arXiv subjects

Jianfeng Zhang

Publications and source records attributed to Jianfeng Zhang.

At least 73 records · Page 4Linked to original sources

arXiv2024

Synthesizer Based Efficient Self-Attention for Vision Tasks

Self-attention module shows outstanding competence in capturing long-range relationships while enhancing performance on vision tasks, such as image classification and image captioning. However, the self-attention module highly relies on the dot product multiplication and dimension alignment among query-key-value features, which cause two problems: (1) The dot product multiplication results in exhaustive and redundant computation. (2) Due to the visual feature map often appearing as a multi-dimensional tensor, reshaping the scale of the tensor feature to adapt to the dimension alignment might destroy the internal structure of the tensor feature map. To address these problems, this paper proposes a self-attention plug-in module with its variants, namely, Synthesizing Tensor Transformations (STT), for directly processing image tensor features. Without computing the dot-product multiplication among query-key-value, the basic STT is composed of the tensor transformation to learn the synthetic attention weight from visual information. The effectiveness of STT series is validated on the image classification and image caption. Experiments show that the proposed STT achieves competitive performance while keeping robustness compared to self-attention in the aforementioned vision tasks.

cs.CV↗

arXiv2024

High Quality Human Image Animation using Regional Supervision and Motion Blur Condition

Recent advances in video diffusion models have enabled realistic and controllable human image animation with temporal coherence. Although generating reasonable results, existing methods often overlook the need for regional supervision in crucial areas such as the face and hands, and neglect the explicit modeling for motion blur, leading to unrealistic low-quality synthesis. To address these limitations, we first leverage regional supervision for detailed regions to enhance face and hand faithfulness. Second, we model the motion blur explicitly to further improve the appearance quality. Third, we explore novel training strategies for high-resolution human animation to improve the overall fidelity. Experimental results demonstrate that our proposed method outperforms state-of-the-art approaches, achieving significant improvements upon the strongest baseline by more than 21.0% and 57.4% in terms of reconstruction precision (L1) and perceptual quality (FVD) on HumanDance dataset. Code and model will be made available.

cs.CV↗

arXiv2024

Non-Hermitian Boundary in a Surface Selective Reconstructed Magnetic Weyl Semimetal

Non-Hermitian physics, studying systems described by non-Hermitian Hamiltonians, reveals unique phenomena not present in Hermitian systems. Unlike Hermitian systems, non-Hermitian systems have complex eigenvalues, making their effects less directly observable. Recently, significant efforts have been devoted to incorporating the non-Hermitian effects into condensed matter physics. However, progress has been hindered by the absence of a viable experimental approach. Here, the discovery of surface-selectively spontaneous reconstructed Weyl semimetal NdAlSi provides a feasible experimental platform for studying non-Hermitian physics. Utilizing angle-resolved photoemission spectroscopy measurements, surface-projected density functional theory calculations, and scanning tunneling microscopy measurements, we demonstrate that surface reconstruction in NdAlSi alters surface Fermi arc connectivity and generates new isolated non-topological surface Fermi arcs. In the presence of a magnetic field, the surface-selective spontaneous reconstructed Weyl semimetal NdAlSi can be viewed as a Hermitian bulk--non-Hermitian boundary system. The isolated non-topological surface Fermi arcs on the reconstructed surface act as a loss mechanism and open boundary condition for the topological electrons and bulk states, serving as non-Hermitian boundary states. This discovery provides a good experimental platform for exploring new physical phenomena and potential applications based on boundary non-Hermitian effects, extending beyond purely mathematical concepts.

