arXiv subjects

Wei Wei

Publications and source records attributed to Wei Wei.

At least 181 records · Page 10Linked to original sources

arXiv2025

Low-Biased General Annotated Dataset Generation

Pre-training backbone networks on a general annotated dataset (e.g., ImageNet) that comprises numerous manually collected images with category annotations has proven to be indispensable for enhancing the generalization capacity of downstream visual tasks. However, those manually collected images often exhibit bias, which is non-transferable across either categories or domains, thus causing the model's generalization capacity degeneration. To mitigate this problem, we present a low-biased general annotated dataset generation framework (lbGen). Instead of expensive manual collection, we aim at directly generating low-biased images with category annotations. To achieve this goal, we propose to leverage the advantage of a multimodal foundation model (e.g., CLIP), in terms of aligning images in a low-biased semantic space defined by language. Specifically, we develop a bi-level semantic alignment loss, which not only forces all generated images to be consistent with the semantic distribution of all categories belonging to the target dataset in an adversarial learning manner, but also requires each generated image to match the semantic description of its category name. In addition, we further cast an existing image quality scoring model into a quality assurance loss to preserve the quality of the generated image. By leveraging these two loss functions, we can obtain a low-biased image generation model by simply fine-tuning a pre-trained diffusion model using only all category names in the target dataset as input. Experimental results confirm that, compared with the manually labeled dataset or other synthetic datasets, the utilization of our generated low-biased dataset leads to stable generalization capacity enhancement of different backbone networks across various tasks, especially in tasks where the manually labeled samples are scarce.

cs.CV↗

arXiv2025

Strong correlation between $H$-linear magnetoresistance and strange metal in FeSe superconductor

In strange metals, a strong and anomalous scattering effect exists and increases linearly with temperature. In FeSe, we observed that the temperature dependence of resistivity exhibits non-Fermi liquid behavior in two regions below and above a critical pressure, $p_\text{c}$$\sim$2 GPa. As pressure increases, a transition from quadratic to nonsaturating magnetoresistance is observed, with a distinct crossover between these two behaviors indicated by $B^{*}$ in the derivative analysis. After subtracting the quadratic term from the magnetoresistance, the residual magnetoresistance clearly exhibits an $H$-linear behavior. Additionally, two segments of $H$-inear magnetoresistance appear with increasing pressure, each arising from distinct origins. Notably, the two $H$-linear magnetoresistances exist and develop within the strange metal states. These results suggest that $H$-linear magnetoresistance is in strong correlation with the strange metal state, which may affect superconductivity in FeSe under pressure. Our study provides valuable insights into the strange metal state and clues for underlying unconventional superconductivity.

cond-mat.supr-con↗

arXiv2025

Beam test result and digitization of TaichuPix-3: A Monolithic Active Pixel Sensors for CEPC vertex detector

The Circular Electron-Positron Collider (CEPC), as the next-generation electron-positron collider, is tasked with advancing not only Higgs physics but also the discovery of new physics. Achieving these goals requires high-precision measurements of particles. Taichu seires, Monolithic Active Pixel Sensor (MAPS), a key component of the vertex detector for CEPC was designed to meet the CEPC's requirements. For the geometry of vertex detector is long barrel with no endcap, and current silicon lacks a complete digitization model, precise estimation of cluster size particularly causing by particle with large incident angle is needed. Testbeam results were conducted at the Beijing Synchrotron Radiation Facility (BSRF) to evaluate cluster size dependence on different incident angles and threshold settings. Experimental results confirmed that cluster size increases with incident angle. Simulations using the Allpix$^2$ framework replicated experimental trends at small angles but exhibited discrepancies at large angles, suggesting limitations in linear electric field assumptions and sensor thickness approximations. The results from both testbeam and simulations have provided insights into the performance of the TaichuPix chip at large incident angles, offering a crucial foundation for the establishment of a digital model and addressing the estimation of cluster size in the forward region of the long barrel. Furthermore, it offers valuable references for future iterations of TaichuPix, the development of digital models, and the simulation and estimation of the vertex detector's performance.

physics.ins-det↗

arXiv2025

Energy-Adaptive Checkpoint-Free Intermittent Inference for Low Power Energy Harvesting Systems

