arXiv subjects

Wei Zhao

Publications and source records attributed to Wei Zhao.

At least 127 records · Page 7Linked to original sources

arXiv2025

Date Fragments: A Hidden Bottleneck of Tokenization for Temporal Reasoning

Modern BPE tokenizers often split calendar dates into meaningless fragments, e.g., 20250312 $\rightarrow$ 202, 503, 12, inflating token counts and obscuring the inherent structure needed for robust temporal reasoning. In this work, we (1) introduce a simple yet interpretable metric, termed date fragmentation ratio, that measures how faithfully a tokenizer preserves multi-digit date components; (2) release DateAugBench, a suite of 6500 examples spanning three temporal reasoning tasks: context-based date resolution, format-invariance puzzles, and date arithmetic across historical, contemporary, and future time periods; and (3) through layer-wise probing and causal attention-hop analyses, uncover an emergent date-abstraction mechanism whereby large language models stitch together the fragments of month, day, and year components for temporal reasoning. Our experiments show that excessive fragmentation correlates with accuracy drops of up to 10 points on uncommon dates like historical and futuristic dates. Further, we find that the larger the model, the faster the emergent date abstraction that heals date fragments is accomplished. Lastly, we observe a reasoning path that LLMs follow to assemble date fragments, typically differing from human interpretation (year $\rightarrow$ month $\rightarrow$ day). Our datasets and code are made publicly available \href{https://github.com/gagan3012/date-fragments}{here}.

cs.CL↗

arXiv2025

VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Model

Vision-Language-Action (VLA) models typically bridge the gap between perceptual and action spaces by pre-training a large-scale Vision-Language Model (VLM) on robotic data. While this approach greatly enhances performance, it also incurs significant training costs. In this paper, we investigate how to effectively bridge vision-language (VL) representations to action (A). We introduce VLA-Adapter, a novel paradigm designed to reduce the reliance of VLA models on large-scale VLMs and extensive pre-training. To this end, we first systematically analyze the effectiveness of various VL conditions and present key findings on which conditions are essential for bridging perception and action spaces. Based on these insights, we propose a lightweight Policy module with Bridge Attention, which autonomously injects the optimal condition into the action space. In this way, our method achieves high performance using only a 0.5B-parameter backbone, without any robotic data pre-training. Extensive experiments on both simulated and real-world robotic benchmarks demonstrate that VLA-Adapter not only achieves state-of-the-art level performance, but also offers the fast inference speed reported to date. Furthermore, thanks to the proposed advanced bridging paradigm, VLA-Adapter enables the training of a powerful VLA model in just 8 hours on a single consumer-grade GPU, greatly lowering the barrier to deploying the VLA model. Project page: https://vla-adapter.github.io/.

cs.RO↗

arXiv2025

BAPFL: Exploring Backdoor Attacks Against Prototype-based Federated Learning

Prototype-based federated learning (PFL) has emerged as a promising paradigm to address data heterogeneity problems in federated learning, as it leverages mean feature vectors as prototypes to enhance model generalization. However, its robustness against backdoor attacks remains largely unexplored. In this paper, we identify that PFL is inherently resistant to existing backdoor attacks due to its unique prototype learning mechanism and local data heterogeneity. To further explore the security of PFL, we propose BAPFL, the first backdoor attack method specifically designed for PFL frameworks. BAPFL integrates a prototype poisoning strategy with a trigger optimization mechanism. The prototype poisoning strategy manipulates the trajectories of global prototypes to mislead the prototype training of benign clients, pushing their local prototypes of clean samples away from the prototypes of trigger-embedded samples. Meanwhile, the trigger optimization mechanism learns a unique and stealthy trigger for each potential target label, and guides the prototypes of trigger-embedded samples to align closely with the global prototype of the target label. Experimental results across multiple datasets and PFL variants demonstrate that BAPFL achieves a 35\%-75\% improvement in attack success rate compared to traditional backdoor attacks, while preserving main task accuracy. These results highlight the effectiveness, stealthiness, and adaptability of BAPFL in PFL.

cs.LG↗

arXiv2025

Probing new light scalars with the lepton anomalous magnetic moment and the weak equivalence principle violation