cond-mat.mtrl-sci↗

arXiv2024

Resolving the Electronic Ground State of La3Ni2O7-δ Films

The recent discovery of a superconductivity signature in La3Ni2O7-δ under a pressure of 14 GPa, with a superconducting transition temperature of around 80 K, has attracted considerable attention. An important aspect of investigating electronic structures is discerning the extent to which the electronic ground state of La3Ni2O7-δ resembles the parent state of the cuprate superconductor, a charge transfer insulator with long-range antiferromagnetism. Through X-ray absorption spectroscopy, we have uncovered the crucial influence of oxygen ligands on the electronic ground states of the Ni ions, displaying a charge transfer nature akin to cuprate but with distinct orbital configurations. Both in-plane and out-of-plane Zhang-Rice singlets associated with Ni d_(x^2-y^2 ) and d_(z^2) orbitals are identified, together with a strong interlayer coupling through inner apical oxygen. Additionally, in La3Ni2O7-δ films, we have detected a superlattice reflection (1/4, 1/4, L) at the Ni L absorption edge using resonant X-ray scattering measurements. Further examination of the resonance profile indicates that the reflection originates from the Ni d orbitals. By evaluating the reflection's azimuthal angle dependence, we have confirmed the presence of collinear antiferromagnetic spin ordering and charge-like anisotropy ordered with the same periodicity. Notably, our findings reveal a microscopic relationship between these two components in the temperature dependence of the scattering intensity of the reflection. This investigation enriches our understanding of high-temperature superconductivity in La3Ni2O7-δ under high pressure.

cond-mat.supr-con↗

arXiv2024

A Shapelet-based Framework for Unsupervised Multivariate Time Series Representation Learning

Recent studies have shown great promise in unsupervised representation learning (URL) for multivariate time series, because URL has the capability in learning generalizable representation for many downstream tasks without using inaccessible labels. However, existing approaches usually adopt the models originally designed for other domains (e.g., computer vision) to encode the time series data and {rely on strong assumptions to design learning objectives, which limits their ability to perform well}. To deal with these problems, we propose a novel URL framework for multivariate time series by learning time-series-specific shapelet-based representation through a popular contrasting learning paradigm. To the best of our knowledge, this is the first work that explores the shapelet-based embedding in the unsupervised general-purpose representation learning. A unified shapelet-based encoder and a novel learning objective with multi-grained contrasting and multi-scale alignment are particularly designed to achieve our goal, and a data augmentation library is employed to improve the generalization. We conduct extensive experiments using tens of real-world datasets to assess the representation quality on many downstream tasks, including classification, clustering, and anomaly detection. The results demonstrate the superiority of our method against not only URL competitors, but also techniques specially designed for downstream tasks. Our code has been made publicly available at https://github.com/real2fish/CSL.

cs.LG↗

arXiv2024

Set Values of Dynamic Nonzero Sum Games and Set Valued Hamiltonians

It is well known that the (unique) value of a stochastic control problem or a two person zero sum game under Isaacs condition can be characterized through a PDE driven by the Hamiltonian. Our goal of this paper is to extend this classical result to nonzero sum games, which typically have multiple Nash equilibria and multiple values. Our object is the set value of the game, which roughly speaking is the set of values over all equilibria and thus is by nature unique. We shall introduce set valued Hamiltonians and characterize the set value of the game through backward SDEs driven by appropriate selectors of the set valued Hamiltonians, where the selectors are typically path dependent. When the set valued Hamiltonian is a singleton, our result covers the standard control problem and two person zero sum game problem under Isaacs condition.

math.OC↗

arXiv2024

Disorder Driven Non-Anderson Transition in a Weyl Semimetal

For several decades, it was widely believed that a non-interacting disordered electronic system could only undergo an Anderson metal-insulator transition due to Anderson localization. However, numerous recent theoretical works have predicted the existence of a disorder-driven non-Anderson phase transition that differ from Anderson localization. The frustration lies in the fact that this non-Anderson disorder-driven transition has not yet been experimentally demonstrated in any system. Here, using angle-resolved photoemission spectroscopy, we present a case study of observing the non-Anderson disorder-driven transition by visualizing the electronic structure of the Weyl semimetal NdAlSi on surfaces with varying amounts of disorder. Our observations reveal that strong disorder can effectively suppress all surface states in the Weyl semimetal NdAlSi, including the topological surface Fermi arcs. This disappearance of surface Fermi arcs is associated with the vanishing of the bulk topological invariant, indicating a quantum phase transition from a Weyl semimetal to a diffusive metal. By analyzing the changes in the electronic structure of NdAlSi, as the surface degrades, we provide a physical picture of this non-Anderson transition from a Weyl semimetal to a diffuse metal. These observations provide the first direct experimental evidence of the non-Anderson disorder-driven transition, a discovery long anticipated by theoretical physicists. The finding dispels longstanding suspicions among researchers that non-Anderson transitions exist in real quantum systems.