Deep neural network (DNN) inference in energy harvesting (EH) devices poses significant challenges due to resource constraints and frequent power interruptions. These power losses not only increase end-to-end latency, but also compromise inference consistency and accuracy, as existing checkpointing and restore mechanisms are prone to errors. Consequently, the quality of service (QoS) for DNN inference on EH devices is severely impacted. In this paper, we propose an energy-adaptive DNN inference mechanism capable of dynamically transitioning the model into a low-power mode by reducing computational complexity when harvested energy is limited. This approach ensures that end-to-end latency requirements are met. Additionally, to address the limitations of error-prone checkpoint-and-restore mechanisms, we introduce a checkpoint-free intermittent inference framework that ensures consistent, progress-preserving DNN inference during power failures in energy-harvesting systems.

cs.CE↗

arXiv2025

Improving Data Efficiency via Curating LLM-Driven Rating Systems

Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce DS2, a Diversity-aware Score curation method for Data Selection. By systematically modeling error patterns through a score transition matrix, DS2 corrects LLM-based scores and promotes diversity in the selected data samples. Our approach shows that a curated subset (just 3.3% of the original dataset) outperforms full-scale datasets (300k samples) across various machine-alignment benchmarks, and matches or surpasses human-aligned datasets such as LIMA with the same sample size (1k samples). These findings challenge conventional data scaling assumptions, highlighting that redundant, low-quality samples can degrade performance and reaffirming that "more can be less."

cs.CL↗

arXiv2025

Extrapolating and Decoupling Image-to-Video Generation Models: Motion Modeling is Easier Than You Think

Image-to-Video (I2V) generation aims to synthesize a video clip according to a given image and condition (e.g., text). The key challenge of this task lies in simultaneously generating natural motions while preserving the original appearance of the images. However, current I2V diffusion models (I2V-DMs) often produce videos with limited motion degrees or exhibit uncontrollable motion that conflicts with the textual condition. To address these limitations, we propose a novel Extrapolating and Decoupling framework, which introduces model merging techniques to the I2V domain for the first time. Specifically, our framework consists of three separate stages: (1) Starting with a base I2V-DM, we explicitly inject the textual condition into the temporal module using a lightweight, learnable adapter and fine-tune the integrated model to improve motion controllability. (2) We introduce a training-free extrapolation strategy to amplify the dynamic range of the motion, effectively reversing the fine-tuning process to enhance the motion degree significantly. (3) With the above two-stage models excelling in motion controllability and degree, we decouple the relevant parameters associated with each type of motion ability and inject them into the base I2V-DM. Since the I2V-DM handles different levels of motion controllability and dynamics at various denoising time steps, we adjust the motion-aware parameters accordingly over time. Extensive qualitative and quantitative experiments have been conducted to demonstrate the superiority of our framework over existing methods.

cs.CV↗

arXiv2025

H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking

Large Reasoning Models (LRMs) have recently extended their powerful reasoning capabilities to safety checks-using chain-of-thought reasoning to decide whether a request should be answered. While this new approach offers a promising route for balancing model utility and safety, its robustness remains underexplored. To address this gap, we introduce Malicious-Educator, a benchmark that disguises extremely dangerous or malicious requests beneath seemingly legitimate educational prompts. Our experiments reveal severe security flaws in popular commercial-grade LRMs, including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking. For instance, although OpenAI's o1 model initially maintains a high refusal rate of about 98%, subsequent model updates significantly compromise its safety; and attackers can easily extract criminal strategies from DeepSeek-R1 and Gemini 2.0 Flash Thinking without any additional tricks. To further highlight these vulnerabilities, we propose Hijacking Chain-of-Thought (H-CoT), a universal and transferable attack method that leverages the model's own displayed intermediate reasoning to jailbreak its safety reasoning mechanism. Under H-CoT, refusal rates sharply decline-dropping from 98% to below 2%-and, in some instances, even transform initially cautious tones into ones that are willing to provide harmful content. We hope these findings underscore the urgent need for more robust safety mechanisms to preserve the benefits of advanced reasoning capabilities without compromising ethical standards.

cs.CL↗

arXiv2025

Brain-inspired analogical mixture prototypes for few-shot class-incremental learning

Few-shot class-incremental learning (FSCIL) poses significant challenges for artificial neural networks due to the need to efficiently learn from limited data while retaining knowledge of previously learned tasks. Inspired by the brain's mechanisms for categorization and analogical learning, we propose a novel approach called Brain-inspired Analogical Mixture Prototypes (BAMP). BAMP has three components: mixed prototypical feature learning, statistical analogy, and soft voting. Starting from a pre-trained Vision Transformer (ViT), mixed prototypical feature learning represents each class using a mixture of prototypes and fine-tunes these representations during the base session. The statistical analogy calibrates the mean and covariance matrix of prototypes for new classes according to similarity to the base classes, and computes classification score with Mahalanobis distance. Soft voting combines both merits of statistical analogy and an off-shelf FSCIL method. Our experiments on benchmark datasets demonstrate that BAMP outperforms state-of-the-art on both traditional big start FSCIL setting and challenging small start FSCIL setting. The study suggests that brain-inspired analogical mixture prototypes can alleviate catastrophic forgetting and over-fitting problems in FSCIL.

