arXiv subjects

Weitao Chen

Publications and source records attributed to Weitao Chen.

At least 19 recordsLinked to original sources

arXiv2026

Connecting finite-size scaling and renormalization-group flows in Anderson localization on random graphs

We investigate the relation between two complementary descriptions of the Anderson transition on small-world random graphs: finite-size scaling of eigenfunction moments and renormalization-group (RG) flows of multifractal dimensions $D_q$. The RG flows recover the qualitative structure previously reported for the information dimension $D_1$, while their extension to other $D_q$ with smaller moment orders $q<1/2$ confirms the strongly multifractal nature of the localized phase characterized by a clear separation between multifractal and localized behaviors at strong disorder. On the other hand, we show that we can construct another beta function characterizing the flow of eigenfunction moments, whose collapse onto a single function across different disorder strengths provides a clear confirmation of single-parameter scaling property. These results indicate that the family of RG trajectories observed for the information dimension $D_1$ need not imply two-parameter scaling and Kosterlitz-Thouless like critical behavior. To characterize these properties, one should consider different observables, such as large $q>1/2$ and small $q<1/2$ eigenfunction moments, which are associated with distinct critical behaviors, in particular different critical exponents.

cond-mat.dis-nn↗

arXiv2026

Long-range Nonlinear Sigma Model for a Singular Quantum Kicked Rotor

Singular kicked rotors have long been compared with power-law random banded matrices (PRBM) because their momentum-space Floquet matrix elements decay algebraically. However, it has remained unclear whether the deterministic correlations of the rotor become irrelevant at long distances and, consequently, under what conditions the two systems share the same infrared theory. To address this question, we derive a nonlocal supersymmetric nonlinear sigma model directly from a quantum kicked rotor with a power-law or logarithmic singularity. By carrying out the renormalization-group analysis up to two-loop order, we show that, after matching the symmetry class and coupling convention, the rotor reproduces the long-range Anderson transition of the corresponding PRBM, including its localized, critical, and extended infrared regimes.

cond-mat.dis-nn↗

arXiv2026

EgoMonth: A Month-Level Egocentric Video Benchmark for Long-Term Spatiotemporal Memory

Recent advances in Multimodal Large Language Models (MLLMs) have led to substantial progress in video understanding, accompanied by a growing number of long video benchmarks. However, existing benchmarks rely predominantly on web-sourced videos that lack inter-clip spatiotemporal continuity, making it difficult to assess whether models can maintain consistent memory across days or weeks of real-world experience. We introduce EgoMonth, the first month-level egocentric video understanding benchmark. EgoMonth comprises over 300 hours of first-person daily-life recordings from 20 participants spanning 20 to 120 days, paired with 1,443 human-crafted multiple-choice question-answer pairs. We design a cognitively grounded 14-task evaluation framework organized into three hierarchical cognitive levels: Schema Consolidation, Episodic Indexing, and Cascading Reasoning. Evaluation of state-of-the-art open-source and closed-source MLLMs reveals that even the best-performing model, Gemini 2.5 Pro, achieves only 71.8% macro-average accuracy, remaining 22.4 percentage points below the corrected human baseline of 94.2%. Several models perform near or below the 25% chance level on tasks such as Route Reasoning, Cross-view Spatial Reasoning, and Direction Judgement, while even the strongest closed-source model remains substantially below human performance. These results indicate that current MLLMs function as lossy summarizers rather than faithful memorizers, highlighting the need for architectures with genuine long-term spatiotemporal memory.

cs.CV↗

arXiv2026

Learning-guided Kansa collocation for forward and inverse PDEs beyond linearity

Partial Differential Equations are precise in modelling the physical, biological and graphical phenomena. However, the numerical methods suffer from the curse of dimensionality, high computation costs and domain-specific discretization. We aim to explore pros and cons of different PDE solvers, and apply them to specific scientific simulation problems, including forwarding solution, inverse problems and equations discovery. In particular, we extend the recent CNF (NeurIPS 2023) framework solver to coupled and non-linear settings, together with down-stream applications. The outcomes include implementation of selected methods, self-tuning techniques, evaluation on benchmark problems and a comprehensive survey of neural PDE solvers and scientific simulation applications.