cond-mat.mtrl-sci↗

arXiv2024

SEGAN: semi-supervised learning approach for missing data imputation

In many practical real-world applications, data missing is a very common phenomenon, making the development of data-driven artificial intelligence theory and technology increasingly difficult. Data completion is an important method for missing data preprocessing. Most existing miss-ing data completion models directly use the known information in the missing data set but ignore the impact of the data label information contained in the data set on the missing data completion model. To this end, this paper proposes a missing data completion model SEGAN based on semi-supervised learning, which mainly includes three important modules: generator, discriminator and classifier. In the SEGAN model, the classifier enables the generator to make more full use of known data and its label information when predicting missing data values. In addition, the SE-GAN model introduces a missing hint matrix to allow the discriminator to more effectively distinguish between known data and data filled by the generator. This paper theoretically proves that the SEGAN model that introduces a classifier and a missing hint matrix can learn the real known data distribution characteristics when reaching Nash equilibrium. Finally, a large number of experiments were conducted in this article, and the experimental results show that com-pared with the current state-of-the-art multivariate data completion method, the performance of the SEGAN model is improved by more than 3%.

cs.LG↗

arXiv2024

A Dynamic Principal Agent Problem with One-sided Commitment

In this paper we consider a principal agent problem where the agent is allowed to quit, by incurring a cost. When the current agent quits the job, the principal will hire a new one, possibly with a different type. We characterize the principal's dynamic value function, which could be discontinuous at the boundary, as the (unique) minimal solution of an infinite dimensional system of HJB equations, parametrized by the agent's type. This dynamic problem is time consistent in certain sense. Some interesting findings are worth mentioning. First, self-enforcing contracts are typically suboptimal. The principal would rather let the agent quit and hire a new one. Next, the standard contract for a committed agent may also be suboptimal, due to the presence of different types of agents in our model. The principal may prefer no commitment from the agent, then she can hire a cheaper one from the market at a later time by designing the contract to induce the current agent to quit. Moreover, due to the cost incurring to the agent, the principal will see only finitely many quittings.

math.OC↗

arXiv2024

ViT-Lens: Initiating Omni-Modal Exploration through 3D Insights

Though the success of CLIP-based training recipes in vision-language models, their scalability to more modalities (e.g., 3D, audio, etc.) is limited to large-scale data, which is expensive or even inapplicable for rare modalities. In this paper, we present ViT-Lens that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning to a pre-defined space. Specifically, the modality-specific lens is tuned to project multimodal signals to the shared embedding space, which are then processed by a strong ViT that carries pre-trained image knowledge. The encoded multimodal representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. A well-trained lens with a ViT backbone has the potential to serve as one of these foundation models, supervising the learning of subsequent modalities. ViT-Lens provides a unified solution for representation learning of increasing modalities with two appealing benefits: (i) Exploiting the pretrained ViT across tasks and domains effectively with efficient data regime; (ii) Emergent downstream capabilities of novel modalities are demonstrated due to the modality alignment space. We evaluate ViT-Lens in the context of 3D as an initial verification. In zero-shot 3D classification, ViT-Lens achieves substantial improvements over previous state-of-the-art, showing 52.0% accuracy on Objaverse-LVIS, 87.4% on ModelNet40, and 60.6% on ScanObjectNN. Furthermore, we enable zero-shot 3D question-answering by simply integrating the trained 3D lens into the InstructBLIP model without any adaptation. We will release the results of ViT-Lens on more modalities in the near future.

cs.CV↗

arXiv2024

ViT-Lens: Towards Omni-modal Representations

Aiming to advance AI agents, large foundation models significantly improve reasoning and instruction execution, yet the current focus on vision and language neglects the potential of perceiving diverse modalities in open-world environments. However, the success of data-driven vision and language models is costly or even infeasible to be reproduced for rare modalities. In this paper, we present ViT-Lens-2 that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning them to a pre-defined space. Specifically, the modality-specific lens is tuned to project any-modal signals to an intermediate embedding space, which are then processed by a strong ViT with pre-trained visual knowledge. The encoded representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. ViT-Lens-2 provides a unified solution for representation learning of increasing modalities with two appealing advantages: (i) Unlocking the great potential of pretrained ViTs to novel modalities effectively with efficient data regime; (ii) Enabling emergent downstream capabilities through modality alignment and shared ViT parameters. We tailor ViT-Lens-2 to learn representations for 3D point cloud, depth, audio, tactile and EEG, and set new state-of-the-art results across various understanding tasks, such as zero-shot classification. By seamlessly integrating ViT-Lens-2 into Multimodal Foundation Models, we enable Any-modality to Text and Image Generation in a zero-shot manner. Code and models are available at https://github.com/TencentARC/ViT-Lens.

