arXiv subjects

Meng Wang

Publications and source records attributed to Meng Wang.

At least 271 records · Page 15Linked to original sources

arXiv2025

Revisiting Audio-Visual Segmentation with Vision-Centric Transformer

Audio-Visual Segmentation (AVS) aims to segment sound-producing objects in video frames based on the associated audio signal. Prevailing AVS methods typically adopt an audio-centric Transformer architecture, where object queries are derived from audio features. However, audio-centric Transformers suffer from two limitations: perception ambiguity caused by the mixed nature of audio, and weakened dense prediction ability due to visual detail loss. To address these limitations, we propose a new Vision-Centric Transformer (VCT) framework that leverages vision-derived queries to iteratively fetch corresponding audio and visual information, enabling queries to better distinguish between different sounding objects from mixed audio and accurately delineate their contours. Additionally, we also introduce a Prototype Prompted Query Generation (PPQG) module within our VCT framework to generate vision-derived queries that are both semantically aware and visually rich through audio prototype prompting and pixel context grouping, facilitating audio-visual information aggregation. Extensive experiments demonstrate that our VCT framework achieves new state-of-the-art performances on three subsets of the AVSBench dataset. The code is available at https://github.com/spyflying/VCT_AVS.

cs.CV↗

arXiv2025

Mettle: Meta-Token Learning for Memory-Efficient Audio-Visual Adaptation

We present \textbf{Met}a-\textbf{T}oken \textbf{Le}arning (Mettle), a simple and memory-efficient method for adapting large-scale pretrained transformer models to downstream audio-visual tasks. Instead of sequentially modifying the output feature distribution of the transformer backbone, Mettle utilizes a lightweight \textit{Layer-Centric Distillation (LCD)} module to distill in parallel the intact audio or visual features embedded by each transformer layer into compact meta-tokens. This distillation process considers both pretrained knowledge preservation and task-specific adaptation. The obtained meta-tokens can be directly applied to classification tasks, such as audio-visual event localization and audio-visual video parsing. To further support fine-grained segmentation tasks, such as audio-visual segmentation, we introduce a \textit{Meta-Token Injection (MTI)} module, which utilizes the audio and visual meta-tokens distilled from the top transformer layer to guide feature adaptation in earlier layers. Extensive experiments on multiple audiovisual benchmarks demonstrate that our method significantly reduces memory usage and training time while maintaining parameter efficiency and competitive accuracy.

cs.CV↗

arXiv2025

MUPA: Towards Multi-Path Agentic Reasoning for Grounded Video Question Answering

Grounded Video Question Answering (Grounded VideoQA) requires aligning textual answers with explicit visual evidence. However, modern multimodal models often rely on linguistic priors and spurious correlations, resulting in poorly grounded predictions. In this work, we propose MUPA, a cooperative MUlti-Path Agentic approach that unifies video grounding, question answering, answer reflection and aggregation to tackle Grounded VideoQA. MUPA features three distinct reasoning paths on the interplay of grounding and QA agents in different chronological orders, along with a dedicated reflection agent to judge and aggregate the multi-path results to accomplish consistent QA and grounding. This design markedly improves grounding fidelity without sacrificing answer accuracy. Despite using only 2B parameters, our method outperforms all 7B-scale competitors. When scaled to 7B parameters, MUPA establishes new state-of-the-art results, with Acc@GQA of 30.3% and 47.4% on NExT-GQA and DeVE-QA respectively, demonstrating MUPA' effectiveness towards trustworthy video-language understanding. Our code is available in https://github.com/longmalongma/MUPA.

cs.CV↗

arXiv2025

Unveiling the multiband metallic nature of the normal state in nickelate La3Ni2O7

The discovery of unconventional superconductivity around 80 K in perovskite nickelates under high pressure has furnished a new platform to explore high-temperature unconventional superconductivity in addition to cuprates. Understanding the normal state of nickelate superconductors is crucial to uncovering the origin of this unconventional superconductivity and gaining further insight into its underlying mechanism. In this study, we systemically studied the transport properties of La3Ni2O7 by tuning the pressure under high magnetic fields. Magnetoresistance (MR) consistently exhibits a quasi-quadratic dependence on the magnetic field across all measured pressures and temperatures. Increased pressure enhances the metallicity of the system and leads to a monotonic increase in MR, which follows the extended Kohler's rule. These results suggest that the normal state of La3Ni2O7 to be a multiband metallic nature.