cs.CV↗

arXiv2025

UniQ: Unified Decoder with Task-specific Queries for Efficient Scene Graph Generation

Scene Graph Generation(SGG) is a scene understanding task that aims at identifying object entities and reasoning their relationships within a given image. In contrast to prevailing two-stage methods based on a large object detector (e.g., Faster R-CNN), one-stage methods integrate a fixed-size set of learnable queries to jointly reason relational triplets . This paradigm demonstrates robust performance with significantly reduced parameters and computational overhead. However, the challenge in one-stage methods stems from the issue of weak entanglement, wherein entities involved in relationships require both coupled features shared within triplets and decoupled visual features. Previous methods either adopt a single decoder for coupled triplet feature modeling or multiple decoders for separate visual feature extraction but fail to consider both. In this paper, we introduce UniQ, a Unified decoder with task-specific Queries architecture, where task-specific queries generate decoupled visual features for subjects, objects, and predicates respectively, and unified decoder enables coupled feature modeling within relational triplets. Experimental results on the Visual Genome dataset demonstrate that UniQ has superior performance to both one-stage and two-stage methods.

cs.CV↗

arXiv2025

Overcoming Language Priors for Visual Question Answering Based on Knowledge Distillation

Previous studies have pointed out that visual question answering (VQA) models are prone to relying on language priors for answer predictions. In this context, predictions often depend on linguistic shortcuts rather than a comprehensive grasp of multimodal knowledge, which diminishes their generalization ability. In this paper, we propose a novel method, namely, KDAR, leveraging knowledge distillation to address the prior-dependency dilemmas within the VQA task. Specifically, the regularization effect facilitated by soft labels from a well-trained teacher is employed to penalize overfitting to the most common answers. The soft labels, which serve a regularization role, also provide semantic guidance that narrows the range of candidate answers. Additionally, we design an adaptive sample-wise reweighting learning strategy to further mitigate bias by dynamically adjusting the importance of each sample. Experimental results demonstrate that our method enhances performance in both OOD and IID settings. Our method achieves state-of-the-art performance on the VQA-CPv2 out-of-distribution (OOD) benchmark, significantly outperforming previous state-of-the-art approaches.

cs.CV↗

arXiv2025

Critical Current Density and AC Magnetic Susceptibility of High-quality FeTe$_{0.5}$Se$_{0.5}$ Superconducting Tapes

Iron telluride-selenium superconducting materials, known for their non-toxicity, ease of preparation, simple structure, and high upper critical fields, have attracted much research interest in practical application. In this work, we conducted electrical transport measurements, magneto-optical imaging, and AC magnetic susceptibility measurements on FeTe$_{0.5}$Se$_{0.5}$ superconducting long tapes fabricated via reel-to-reel pulsed laser deposition. Our transport measurements revealed a high critical current density that remains relatively stable even with increasing external magnetic fields, reaching over $1\times 10^5$ A/cm$^2$ at 8 K and 9 T. The calculated pinning force density indicates that normal point pinning is the primary mechanism in these tapes. The magneto-optical images demonstrated that the tapes show homogeneous superconductivity and uniform distribution of critical current density. The AC magnetic susceptibility measurements also confirmed their strong flux pinning nature of withstanding high magnetic field. Based on these characteristics, FeTe$_{0.5}$Se$_{0.5}$ superconducting tapes show promising prospects for applications under high magnetic fields.