cs.CV↗

arXiv2024

An edge detection-based deep learning approach for tear meniscus height measurement

Automatic measurements of tear meniscus height (TMH) have been achieved by using deep learning techniques; however, annotation is significantly influenced by subjective factors and is both time-consuming and labor-intensive. In this paper, we introduce an automatic TMH measurement technique based on edge detection-assisted annotation within a deep learning framework. This method generates mask labels less affected by subjective factors with enhanced efficiency compared to previous annotation approaches. For improved segmentation of the pupil and tear meniscus areas, the convolutional neural network Inceptionv3 was first implemented as an image quality assessment model, effectively identifying higher-quality images with an accuracy of 98.224%. Subsequently, by using the generated labels, various algorithms, including Unet, ResUnet, Deeplabv3+FcnResnet101, Deeplabv3+FcnResnet50, FcnResnet50, and FcnResnet101 were trained, with Unet demonstrating the best performance. Finally, Unet was used for automatic pupil and tear meniscus segmentation to locate the center of the pupil and calculate TMH,respectively. An evaluation of the mask quality predicted by Unet indicated a Mean Intersection over Union of 0.9362, a recall of 0.9261, a precision of 0.9423, and an F1-Score of 0.9326. Additionally, the TMH predicted by the model was assessed, with the fitting curve represented as y= 0.982x-0.862, an overall correlation coefficient of r^2=0.961 , and an accuracy of 94.80% (237/250). In summary, the algorithm can automatically screen images based on their quality,segment the pupil and tear meniscus areas, and automatically measure TMH. Measurement results using the AI algorithm demonstrate a high level of consistency with manual measurements, offering significant support to clinical doctors in diagnosing dry eye disease.

eess.IV↗

arXiv2024

Set Values for Mean Field Games

In this paper we study mean field games with possibly multiple mean field equilibria. Instead of focusing on the individual equilibria, we propose to study the set of values over all possible equilibria, which we call the set value of the mean field game. When the mean field equilibrium is unique, typically under certain monotonicity conditions, our set value reduces to the singleton of the standard value function which solves the master equation. The set value is by nature unique, and we shall establish two crucial properties: (i) the dynamic programming principle, also called time consistency; and (ii) the convergence of the set values of the corresponding $N$-player games, which can be viewed as a type of stability result. To our best knowledge, this is the first work in the literature which studies the dynamic value of mean field games without requiring the uniqueness of mean field equilibria. We emphasize that the set value is very sensitive to the type of the admissible controls. In particular, for the convergence one has to restrict to corresponding types of equilibria for the N-player game and for the mean field game. We shall illustrate this point by investigating three cases, two in finite state space models and the other in a continuous time model with controlled diffusions.

math.OC↗

arXiv2023

DREAM-Talk: Diffusion-based Realistic Emotional Audio-driven Method for Single Image Talking Face Generation

The generation of emotional talking faces from a single portrait image remains a significant challenge. The simultaneous achievement of expressive emotional talking and accurate lip-sync is particularly difficult, as expressiveness is often compromised for the accuracy of lip-sync. As widely adopted by many prior works, the LSTM network often fails to capture the subtleties and variations of emotional expressions. To address these challenges, we introduce DREAM-Talk, a two-stage diffusion-based audio-driven framework, tailored for generating diverse expressions and accurate lip-sync concurrently. In the first stage, we propose EmoDiff, a novel diffusion module that generates diverse highly dynamic emotional expressions and head poses in accordance with the audio and the referenced emotion style. Given the strong correlation between lip motion and audio, we then refine the dynamics with enhanced lip-sync accuracy using audio features and emotion style. To this end, we deploy a video-to-video rendering module to transfer the expressions and lip motions from our proxy 3D avatar to an arbitrary portrait. Both quantitatively and qualitatively, DREAM-Talk outperforms state-of-the-art methods in terms of expressiveness, lip-sync accuracy and perceptual quality.