cond-mat.supr-con↗

arXiv2025

Merging Smarter, Generalizing Better: Enhancing Model Merging on OOD Data

Multi-task learning (MTL) concurrently trains a model on diverse task datasets to exploit common features, thereby improving overall performance across the tasks. Recent studies have dedicated efforts to merging multiple independent model parameters into a unified model for MTL, thus circumventing the need for training data and expanding the scope of applicable scenarios of MTL. However, current approaches to model merging predominantly concentrate on enhancing performance within in-domain (ID) datasets, often overlooking their efficacy on out-of-domain (OOD) datasets. In this work, we proposed LwPTV (Layer-wise Pruning Task Vector) by building a saliency score, measuring the redundancy of parameters in task vectors. Designed in this way ours can achieve mask vector for each task and thus perform layer-wise pruning on the task vectors, only keeping the pre-trained model parameters at the corresponding layer in merged model. Owing to its flexibility, our method can be seamlessly integrated with most of existing model merging methods to improve their performance on OOD tasks. Extensive experiments demonstrate that the application of our method results in substantial enhancements in OOD performance while preserving the ability on ID tasks.

cs.LG↗

arXiv2025

Reinforcing Video Reasoning with Focused Thinking

Recent advancements in reinforcement learning, particularly through Group Relative Policy Optimization (GRPO), have significantly improved multimodal large language models for complex reasoning tasks. However, two critical limitations persist: 1) they often produce unfocused, verbose reasoning chains that obscure salient spatiotemporal cues and 2) binary rewarding fails to account for partially correct answers, resulting in high reward variance and inefficient learning. In this paper, we propose TW-GRPO, a novel framework that enhances visual reasoning with focused thinking and dense reward granularity. Specifically, we employs a token weighting mechanism that prioritizes tokens with high informational density (estimated by intra-group information entropy), suppressing redundant tokens like generic reasoning prefixes. Furthermore, we reformulate RL training by shifting from single-choice to multi-choice QA tasks, where soft rewards enable finer-grained gradient estimation by distinguishing partial correctness. Additionally, we propose question-answer inversion, a data augmentation strategy to generate diverse multi-choice samples from existing benchmarks. Experiments demonstrate state-of-the-art performance on several video reasoning and general understanding benchmarks. Notably, TW-GRPO achieves 50.4\% accuracy on CLEVRER (18.8\% improvement over Video-R1) and 65.8\% on MMVU. Our codes are available at \href{https://github.com/longmalongma/TW-GRPO}.

cs.CV↗

arXiv2025

SSAM: Self-Supervised Association Modeling for Test-Time Adaption

Test-time adaption (TTA) has witnessed important progress in recent years, the prevailing methods typically first encode the image and the text and design strategies to model the association between them. Meanwhile, the image encoder is usually frozen due to the absence of explicit supervision in TTA scenarios. We identify a critical limitation in this paradigm: While test-time images often exhibit distribution shifts from training data, existing methods persistently freeze the image encoder due to the absence of explicit supervision during adaptation. This practice overlooks the image encoder's crucial role in bridging distribution shift between training and test. To address this challenge, we propose SSAM (Self-Supervised Association Modeling), a new TTA framework that enables dynamic encoder refinement through dual-phase association learning. Our method operates via two synergistic components: 1) Soft Prototype Estimation (SPE), which estimates probabilistic category associations to guide feature space reorganization, and 2) Prototype-anchored Image Reconstruction (PIR), enforcing encoder stability through cluster-conditional image feature reconstruction. Comprehensive experiments across diverse baseline methods and benchmarks demonstrate that SSAM can surpass state-of-the-art TTA baselines by a clear margin while maintaining computational efficiency. The framework's architecture-agnostic design and minimal hyperparameter dependence further enhance its practical applicability.

cs.CV↗

arXiv2025

Adversarially Robust AI-Generated Image Detection for Free: An Information Theoretic Perspective

Rapid advances in Artificial Intelligence Generated Images (AIGI) have facilitated malicious use, such as forgery and misinformation. Therefore, numerous methods have been proposed to detect fake images. Although such detectors have been proven to be universally vulnerable to adversarial attacks, defenses in this field are scarce. In this paper, we first identify that adversarial training (AT), widely regarded as the most effective defense, suffers from performance collapse in AIGI detection. Through an information-theoretic lens, we further attribute the cause of collapse to feature entanglement, which disrupts the preservation of feature-label mutual information. Instead, standard detectors show clear feature separation. Motivated by this difference, we propose Training-free Robust Detection via Information-theoretic Measures (TRIM), the first training-free adversarial defense for AIGI detection. TRIM builds on standard detectors and quantifies feature shifts using prediction entropy and KL divergence. Extensive experiments across multiple datasets and attacks validate the superiority of our TRIM, e.g., outperforming the state-of-the-art defense by 33.88% (28.91%) on ProGAN (GenImage), while well maintaining original accuracy.