cond-mat.supr-con↗

arXiv2025

Exploring EEG and Eye Movement Fusion for Multi-Class Target RSVP-BCI

Rapid Serial Visual Presentation (RSVP)-based Brain-Computer Interfaces (BCIs) facilitate high-throughput target image detection by identifying event-related potentials (ERPs) evoked in EEG signals. The RSVP-BCI systems effectively detect single-class targets within a stream of images but have limited applicability in scenarios that require detecting multiple target categories. Multi-class RSVP-BCI systems address this limitation by simultaneously identifying the presence of a target and distinguishing its category. However, existing multi-class RSVP decoding algorithms predominantly rely on single-modality EEG decoding, which restricts their performance improvement due to the high similarity between ERPs evoked by different target categories. In this work, we introduce eye movement (EM) modality into multi-class RSVP decoding and explore EEG and EM fusion to enhance decoding performance. First, we design three independent multi-class target RSVP tasks and build an open-source dataset comprising EEG and EM signals from 43 subjects. Then, we propose the Multi-class Target RSVP EEG and EM fusion Network (MTREE-Net) to enhance multi-class RSVP decoding. Specifically, a dual-complementary module is proposed to strengthen the differentiation of uni-modal features across categories. To improve multi-modal fusion performance, we adopt a dynamic reweighting fusion strategy guided by theoretically derived modality contribution ratios. Furthermore, we reduce the misclassification of non-target samples through knowledge transfer between two hierarchical classifiers. Extensive experiments demonstrate the feasibility of integrating EM signals into multi-class RSVP decoding and highlight the superior performance of MTREE-Net compared to existing RSVP decoding methods. The proposed MTREE-Net and open-source dataset provide a promising framework for developing practical multi-class RSVP-BCI systems.

cs.HC↗

arXiv2025

Integrating Language-Image Prior into EEG Decoding for Cross-Task Zero-Calibration RSVP-BCI

Rapid Serial Visual Presentation (RSVP)-based Brain-Computer Interface (BCI) is an effective technology used for information detection by detecting Event-Related Potentials (ERPs). The current RSVP decoding methods can perform well in decoding EEG signals within a single RSVP task, but their decoding performance significantly decreases when directly applied to different RSVP tasks without calibration data from the new tasks. This limits the rapid and efficient deployment of RSVP-BCI systems for detecting different categories of targets in various scenarios. To overcome this limitation, this study aims to enhance the cross-task zero-calibration RSVP decoding performance. First, we design three distinct RSVP tasks for target image retrieval and build an open-source dataset containing EEG signals and corresponding stimulus images. Then we propose an EEG with Language-Image Prior fusion Transformer (ELIPformer) for cross-task zero-calibration RSVP decoding. Specifically, we propose a prompt encoder based on the language-image pre-trained model to extract language-image features from task-specific prompts and stimulus images as prior knowledge for enhancing EEG decoding. A cross bidirectional attention mechanism is also adopted to facilitate the effective feature fusion and alignment between the EEG and language-image features. Extensive experiments demonstrate that the proposed model achieves superior performance in cross-task zero-calibration RSVP decoding, which promotes the RSVP-BCI system from research to practical application.

cs.HC↗

arXiv2024

Large Language Models-guided Dynamic Adaptation for Temporal Knowledge Graph Reasoning

Temporal Knowledge Graph Reasoning (TKGR) is the process of utilizing temporal information to capture complex relations within a Temporal Knowledge Graph (TKG) to infer new knowledge. Conventional methods in TKGR typically depend on deep learning algorithms or temporal logical rules. However, deep learning-based TKGRs often lack interpretability, whereas rule-based TKGRs struggle to effectively learn temporal rules that capture temporal patterns. Recently, Large Language Models (LLMs) have demonstrated extensive knowledge and remarkable proficiency in temporal reasoning. Consequently, the employment of LLMs for Temporal Knowledge Graph Reasoning (TKGR) has sparked increasing interest among researchers. Nonetheless, LLMs are known to function as black boxes, making it challenging to comprehend their reasoning process. Additionally, due to the resource-intensive nature of fine-tuning, promptly updating LLMs to integrate evolving knowledge within TKGs for reasoning is impractical. To address these challenges, in this paper, we propose a Large Language Models-guided Dynamic Adaptation (LLM-DA) method for reasoning on TKGs. Specifically, LLM-DA harnesses the capabilities of LLMs to analyze historical data and extract temporal logical rules. These rules unveil temporal patterns and facilitate interpretable reasoning. To account for the evolving nature of TKGs, a dynamic adaptation strategy is proposed to update the LLM-generated rules with the latest events. This ensures that the extracted rules always incorporate the most recent knowledge and better generalize to the predictions on future events. Experimental results show that without the need of fine-tuning, LLM-DA significantly improves the accuracy of reasoning over several common datasets, providing a robust framework for TKGR tasks.

cs.AI↗

arXiv2024

Tackling the Dynamicity in a Production LLM Serving System with SOTA Optimizations via Hybrid Prefill/Decode/Verify Scheduling on Efficient Meta-kernels