cs.CE↗

arXiv2025

Cyc3D: Fine-grained Controllable 3D Generation via Cycle Consistency Regularization

Despite the remarkable progress of 3D generation, achieving controllability, i.e., ensuring consistency between generated 3D content and input conditions like edge and depth, remains a significant challenge. Existing methods often struggle to maintain accurate alignment, leading to noticeable discrepancies. To address this issue, we propose \name{}, a new framework that enhances controllable 3D generation by explicitly encouraging cyclic consistency between the second-order 3D content, generated based on extracted signals from the first-order generation, and its original input controls. Specifically, we employ an efficient feed-forward backbone that can generate a 3D object from an input condition and a text prompt. Given an initial viewpoint and a control signal, a novel view is rendered from the generated 3D content, from which the extracted condition is used to regenerate the 3D content. This re-generated output is then rendered back to the initial viewpoint, followed by another round of control signal extraction, forming a cyclic process with two consistency constraints. \emph{View consistency} ensures coherence between the two generated 3D objects, measured by semantic similarity to accommodate generative diversity. \emph{Condition consistency} aligns the final extracted signal with the original input control, preserving structural or geometric details throughout the process. Extensive experiments on popular benchmarks demonstrate that \name{} significantly improves controllability, especially for fine-grained details, outperforming existing methods across various conditions (e.g., +14.17\% PSNR for edge, +6.26\% PSNR for sketch).

cs.CV↗

arXiv2025

Critical Dynamics of the Anderson Transition on Small-World Graphs

The Anderson transition on random graphs draws interest through its resemblance to the many-body localization (MBL) transition with similarly debated properties. In this Letter, we construct a unitary Anderson model on Small-World graphs to characterize long time and large size wave-packet dynamics across the Anderson transition. We reveal the logarithmically slow non-ergodic dynamics in the critical regime, confirming recent random matrix predictions. Our data clearly indicate two localization times: an average localization time that diverges, while the typical one saturates. In the delocalized regime, the dynamics are initially non-ergodic but cross over to ergodic diffusion at long times and large distances. Finite-time scaling then allows us to characterize the critical dynamical properties: the logarithm of the average localization time diverges algebraically, while the ergodic time diverges exponentially. Our results could be used to clarify the dynamical properties of MBL and could guide future experiments with quantum simulators.

cond-mat.dis-nn↗

arXiv2025

Enhancing Quantum Metrology by Quantum Resonance Dynamics

Quantum effects in metrology can in principle enhance measurement precision from the so-called standard quantum limit to the Heisenberg Limit. Further advancements in quantum metrology largely rely on innovative metrology protocols that can avoid a number of known obstacles, including the challenge of preparing entangled states with sufficient fidelity, the readout noise in measuring highly entangled states, and no-go theorems for quantum metrology under noisy environments. In this Letter, exploiting some peculiar but experimentally feasible dynamical features of a collection of spins with all-to-all time-periodic interactions, we propose a metrology protocol that can circumvent all the three mentioned obstacles and yet still make good use of time as a resource for metrology. Specifically, by mapping the dynamics of such a periodically driven spin system to that of a paradigm of quantum chaos but tuned to some high-order quantum resonance, it is shown that a simple $SU(2)$ coherent state can, after evolving to highly entangled states in the ensuing dynamics, be dynamically brought back to the same initial coherent state. The associated quantum Fisher information is found to exhibit quadratic scaling with both the number of spins and the duration of the metrology protocol. The achieved Heisenberg scaling can also largely survive in the presence of Markovian noise. Representing a previously unknown strategy for quantum metrology, the protocol proposed here can be tested on available experimental platforms.

quant-ph↗

arXiv2024

4DStyleGaussian: Zero-shot 4D Style Transfer with Gaussian Splatting

3D neural style transfer has gained significant attention for its potential to provide user-friendly stylization with spatial consistency. However, existing 3D style transfer methods often fall short in terms of inference efficiency, generalization ability, and struggle to handle dynamic scenes with temporal consistency. In this paper, we introduce 4DStyleGaussian, a novel 4D style transfer framework designed to achieve real-time stylization of arbitrary style references while maintaining reasonable content affinity, multi-view consistency, and temporal coherence. Our approach leverages an embedded 4D Gaussian Splatting technique, which is trained using a reversible neural network for reducing content loss in the feature distillation process. Utilizing the 4D embedded Gaussians, we predict a 4D style transformation matrix that facilitates spatially and temporally consistent style transfer with Gaussian Splatting. Experiments demonstrate that our method can achieve high-quality and zero-shot stylization for 4D scenarios with enhanced efficiency and spatial-temporal consistency.