cs.CV↗

arXiv2025

Calibrating Undisciplined Over-Smoothing in Transformer for Weakly Supervised Semantic Segmentation

Weakly supervised semantic segmentation (WSSS) has recently attracted considerable attention because it requires fewer annotations than fully supervised approaches, making it especially promising for large-scale image segmentation tasks. Although many vision transformer-based methods leverage self-attention affinity matrices to refine Class Activation Maps (CAMs), they often treat each layer's affinity equally and thus introduce considerable background noise at deeper layers, where attention tends to converge excessively on certain tokens (i.e., over-smoothing). We observe that this deep-level attention naturally converges on a subset of tokens, yet unregulated query-key affinity can generate unpredictable activation patterns (undisciplined over-smoothing), adversely affecting CAM accuracy. To address these limitations, we propose an Adaptive Re-Activation Mechanism (AReAM), which exploits shallow-level affinity to guide deeper-layer convergence in an entropy-aware manner, thereby suppressing background noise and re-activating crucial semantic regions in the CAMs. Experiments on two commonly used datasets demonstrate that AReAM substantially improves segmentation performance compared with existing WSSS methods, reducing noise while sharpening focus on relevant semantic regions. Overall, this work underscores the importance of controlling deep-level attention to mitigate undisciplined over-smoothing, introduces an entropy-aware mechanism that harmonizes shallow and deep-level affinities, and provides a refined approach to enhance transformer-based WSSS accuracy by re-activating CAMs.

cs.CV↗

arXiv2025

Diff-FlowFSI: A GPU-Optimized Differentiable CFD Platform for High-Fidelity Turbulence and FSI Simulations

Turbulent flows and fluid-structure interactions (FSI) are ubiquitous in scientific and engineering applications, but their accurate and efficient simulation remains a major challenge due to strong nonlinearities, multiscale interactions, and high computational demands. Traditional CFD solvers, though effective, struggle with scalability and adaptability for tasks such as inverse modeling, optimization, and data assimilation. Recent advances in machine learning (ML) have inspired hybrid modeling approaches that integrate neural networks with physics-based solvers to enhance generalization and capture unresolved dynamics. However, realizing this integration requires solvers that are not only physically accurate but also differentiable and GPU-efficient. In this work, we introduce Diff-FlowFSI, a GPU-accelerated, fully differentiable CFD platform designed for high-fidelity turbulence and FSI simulations. Implemented in JAX, Diff-FlowFSI features a vectorized finite volume solver combined with the immersed boundary method to handle complex geometries and fluid-structure coupling. The platform enables GPU-enabled fast forward simulations, supports automatic differentiation for gradient-based inverse problems, and integrates seamlessly with deep learning components for hybrid neural-CFD modeling. We validate Diff-FlowFSI across a series of benchmark turbulence and FSI problems, demonstrating its capability to accelerate scientific computing at the intersection of physics and machine learning.

physics.flu-dyn↗

arXiv2025

Training Nonlinear Transformers for Chain-of-Thought Inference: A Theoretical Generalization Analysis

Chain-of-Thought (CoT) is an efficient prompting method that enables the reasoning ability of large language models by augmenting the query using multiple examples with multiple intermediate steps. Despite the empirical success, the theoretical understanding of how to train a Transformer to achieve the CoT ability remains less explored. This is primarily due to the technical challenges involved in analyzing the nonconvex optimization on nonlinear attention models. To the best of our knowledge, this work provides the first theoretical study of training Transformers with nonlinear attention to obtain the CoT generalization capability so that the resulting model can inference on unseen tasks when the input is augmented by examples of the new task. We first quantify the required training samples and iterations to train a Transformer model towards CoT ability. We then prove the success of its CoT generalization on unseen tasks with distribution-shifted testing data. Moreover, we theoretically characterize the conditions for an accurate reasoning output by CoT even when the provided reasoning examples contain noises and are not always accurate. In contrast, in-context learning (ICL), which can be viewed as one-step CoT without intermediate steps, may fail to provide an accurate output when CoT does. These theoretical findings are justified through experiments.