cs.CV↗

arXiv2023

From finite population optimal stopping to mean field optimal stopping

This paper analyzes the convergence of the finite population optimal stopping problem towards the corresponding mean field limit. Building on the viscosity solution characterization of the mean field optimal stopping problem of our previous papers [Talbi, Touzi & Zhang 2021 & 2022], we prove the convergence of the value functions by adapting the Barles-Souganidis [1991] monotone scheme method to our context. We next characterize the optimal stopping policies of the mean field problem by the accumulation points of the finite population optimal stopping strategies. In particular, if the limiting problem has a unique optimal stopping policy, then the finite population optimal stopping strategies do converge towards this solution. As a by-product of our analysis, we provide an extension of the standard propagation of chaos to the context of stopped McKean-Vlasov diffusions.

math.PR↗

arXiv2023

AvatarStudio: High-fidelity and Animatable 3D Avatar Creation from Text

We study the problem of creating high-fidelity and animatable 3D avatars from only textual descriptions. Existing text-to-avatar methods are either limited to static avatars which cannot be animated or struggle to generate animatable avatars with promising quality and precise pose control. To address these limitations, we propose AvatarStudio, a coarse-to-fine generative model that generates explicit textured 3D meshes for animatable human avatars. Specifically, AvatarStudio begins with a low-resolution NeRF-based representation for coarse generation, followed by incorporating SMPL-guided articulation into the explicit mesh representation to support avatar animation and high resolution rendering. To ensure view consistency and pose controllability of the resulting avatars, we introduce a 2D diffusion model conditioned on DensePose for Score Distillation Sampling supervision. By effectively leveraging the synergy between the articulated mesh representation and the DensePose-conditional diffusion model, AvatarStudio can create high-quality avatars from text that are ready for animation, significantly outperforming previous methods. Moreover, it is competent for many applications, e.g., multimodal avatar animations and style-guided avatar creation. For more results, please refer to our project page: http://jeff95.me/projects/avatarstudio.html

cs.GR↗

arXiv2023

MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model

This paper studies the human image animation task, which aims to generate a video of a certain reference identity following a particular motion sequence. Existing animation works typically employ the frame-warping technique to animate the reference image towards the target motion. Despite achieving reasonable results, these approaches face challenges in maintaining temporal consistency throughout the animation due to the lack of temporal modeling and poor preservation of reference identity. In this work, we introduce MagicAnimate, a diffusion-based framework that aims at enhancing temporal consistency, preserving reference image faithfully, and improving animation fidelity. To achieve this, we first develop a video diffusion model to encode temporal information. Second, to maintain the appearance coherence across frames, we introduce a novel appearance encoder to retain the intricate details of the reference image. Leveraging these two innovations, we further employ a simple video fusion technique to encourage smooth transitions for long video animation. Empirical results demonstrate the superiority of our method over baseline approaches on two benchmarks. Notably, our approach outperforms the strongest baseline by over 38% in terms of video fidelity on the challenging TikTok dancing dataset. Code and model will be made available.

cs.CV↗

arXiv2023

XAGen: 3D Expressive Human Avatars Generation

Recent advances in 3D-aware GAN models have enabled the generation of realistic and controllable human body images. However, existing methods focus on the control of major body joints, neglecting the manipulation of expressive attributes, such as facial expressions, jaw poses, hand poses, and so on. In this work, we present XAGen, the first 3D generative model for human avatars capable of expressive control over body, face, and hands. To enhance the fidelity of small-scale regions like face and hands, we devise a multi-scale and multi-part 3D representation that models fine details. Based on this representation, we propose a multi-part rendering technique that disentangles the synthesis of body, face, and hands to ease model training and enhance geometric quality. Furthermore, we design multi-part discriminators that evaluate the quality of the generated avatars with respect to their appearance and fine-grained control capabilities. Experiments show that XAGen surpasses state-of-the-art methods in terms of realism, diversity, and expressive control abilities. Code and data will be made available at https://showlab.github.io/xagen.

cs.CV↗