A new scalar particle with generic couplings to the standard-model particles is a possible source for the lepton anomalous magnetic moment and the violation of the weak equivalence principle. Here, one-loop contributions to the lepton anomalous magnetic moment, involving the scalar-photon and scalar-lepton couplings, are calculated. Then, employing the recent experimental results of the electron anomalous magnetic moment, the muon anomalous magnetic moment, and the MICROSCOPE mission, we find the improved constraints on scalar-lepton and scalar-photon couplings: $|λ_e|\leq 6.0 \times 10^{-6}$, $|λ_μ|\leq 3.5\times 10^{-4}$, and $|λ_γ|\leq 4.5 \times 10^{-13}$ ${\rm eV^{-1}}$ for scalar mass below $10^4$ eV. We find that the naive scaling relationship between the scalar-muon coupling and the scalar-electron coupling is favored by three experimental results. Furthermore, the minimal standard-model extension by one scalar is also favored by all three experiments, and the model parameter is constrained best to $|\mathcal{A}|\leq 1.7 \times 10^{-11}$ eV for $m_ϕ< 10^{-13}$ eV.

hep-ph↗

arXiv2025

Analysis on asymmetric metric measure spaces: $q$-heat flow, $q$-Laplacian and Sobolev spaces

We investigate geometric analysis on metric measure spaces equipped with asymmetric distance functions. Extending concepts from the symmetric case, we introduce upper gradients and corresponding $L^q$-energy functionals as well as $q$-Laplacian in the asymmetric setting. Along the lines of gradient flow theory, we then study $q$-heat flow as the $L^2$-gradient flow for $L^q$-energy. We also study the Sobolev spaces over asymmetric metric measure spaces, extending some results of Ambrosio--Gigli--Savaré to asymmetric distances. A wide class of asymmetric metric measure spaces is provided by irreversible Finsler manifolds, which serve to construct various model examples by pointing out genuine differences between the symmetric and asymmetric settings.

math.DG↗

arXiv2025

An Evaluation of External Magnetic Flux Error in Magnet-Moving Kibble balances

The magnet-moving measurement scheme in Kibble balances avoids displacing force-sensitive components, such as the weighing cell, and enables a broader magnetic profile measurement range during the velocity phase. However, this mechanism introduces the risk of asymmetry in the $Bl$ measurement due to external magnetic flux, leading to a potential systematic error in the final measurement results. Using the Tsinghua tabletop Kibble balance magnet as a case study, this paper investigates the error mechanism through finite element analysis and experimental investigations. An evaluation method combining external weak-field measurements with attenuation factor analysis is proposed to assess external magnetic flux errors in magnet-moving measurement schemes. The findings demonstrate that selecting an optimal weighing position can reduce the far-end flux effect to the order of $10^{-9}$. In contrast, the near-end flux effect can be quantified by monitoring the magnetic field surrounding the magnet system. In the Tsinghua Kibble balance system, we show that with proper control of external flux sources, the relative error can be reduced below $1 \times 10^{-8}$ without requiring additional magnetic shielding.

physics.ins-det↗

arXiv2025

Long-VLA: Unleashing Long-Horizon Capability of Vision Language Action Model for Robot Manipulation

Vision-Language-Action (VLA) models have become a cornerstone in robotic policy learning, leveraging large-scale multimodal data for robust and scalable control. However, existing VLA frameworks primarily address short-horizon tasks, and their effectiveness on long-horizon, multi-step robotic manipulation remains limited due to challenges in skill chaining and subtask dependencies. In this work, we introduce Long-VLA, the first end-to-end VLA model specifically designed for long-horizon robotic tasks. Our approach features a novel phase-aware input masking strategy that adaptively segments each subtask into moving and interaction phases, enabling the model to focus on phase-relevant sensory cues and enhancing subtask compatibility. This unified strategy preserves the scalability and data efficiency of VLA training, and our architecture-agnostic module can be seamlessly integrated into existing VLA models. We further propose the L-CALVIN benchmark to systematically evaluate long-horizon manipulation. Extensive experiments on both simulated and real-world tasks demonstrate that Long-VLA significantly outperforms prior state-of-the-art methods, establishing a new baseline for long-horizon robotic control.

cs.RO↗

arXiv2025

Estimating Covariate Effects on Functional Connectivity using Voxel-Level fMRI Data