cs.CV↗

arXiv2024

ControLRM: Fast and Controllable 3D Generation via Large Reconstruction Model

Despite recent advancements in 3D generation methods, achieving controllability still remains a challenging issue. Current approaches utilizing score-distillation sampling are hindered by laborious procedures that consume a significant amount of time. Furthermore, the process of first generating 2D representations and then mapping them to 3D lacks internal alignment between the two forms of representation. To address these challenges, we introduce ControLRM, an end-to-end feed-forward model designed for rapid and controllable 3D generation using a large reconstruction model (LRM). ControLRM comprises a 2D condition generator, a condition encoding transformer, and a triplane decoder transformer. Instead of training our model from scratch, we advocate for a joint training framework. In the condition training branch, we lock the triplane decoder and reuses the deep and robust encoding layers pretrained with millions of 3D data in LRM. In the image training branch, we unlock the triplane decoder to establish an implicit alignment between the 2D and 3D representations. To ensure unbiased evaluation, we curate evaluation samples from three distinct datasets (G-OBJ, GSO, ABO) rather than relying on cherry-picking manual generation. The comprehensive experiments conducted on quantitative and qualitative comparisons of 3D controllability and generation quality demonstrate the strong generalization capacity of our proposed approach.

cs.CV↗

arXiv2024

RobustMVS: Single Domain Generalized Deep Multi-view Stereo

Despite the impressive performance of Multi-view Stereo (MVS) approaches given plenty of training samples, the performance degradation when generalizing to unseen domains has not been clearly explored yet. In this work, we focus on the domain generalization problem in MVS. To evaluate the generalization results, we build a novel MVS domain generalization benchmark including synthetic and real-world datasets. In contrast to conventional domain generalization benchmarks, we consider a more realistic but challenging scenario, where only one source domain is available for training. The MVS problem can be analogized back to the feature matching task, and maintaining robust feature consistency among views is an important factor for improving generalization performance. To address the domain generalization problem in MVS, we propose a novel MVS framework, namely RobustMVS. A DepthClustering-guided Whitening (DCW) loss is further introduced to preserve the feature consistency among different views, which decorrelates multi-view features from viewpoint-specific style information based on geometric priors from depth maps. The experimental results further show that our method achieves superior performance on the domain generalization benchmark.

cs.CV↗

arXiv2024

Describing the critical behavior of the Anderson transition in infinite dimension by random-matrix ensembles: logarithmic multifractality and critical localization

Due to their analytical tractability, random matrix ensembles serve as robust platforms for exploring exotic phenomena in systems that are computationally demanding. Building on a companion letter [arXiv:2312.17481], this paper investigates two random matrix ensembles tailored to capture the critical behavior of the Anderson transition in infinite dimension, employing both analytical techniques and extensive numerical simulations. Our study unveils two types of critical behaviors: logarithmic multifractality and critical localization. In contrast to conventional multifractality, the novel logarithmic multifractality features eigenstate moments scaling algebraically with the logarithm of the system size. Critical localization, characterized by eigenstate moments of order $q>1/2$ converging to a finite value indicating localization, exhibits characteristic logarithmic finite-size or time effects, consistent with the critical behavior observed in random regular and Erdös-Rényi graphs of effective infinite dimensionality. Using perturbative methods, we establish the existence of logarithmic multifractality and critical localization in our models. Furthermore, we explore the emergence of novel scaling behaviors in the time dynamics and spatial correlation functions. Our models provide a valuable framework for studying infinite-dimensional quantum disordered systems, and the universality of our findings enables broad applicability to systems with pronounced finite-size effects and slow dynamics, including the contentious many-body localization transition, akin to the Anderson transition in infinite dimension.