cs.LG↗

arXiv2025

When is Task Vector Provably Effective for Model Editing? A Generalization Analysis of Nonlinear Transformers

Task arithmetic refers to editing the pre-trained model by adding a weighted sum of task vectors, each of which is the weight update from the pre-trained model to fine-tuned models for certain tasks. This approach recently gained attention as a computationally efficient inference method for model editing, e.g., multi-task learning, forgetting, and out-of-domain generalization capabilities. However, the theoretical understanding of why task vectors can execute various conceptual operations remains limited, due to the highly non-convexity of training Transformer-based models. To the best of our knowledge, this paper provides the first theoretical characterization of the generalization guarantees of task vector methods on nonlinear Transformers. We consider a conceptual learning setting, where each task is a binary classification problem based on a discriminative pattern. We theoretically prove the effectiveness of task addition in simultaneously learning a set of irrelevant or aligned tasks, as well as the success of task negation in unlearning one task from irrelevant or contradictory tasks. Moreover, we prove the proper selection of linear coefficients for task arithmetic to achieve guaranteed generalization to out-of-domain tasks. All of our theoretical results hold for both dense-weight parameters and their low-rank approximations. Although established in a conceptual setting, our theoretical findings were validated on a practical machine unlearning task using the large language model Phi-1.5 (1.3B).

cs.LG↗

arXiv2025

Scene-Text Grounding for Text-Based Video Question Answering

Existing efforts in text-based video question answering (TextVideoQA) are criticized for their opaque decisionmaking and heavy reliance on scene-text recognition. In this paper, we propose to study Grounded TextVideoQA by forcing models to answer questions and spatio-temporally localize the relevant scene-text regions, thus decoupling QA from scenetext recognition and promoting research towards interpretable QA. The task has three-fold significance. First, it encourages scene-text evidence versus other short-cuts for answer predictions. Second, it directly accepts scene-text regions as visual answers, thus circumventing the problem of ineffective answer evaluation by stringent string matching. Third, it isolates the challenges inherited in VideoQA and scene-text recognition. This enables the diagnosis of the root causes for failure predictions, e.g., wrong QA or wrong scene-text recognition? To achieve Grounded TextVideoQA, we propose the T2S-QA model that highlights a disentangled temporal-to-spatial contrastive learning strategy for weakly-supervised scene-text grounding and grounded TextVideoQA. To facilitate evaluation, we construct a new dataset ViTXT-GQA which features 52K scene-text bounding boxes within 2.2K temporal segments related to 2K questions and 729 videos. With ViTXT-GQA, we perform extensive experiments and demonstrate the severe limitations of existing techniques in Grounded TextVideoQA. While T2S-QA achieves superior results, the large performance gap with human leaves ample space for improvement. Our further analysis of oracle scene-text inputs posits that the major challenge is scene-text recognition. To advance the research of Grounded TextVideoQA, our dataset and code are at https://github.com/zhousheng97/ViTXT-GQA.git

cs.CV↗

arXiv2025

AdaptCLIP: Adapting CLIP for Universal Visual Anomaly Detection

Universal visual anomaly detection aims to identify anomalies from novel or unseen vision domains without additional fine-tuning, which is critical in open scenarios. Recent studies have demonstrated that pre-trained vision-language models like CLIP exhibit strong generalization with just zero or a few normal images. However, existing methods struggle with designing prompt templates, complex token interactions, or requiring additional fine-tuning, resulting in limited flexibility. In this work, we present a simple yet effective method called AdaptCLIP based on two key insights. First, adaptive visual and textual representations should be learned alternately rather than jointly. Second, comparative learning between query and normal image prompt should incorporate both contextual and aligned residual features, rather than relying solely on residual features. AdaptCLIP treats CLIP models as a foundational service, adding only three simple adapters, visual adapter, textual adapter, and prompt-query adapter, at its input or output ends. AdaptCLIP supports zero-/few-shot generalization across domains and possesses a training-free manner on target domains once trained on a base dataset. AdaptCLIP achieves state-of-the-art performance on 12 anomaly detection benchmarks from industrial and medical domains, significantly outperforming existing competitive methods. We will make the code and model of AdaptCLIP available at https://github.com/gaobb/AdaptCLIP.

