to accurately estimate the fairness of a model without infringing the data privacy of any party. Then, we use the fairness estimation to formulate a novel problem of training fair models in federated learning. We develop FedFair, a well-designed federated learning framework, which can successfully train a fair model with high performance without data privacy infringement. Our extensive experiments on three real-world data sets demonstrate the excellent fair model training performance of our method.

cs.LG↗