cond-mat.dis-nn↗

arXiv2024

StyleDyRF: Zero-shot 4D Style Transfer for Dynamic Neural Radiance Fields

4D style transfer aims at transferring arbitrary visual style to the synthesized novel views of a dynamic 4D scene with varying viewpoints and times. Existing efforts on 3D style transfer can effectively combine the visual features of style images and neural radiance fields (NeRF) but fail to handle the 4D dynamic scenes limited by the static scene assumption. Consequently, we aim to handle the novel challenging problem of 4D style transfer for the first time, which further requires the consistency of stylized results on dynamic objects. In this paper, we introduce StyleDyRF, a method that represents the 4D feature space by deforming a canonical feature volume and learns a linear style transformation matrix on the feature volume in a data-driven fashion. To obtain the canonical feature volume, the rays at each time step are deformed with the geometric prior of a pre-trained dynamic NeRF to render the feature map under the supervision of pre-trained visual encoders. With the content and style cues in the canonical feature volume and the style image, we can learn the style transformation matrix from their covariance matrices with lightweight neural networks. The learned style transformation matrix can reflect a direct matching of feature covariance from the content volume to the given style pattern, in analogy with the optimization of the Gram matrix in traditional 2D neural style transfer. The experimental results show that our method not only renders 4D photorealistic style transfer results in a zero-shot manner but also outperforms existing methods in terms of visual quality and consistency.

cs.CV↗

arXiv2024

PointDC:Unsupervised Semantic Segmentation of 3D Point Clouds via Cross-modal Distillation and Super-Voxel Clustering

Semantic segmentation of point clouds usually requires exhausting efforts of human annotations, hence it attracts wide attention to the challenging topic of learning from unlabeled or weaker forms of annotations. In this paper, we take the first attempt for fully unsupervised semantic segmentation of point clouds, which aims to delineate semantically meaningful objects without any form of annotations. Previous works of unsupervised pipeline on 2D images fails in this task of point clouds, due to: 1) Clustering Ambiguity caused by limited magnitude of data and imbalanced class distribution; 2) Irregularity Ambiguity caused by the irregular sparsity of point cloud. Therefore, we propose a novel framework, PointDC, which is comprised of two steps that handle the aforementioned problems respectively: Cross-Modal Distillation (CMD) and Super-Voxel Clustering (SVC). In the first stage of CMD, multi-view visual features are back-projected to the 3D space and aggregated to a unified point feature to distill the training of the point representation. In the second stage of SVC, the point features are aggregated to super-voxels and then fed to the iterative clustering process for excavating semantic classes. PointDC yields a significant improvement over the prior state-of-the-art unsupervised methods, on both the ScanNet-v2 (+18.4 mIoU) and S3DIS (+11.5 mIoU) semantic segmentation benchmarks.

cs.CV↗

arXiv2023

Quantum logarithmic multifractality

Through a combination of rigorous analytical derivations and extensive numerical simulations, this work reports an exotic multifractal behavior, dubbed "logarithmic multifractality", in effectively infinite-dimensional systems undergoing the Anderson transition. In marked contrast to conventional multifractal critical properties observed at finite-dimensional Anderson transitions or scale-invariant second-order phase transitions, in the presence of logarithmic multifractality, eigenstate statistics, spatial correlations, and wave packet dynamics can all exhibit scaling laws which are algebraic in the logarithm of system size or time. Our findings offer crucial insights into strong finite-size effects and slow dynamics in complex systems undergoing the Anderson transition, such as the many-body localization transition.

cond-mat.dis-nn↗

arXiv2023

FaceChain: A Playground for Human-centric Artificial Intelligence Generated Content

Recent advancement in personalized image generation have unveiled the intriguing capability of pre-trained text-to-image models on learning identity information from a collection of portrait images. However, existing solutions are vulnerable in producing truthful details, and usually suffer from several defects such as (i) The generated face exhibit its own unique characteristics, \ie facial shape and facial feature positioning may not resemble key characteristics of the input, and (ii) The synthesized face may contain warped, blurred or corrupted regions. In this paper, we present FaceChain, a personalized portrait generation framework that combines a series of customized image-generation model and a rich set of face-related perceptual understanding models (\eg, face detection, deep face embedding extraction, and facial attribute recognition), to tackle aforementioned challenges and to generate truthful personalized portraits, with only a handful of portrait images as input. Concretely, we inject several SOTA face models into the generation procedure, achieving a more efficient label-tagging, data-processing, and model post-processing compared to previous solutions, such as DreamBooth ~\cite{ruiz2023dreambooth} , InstantBooth ~\cite{shi2023instantbooth} , or other LoRA-only approaches ~\cite{hu2021lora} . Besides, based on FaceChain, we further develop several applications to build a broader playground for better showing its value, including virtual try on and 2D talking head. We hope it can grow to serve the burgeoning needs from the communities. Note that this is an ongoing work that will be consistently refined and improved upon. FaceChain is open-sourced under Apache-2.0 license at \url{https://github.com/modelscope/facechain}.