Meeting growing demands for low latency and cost efficiency in production-grade large language model (LLM) serving systems requires integrating advanced optimization techniques. However, dynamic and unpredictable input-output lengths of LLM, compounded by these optimizations, exacerbate the issues of workload variability, making it difficult to maintain high efficiency on AI accelerators, especially DSAs with tile-based programming models. To address this challenge, we introduce XY-Serve, a versatile, Ascend native, end-to-end production LLM-serving system. The core idea is an abstraction mechanism that smooths out the workload variability by decomposing computations into unified, hardware-friendly, fine-grained meta primitives. For attention, we propose a meta-kernel that computes the basic pattern of matmul-softmax-matmul with architectural-aware tile sizes. For GEMM, we introduce a virtual padding scheme that adapts to dynamic shape changes while using highly efficient GEMM primitives with assorted fixed tile sizes. XY-Serve sits harmoniously with vLLM. Experimental results show up to 89% end-to-end throughput improvement compared with current publicly available baselines on Ascend NPUs. Additionally, our approach outperforms existing GEMM (average 14.6% faster) and attention (average 21.5% faster) kernels relative to existing libraries. While the work is Ascend native, we believe the approach can be readily applicable to SIMT architectures as well.

cs.AI↗

arXiv2024

SpatialMe: Stereo Video Conversion Using Depth-Warping and Blend-Inpainting

Stereo video conversion aims to transform monocular videos into immersive stereo format. Despite the advancements in novel view synthesis, it still remains two major challenges: i) difficulty of achieving high-fidelity and stable results, and ii) insufficiency of high-quality stereo video data. In this paper, we introduce SpatialMe, a novel stereo video conversion framework based on depth-warping and blend-inpainting. Specifically, we propose a mask-based hierarchy feature update (MHFU) refiner, which integrate and refine the outputs from designed multi-branch inpainting module, using feature update unit (FUU) and mask mechanism. We also propose a disparity expansion strategy to address the problem of foreground bleeding. Furthermore, we conduct a high-quality real-world stereo video dataset -- StereoV1K, to alleviate the data shortage. It contains 1000 stereo videos captured in real-world at a resolution of 1180 x 1180, covering various indoor and outdoor scenes. Extensive experiments demonstrate the superiority of our approach in generating stereo videos over state-of-the-art methods.

cs.CV↗

arXiv2024

Representation Learning with Large Language Models for Recommendation

Recommender systems have seen significant advancements with the influence of deep learning and graph neural networks, particularly in capturing complex user-item relationships. However, these graph-based recommenders heavily depend on ID-based data, potentially disregarding valuable textual information associated with users and items, resulting in less informative learned representations. Moreover, the utilization of implicit feedback data introduces potential noise and bias, posing challenges for the effectiveness of user preference learning. While the integration of large language models (LLMs) into traditional ID-based recommenders has gained attention, challenges such as scalability issues, limitations in text-only reliance, and prompt input constraints need to be addressed for effective implementation in practical recommender systems. To address these challenges, we propose a model-agnostic framework RLMRec that aims to enhance existing recommenders with LLM-empowered representation learning. It proposes a recommendation paradigm that integrates representation learning with LLMs to capture intricate semantic aspects of user behaviors and preferences. RLMRec incorporates auxiliary textual signals, develops a user/item profiling paradigm empowered by LLMs, and aligns the semantic space of LLMs with the representation space of collaborative relational signals through a cross-view alignment framework. This work further establish a theoretical foundation demonstrating that incorporating textual signals through mutual information maximization enhances the quality of representations. In our evaluation, we integrate RLMRec with state-of-the-art recommender models, while also analyzing its efficiency and robustness to noise data. Our implementation codes are available at https://github.com/HKUDS/RLMRec.

cs.IR↗

arXiv2024

Local and Global Feature Attention Fusion Network for Face Recognition

Recognition of low-quality face images remains a challenge due to invisible or deformation in partial facial regions. For low-quality images dominated by missing partial facial regions, local region similarity contributes more to face recognition (FR). Conversely, in cases dominated by local face deformation, excessive attention to local regions may lead to misjudgments, while global features exhibit better robustness. However, most of the existing FR methods neglect the bias in feature quality of low-quality images introduced by different factors. To address this issue, we propose a Local and Global Feature Attention Fusion (LGAF) network based on feature quality. The network adaptively allocates attention between local and global features according to feature quality and obtains more discriminative and high-quality face features through local and global information complementarity. In addition, to effectively obtain fine-grained information at various scales and increase the separability of facial features in high-dimensional space, we introduce a Multi-Head Multi-Scale Local Feature Extraction (MHMS) module. Experimental results demonstrate that the LGAF achieves the best average performance on $4$ validation sets (CFP-FP, CPLFW, AgeDB, and CALFW), and the performance on TinyFace and SCFace outperforms the state-of-the-art methods (SoTA).

cs.CV↗