Functional connectivity (FC) analysis of resting-state fMRI data provides a framework for characterizing brain networks and their association with participant-level covariates. Due to the high dimensionality of neuroimaging data, standard approaches often average signals within regions of interest (ROIs), which ignores the underlying spatiotemporal dependence among voxels and can lead to biased or inefficient inference. We propose to use a summary statistic -- the empirical voxel-wise correlations between ROIs -- and, crucially, model the complex covariance structure among these correlations through a new positive definite covariance function. Building on this foundation, we develop a computationally efficient two-step estimation procedure that enables statistical inference on covariate effects on region-level connectivity. Simulation studies show calibrated uncertainty quantification, and substantial gains in validity of the statistical inference over the standard averaging method. With data from the Autism Brain Imaging Data Exchange, we show that autism spectrum disorder is associated with altered FC between attention-related ROIs after adjusting for age and gender. The proposed framework offers an interpretable and statistically rigorous approach to estimation of covariate effects on FC suitable for large-scale neuroimaging studies.

stat.ME↗

arXiv2025

Discrepancy-Aware Graph Mask Auto-Encoder

Masked Graph Auto-Encoder, a powerful graph self-supervised training paradigm, has recently shown superior performance in graph representation learning. Existing works typically rely on node contextual information to recover the masked information. However, they fail to generalize well to heterophilic graphs where connected nodes may be not similar, because they focus only on capturing the neighborhood information and ignoring the discrepancy information between different nodes, resulting in indistinguishable node representations. In this paper, to address this issue, we propose a Discrepancy-Aware Graph Mask Auto-Encoder (DGMAE). It obtains more distinguishable node representations by reconstructing the discrepancy information of neighboring nodes during the masking process. We conduct extensive experiments on 17 widely-used benchmark datasets. The results show that our DGMAE can effectively preserve the discrepancies of nodes in low-dimensional space. Moreover, DGMAE significantly outperforms state-of-the-art graph self-supervised learning methods on three graph analytic including tasks node classification, node clustering, and graph classification, demonstrating its remarkable superiority. The code of DGMAE is available at https://github.com/zhengziyu77/DGMAE.

cs.LG↗

arXiv2025

Influence and Connectivity in Networks: A Generating Function Approach

Many widely used network centralities are based on counting walks that meet specific criteria. This paper introduces a systematic framework for walk enumeration using generating functions. We introduce a first-passage decomposition that uniquely divides any walk passing through specified nodes or links into two components: a first-reaching walk and a subsequent walk. This decomposition yields a system of interconnected equations that relate three disjoint categories of walks: unrestricted walks, walks that avoid specific elements, and walks that pass through designated sets. The framework offers a range of applications, including evaluating the effects of structural interventions, such as node or link modifications, on network walks, generalizing target centrality to multi-receiver scenarios in information networks, and comparing different strategies for adding links.

econ.TH↗

arXiv2025

CARP: Visuomotor Policy Learning via Coarse-to-Fine Autoregressive Prediction

In robotic visuomotor policy learning, diffusion-based models have achieved significant success in improving the accuracy of action trajectory generation compared to traditional autoregressive models. However, they suffer from inefficiency due to multiple denoising steps and limited flexibility from complex constraints. In this paper, we introduce Coarse-to-Fine AutoRegressive Policy (CARP), a novel paradigm for visuomotor policy learning that redefines the autoregressive action generation process as a coarse-to-fine, next-scale approach. CARP decouples action generation into two stages: first, an action autoencoder learns multi-scale representations of the entire action sequence; then, a GPT-style transformer refines the sequence prediction through a coarse-to-fine autoregressive process. This straightforward and intuitive approach produces highly accurate and smooth actions, matching or even surpassing the performance of diffusion-based policies while maintaining efficiency on par with autoregressive policies. We conduct extensive evaluations across diverse settings, including single-task and multi-task scenarios on state-based and image-based simulation benchmarks, as well as real-world tasks. CARP achieves competitive success rates, with up to a 10% improvement, and delivers 10x faster inference compared to state-of-the-art policies, establishing a high-performance, efficient, and flexible paradigm for action generation in robotic tasks.

cs.RO↗

arXiv2025

Weight Distribution of Repeated-Root Cyclic Codes with Prime Power Lengths

Determining the weight distribution of a linear code is a classical and fundamental topic in coding theory that has been extensively investigated. Repeated-root cyclic codes, which form a significant subclass of error-correcting codes, have found broad applications in quantum error-correcting codes, symbol-pair codes, and storage codes. Through polynomial derivation, we derive the monomial equivalent codes for these repeated-root cyclic codes with prime power lengths. Given that monomial equivalent codes exhibit identical weight distributions, we transform the computation of the weight distribution of these repeated-root cyclic codes into the computation of the weight distribution of their monomial equivalent codes. Leveraging the classical results on the weight distribution of MDS codes, we explicitly determine the weight distribution of these repeated-root cyclic codes. Moreover, we apply the weight distribution formula to construct a class of $p$-weight cyclic codes for any prime $p$.