cs.CV↗

arXiv2023

Semi-supervised Deep Multi-view Stereo

Significant progress has been witnessed in learning-based Multi-view Stereo (MVS) under supervised and unsupervised settings. To combine their respective merits in accuracy and completeness, meantime reducing the demand for expensive labeled data, this paper explores the problem of learning-based MVS in a semi-supervised setting that only a tiny part of the MVS data is attached with dense depth ground truth. However, due to huge variation of scenarios and flexible settings in views, it may break the basic assumption in classic semi-supervised learning, that unlabeled data and labeled data share the same label space and data distribution, named as semi-supervised distribution-gap ambiguity in the MVS problem. To handle these issues, we propose a novel semi-supervised distribution-augmented MVS framework, namely SDA-MVS. For the simple case that the basic assumption works in MVS data, consistency regularization encourages the model predictions to be consistent between original sample and randomly augmented sample. For further troublesome case that the basic assumption is conflicted in MVS data, we propose a novel style consistency loss to alleviate the negative effect caused by the distribution gap. The visual style of unlabeled sample is transferred to labeled sample to shrink the gap, and the model prediction of generated sample is further supervised with the label in original labeled sample. The experimental results in semi-supervised settings of multiple MVS datasets show the superior performance of the proposed method. With the same settings in backbone network, our proposed SDA-MVS outperforms its fully-supervised and unsupervised baselines.

cs.CV↗

arXiv2023

Critical dynamics of long-range quantum disordered systems

Long-range hoppings in quantum disordered systems are known to yield quantum multifractality, whose features can go beyond the characteristic properties associated with an Anderson transition. Indeed, critical dynamics of long-range quantum systems can exhibit anomalous dynamical behaviours distinct from those at the Anderson transition in finite dimensions. In this paper, we propose a phenomenological model of wave packet expansion in long-range hopping systems. We consider both their multifractal properties and the algebraic fat tails induced by the long-range hoppings. Using this model, we analytically derive the dynamics of moments and Inverse Participation Ratios of the time-evolving wave packets, in connection with the multifractal dimension of the system. To validate our predictions, we perform numerical simulations of a Floquet model that is analogous to the power law random banded matrix ensemble. Unlike the Anderson transition in finite dimensions, the dynamics of such systems cannot be adequately described by a single parameter scaling law that solely depends on time. Instead, it becomes crucial to establish scaling laws involving both the finite-size and the time. Explicit scaling laws for the observables under consideration are presented. Our findings are of considerable interest towards applications in the fields of many-body localization and Anderson localization on random graphs, where long-range effects arise due to the inherent topology of the Hilbert space.

cond-mat.dis-nn↗

arXiv2023

CostFormer:Cost Transformer for Cost Aggregation in Multi-view Stereo

The core of Multi-view Stereo(MVS) is the matching process among reference and source pixels. Cost aggregation plays a significant role in this process, while previous methods focus on handling it via CNNs. This may inherit the natural limitation of CNNs that fail to discriminate repetitive or incorrect matches due to limited local receptive fields. To handle the issue, we aim to involve Transformer into cost aggregation. However, another problem may occur due to the quadratically growing computational complexity caused by Transformer, resulting in memory overflow and inference latency. In this paper, we overcome these limits with an efficient Transformer-based cost aggregation network, namely CostFormer. The Residual Depth-Aware Cost Transformer(RDACT) is proposed to aggregate long-range features on cost volume via self-attention mechanisms along the depth and spatial dimensions. Furthermore, Residual Regression Transformer(RRT) is proposed to enhance spatial attention. The proposed method is a universal plug-in to improve learning-based MVS methods.

cs.CV↗