cs.CV↗

arXiv2025

Accelerating Simulation of Quantum Circuits under Noise via Computational Reuse

To realize the full potential of quantum computers, we must mitigate qubit errors by developing noise-aware algorithms, compilers, and architectures. Thus, simulating quantum programs on high-performance computing (HPC) systems with different noise models is a de facto tool researchers use. Unfortunately, noisy simulators iteratively execute a similar circuit for thousands of trials, thereby incurring significant performance overheads. To address this, we propose a noisy simulation technique called Tree-Based Quantum Circuit Simulation (TQSim). TQSim exploits the reusability of intermediate results during the noisy simulation, reducing computation. TQSim dynamically partitions a circuit into several subcircuits. It then reuses the intermediate results from these subcircuits during computation. Compared to a noisy Qulacs-based baseline simulator, TQSim achieves a speedup of up to 3.89x for noisy simulations. TQSim is designed to be efficient with multi-node setups while also maintaining tight fidelity bounds.

quant-ph↗

arXiv2025

Measurement of two-neutrino double electron capture half-life of $^{124}$Xe with PandaX-4T

Detailed studies of two-neutrino double electron capture (2$ν$DEC) is a crucial step towards searching for the neutrino-less mode to explore the Majorana nature of neutrinos. We have measured precisely the half-life of the 2$ν$DEC process in $^{124}$Xe, utilizing a total exposure of 1.73 tonne$\cdot$year from the commissioning run and the first science run of the PandaX-4T experiment. A time-dependent background model in the $\mathcal{O}$(10 keV) energy is constructed for the first time in PandaX-4T data. With an unbinned maximum likelihood fit, we determine the half-life of the 2$ν$DEC process to be $(1.03\pm0.15_{\rm stat}\pm0.08_{\rm sys})\times 10^{22}$$\,$yr. Furthermore, we have evaluated the branching ratio for both electrons captured from the $K$ shell ($KK$) to be $(65\pm5)\%$, which aligns with the $^{124}$Xe nuclear model calculations within 1.8$\,$$σ$.

nucl-ex↗

arXiv2025

Search for Solar Boosted Dark Matter Particles at the PandaX-4T Experiment

We present a novel constraint on light dark matter utilizing $1.54$ tonne$\cdot$year of data acquired from the PandaX-4T dual-phase xenon time projection chamber. This constraint is derived through detecting electronic recoil signals resulting from the interaction with solar-enhanced dark matter flux. Low-mass dark matter particles, lighter than a few MeV/$c^2$, can scatter with the thermal electrons in the Sun. Consequently, with higher kinetic energy, the boosted dark matter component becomes detectable via contact scattering with xenon electrons, resulting in a few keV energy deposition that exceeds the threshold of PandaX-4T. We calculate the expected recoil energy in PandaX-4T considering the Sun's acceleration and the detection capabilities of the xenon detector. The first experimental search results using the xenon detector yield the most stringent cross-section of $3.51 \times 10^{-39}~\mathrm{cm}^2$ at $0.08~\mathrm{MeV}$/$c^2$ for a solar boosted dark matter mass ranging from $0.02$ to $10~ \mathrm{MeV}$/$c^2$, achieving a 23 fold improvement compared with earlier experimental studies.

hep-ex↗

arXiv2025

PIDiff: Image Customization for Personalized Identities with Diffusion Models

Text-to-image generation for personalized identities aims at incorporating the specific identity into images using a text prompt and an identity image. Based on the powerful generative capabilities of DDPMs, many previous works adopt additional prompts, such as text embeddings and CLIP image embeddings, to represent the identity information, while they fail to disentangle the identity information and background information. As a result, the generated images not only lose key identity characteristics but also suffer from significantly reduced diversity. To address this issue, previous works have combined the W+ space from StyleGAN with diffusion models, leveraging this space to provide a more accurate and comprehensive representation of identity features through multi-level feature extraction. However, the entanglement of identity and background information in in-the-wild images during training prevents accurate identity localization, resulting in severe semantic interference between identity and background. In this paper, we propose a novel fine-tuning-based diffusion model for personalized identities text-to-image generation, named PIDiff, which leverages the W+ space and an identity-tailored fine-tuning strategy to avoid semantic entanglement and achieves accurate feature extraction and localization. Style editing can also be achieved by PIDiff through preserving the characteristics of identity features in the W+ space, which vary from coarse to fine. Through the combination of the proposed cross-attention block and parameter optimization strategy, PIDiff preserves the identity information and maintains the generation capability for in-the-wild images of the pre-trained model during inference. Our experimental results validate the effectiveness of our method in this task.

cs.CV↗