cs.IT↗

arXiv2025

ProGMLP: A Progressive Framework for GNN-to-MLP Knowledge Distillation with Efficient Trade-offs

GNN-to-MLP (G2M) methods have emerged as a promising approach to accelerate Graph Neural Networks (GNNs) by distilling their knowledge into simpler Multi-Layer Perceptrons (MLPs). These methods bridge the gap between the expressive power of GNNs and the computational efficiency of MLPs, making them well-suited for resource-constrained environments. However, existing G2M methods are limited by their inability to flexibly adjust inference cost and accuracy dynamically, a critical requirement for real-world applications where computational resources and time constraints can vary significantly. To address this, we introduce a Progressive framework designed to offer flexible and on-demand trade-offs between inference cost and accuracy for GNN-to-MLP knowledge distillation (ProGMLP). ProGMLP employs a Progressive Training Structure (PTS), where multiple MLP students are trained in sequence, each building on the previous one. Furthermore, ProGMLP incorporates Progressive Knowledge Distillation (PKD) to iteratively refine the distillation process from GNNs to MLPs, and Progressive Mixup Augmentation (PMA) to enhance generalization by progressively generating harder mixed samples. Our approach is validated through comprehensive experiments on eight real-world graph datasets, demonstrating that ProGMLP maintains high accuracy while dynamically adapting to varying runtime scenarios, making it highly effective for deployment in diverse application settings.

cs.LG↗

arXiv2025

Trusted Multi-view Learning under Noisy Supervision

Multi-view learning methods often focus on improving decision accuracy while neglecting the decision uncertainty, which significantly restricts their applications in safety-critical scenarios. To address this, trusted multi-view learning methods estimate prediction uncertainties by learning class distributions from each instance. However, these methods heavily rely on high quality ground-truth labels. This motivates us to delve into a new problem: how to develop a reliable multi-view learning model under the guidance of noisy labels? We propose the Trusted Multi view Noise Refining (TMNR) method to address this challenge by modeling label noise arising from low-quality data features and easily-confused classes. TMNR employs evidential deep neural networks to construct view-specific opinions that capture both beliefs and uncertainty. These opinions are then transformed through noise correlation matrices to align with the noisy supervision, where matrix elements are constrained by sample uncertainty to reflect label reliability. Furthermore, considering the challenge of jointly optimizing the evidence network and noise correlation matrices under noisy supervision, we further propose Trusted Multi-view Noise Re-Refining (TMNR^2 ), which disentangles this complex co-training problem by establishing different training objectives for distinct modules. TMNR^2 identifies potentially mislabeled samples through evidence-label consistency and generates pseudo-labels from neighboring information. By assigning clean samples to optimize evidential networks and noisy samples to guide noise correlation matrices, respectively, TMNR^2 reduces mapping interference and achieves stabilizes training. Experimental results demonstrate that TMNR^2 significantly outperforms baseline methods, with average accuracy improvements of 7% on datasets with 50% label noise.

cs.LG↗

arXiv2025

LOVO: Efficient Complex Object Query in Large-Scale Video Datasets

The widespread deployment of cameras has led to an exponential increase in video data, creating vast opportunities for applications such as traffic management and crime surveillance. However, querying specific objects from large-scale video datasets presents challenges, including (1) processing massive and continuously growing data volumes, (2) supporting complex query requirements, and (3) ensuring low-latency execution. Existing video analysis methods struggle with either limited adaptability to unseen object classes or suffer from high query latency. In this paper, we present LOVO, a novel system designed to efficiently handle comp$\underline{L}$ex $\underline{O}$bject queries in large-scale $\underline{V}$ide$\underline{O}$ datasets. Agnostic to user queries, LOVO performs one-time feature extraction using pre-trained visual encoders, generating compact visual embeddings for key frames to build an efficient index. These visual embeddings, along with associated bounding boxes, are organized in an inverted multi-index structure within a vector database, which supports queries for any objects. During the query phase, LOVO transforms object queries to query embeddings and conducts fast approximate nearest-neighbor searches on the visual embeddings. Finally, a cross-modal rerank is performed to refine the results by fusing visual features with detailed textual features. Evaluation on real-world video datasets demonstrates that LOVO outperforms existing methods in handling complex queries, with near-optimal query accuracy and up to 85x lower search latency, while significantly reducing index construction costs. This system redefines the state-of-the-art object query approaches in video analysis, setting a new benchmark for complex object queries with a novel, scalable, and efficient approach that excels in dynamic environments.

cs.IR↗

arXiv2025

Unveiling the Complex Jet Dynamics in the Blazar 2021+317 through Multi-Epoch VLBI Observations

We present an investigation of the compact structure of the AGN 2021+317 based on multi-epoch Very Long Baseline Interferometry (VLBI) observations at 15, 22, and 43 GHz in the period from 2013 through 2024. The VLBI images show a core-jet structure extended to the south, with two stationary components in the northern region, one of which likely to be the core of the source. We also detected two new moving jet components (S4 and S5) in the observations of 2021. Based on these observational findings, we analyzed two distinctive jet models, involving one or another stationary component mentioned above as the jet core. One model assumes a moderate bulk motion velocity, a wider viewing angle, and a lower Doppler factor, with the magnetic field energy density significantly dominating over non-thermal particle energy density. The other model involves a higher bulk motion velocity, a narrower viewing angle, and a higher Doppler factor, with an even greater dominance of magnetic field energy in the core. The position angle of the jet ridge line rotates counter-clockwise over the observed period. The apparent kinematics of the jet components is more consistent with a model of the precessing jet, which has recently completed the first half of the precession cycle. Our results provide constraints on the dynamic evolution of the jet and its interaction with the surrounding medium.

astro-ph.HE↗

arXiv2025

CT-Mamba: A Hybrid Convolutional State Space Model for Low-Dose CT Denoising

Low-dose CT (LDCT) significantly reduces the radiation dose received by patients, however, dose reduction introduces additional noise and artifacts. Currently, denoising methods based on convolutional neural networks (CNNs) face limitations in long-range modeling capabilities, while Transformer-based denoising methods, although capable of powerful long-range modeling, suffer from high computational complexity. Furthermore, the denoised images predicted by deep learning-based techniques inevitably exhibit differences in noise distribution compared to normal-dose CT (NDCT) images, which can also impact the final image quality and diagnostic outcomes. This paper proposes CT-Mamba, a hybrid convolutional State Space Model for LDCT image denoising. The model combines the local feature extraction advantages of CNNs with Mamba's strength in capturing long-range dependencies, enabling it to capture both local details and global context. Additionally, we introduce an innovative spatially coherent Z-shaped scanning scheme to ensure spatial continuity between adjacent pixels in the image. We design a Mamba-driven deep noise power spectrum (NPS) loss function to guide model training, ensuring that the noise texture of the denoised LDCT images closely resembles that of NDCT images, thereby enhancing overall image quality and diagnostic value. Experimental results have demonstrated that CT-Mamba performs excellently in reducing noise in LDCT images, enhancing detail preservation, and optimizing noise texture distribution, and exhibits higher statistical similarity with the radiomics features of NDCT images. The proposed CT-Mamba demonstrates outstanding performance in LDCT denoising and holds promise as a representative approach for applying the Mamba framework to LDCT denoising tasks.

eess.IV↗

arXiv2025

Enhancing Homophily-Heterophily Separation: Relation-Aware Learning in Heterogeneous Graphs

Real-world networks usually have a property of node heterophily, that is, the connected nodes usually have different features or different labels. This heterophily issue has been extensively studied in homogeneous graphs but remains under-explored in heterogeneous graphs, where there are multiple types of nodes and edges. Capturing node heterophily in heterogeneous graphs is very challenging since both node/edge heterogeneity and node heterophily should be carefully taken into consideration. Existing methods typically convert heterogeneous graphs into homogeneous ones to learn node heterophily, which will inevitably lose the potential heterophily conveyed by heterogeneous relations. To bridge this gap, we propose Relation-Aware Separation of Homophily and Heterophily (RASH), a novel contrastive learning framework that explicitly models high-order semantics of heterogeneous interactions and adaptively separates homophilic and heterophilic patterns. Particularly, RASH introduces dual heterogeneous hypergraphs to encode multi-relational bipartite subgraphs and dynamically constructs homophilic graphs and heterophilic graphs based on relation importance. A multi-relation contrastive loss is designed to align heterogeneous and homophilic/heterophilic views by maximizing mutual information. In this way, RASH simultaneously resolves the challenges of heterogeneity and heterophily in heterogeneous graphs. Extensive experiments on benchmark datasets demonstrate the effectiveness of RASH across various downstream tasks. The code is available at: https://github.com/zhengziyu77/RASH.

cs.SI↗