arXiv subjects

Yang Shi

Publications and source records attributed to Yang Shi.

At least 109 records · Page 6Linked to original sources

arXiv2025

Panther: A Cost-Effective Privacy-Preserving Framework for GNN Training and Inference Services in Cloud Environments

Graph Neural Networks (GNNs) have marked significant impact in traffic state prediction, social recommendation, knowledge-aware question answering and so on. As more and more users move towards cloud computing, it has become a critical issue to unleash the power of GNNs while protecting the privacy in cloud environments. Specifically, the training data and inference data for GNNs need to be protected from being stolen by external adversaries. Meanwhile, the financial cost of cloud computing is another primary concern for users. Therefore, although existing studies have proposed privacy-preserving techniques for GNNs in cloud environments, their additional computational and communication overhead remain relatively high, causing high financial costs that limit their widespread adoption among users. To protect GNN privacy while lowering the additional financial costs, we introduce Panther, a cost-effective privacy-preserving framework for GNN training and inference services in cloud environments. Technically, Panther leverages four-party computation to asynchronously executing the secure array access protocol, and randomly pads the neighbor information of GNN nodes. We prove that Panther can protect privacy for both training and inference of GNN models. Our evaluation shows that Panther reduces the training and inference time by an average of 75.28% and 82.80%, respectively, and communication overhead by an average of 52.61% and 50.26% compared with the state-of-the-art, which is estimated to save an average of 55.05% and 59.00% in financial costs (based on on-demand pricing model) for the GNN training and inference process on Google Cloud Platform.

cs.CR↗

arXiv2025

MorphoBench: A Benchmark with Difficulty Adaptive to Model Reasoning

With the advancement of powerful large-scale reasoning models, effectively evaluating the reasoning capabilities of these models has become increasingly important. However, existing benchmarks designed to assess the reasoning abilities of large models tend to be limited in scope and lack the flexibility to adapt their difficulty according to the evolving reasoning capacities of the models. To address this, we propose MorphoBench, a benchmark that incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models. Specifically, we curate the benchmark by selecting and collecting complex reasoning questions from existing benchmarks and sources such as Olympiad-level competitions. Additionally, MorphoBench adaptively modifies the analytical challenge of questions by leveraging key statements generated during the model's reasoning process. Furthermore, it includes questions generated using simulation software, enabling dynamic adjustment of benchmark difficulty with minimal resource consumption. We have gathered over 1,300 test questions and iteratively adjusted the difficulty of MorphoBench based on the reasoning capabilities of models such as o3 and GPT-5. MorphoBench enhances the comprehensiveness and validity of model reasoning evaluation, providing reliable guidance for improving both the reasoning abilities and scientific robustness of large models. The code has been released in https://github.com/OpenDCAI/MorphoBench.

cs.AI↗

arXiv2025

AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestration

Audiovisual video captioning aims to generate semantically rich descriptions with temporal alignment between visual and auditory events, thereby benefiting both video understanding and generation. In this paper, we present AVoCaDO, a powerful audiovisual video captioner driven by the temporal orchestration between audio and visual modalities. We propose a two-stage post-training pipeline: (1) AVoCaDO SFT, which fine-tunes the model on a newly curated dataset of 107K high-quality, temporally-aligned audiovisual captions; and (2) AVoCaDO GRPO, which leverages tailored reward functions to further enhance temporal coherence and dialogue accuracy while regularizing caption length and reducing collapse. Experimental results demonstrate that AVoCaDO significantly outperforms existing open-source models across four audiovisual video captioning benchmarks, and also achieves competitive performance on the VDC and DREAM-1K benchmark under visual-only settings.

cs.CV↗

arXiv2025

Distributionally Robust Control with End-to-End Statistically Guaranteed Metric Learning

Wasserstein distributionally robust control (DRC) recently emerges as a principled paradigm for handling uncertainty in stochastic dynamical systems. However, it constructs data-driven ambiguity sets via uniform distribution shifts before sequentially incorporating them into downstream control synthesis. This segregation between ambiguity set construction and control objectives inherently introduces a structural misalignment, which undesirably leads to conservative control policies with sub-optimal performance. To address this limitation, we propose a novel end-to-end finite-horizon Wasserstein DRC framework that integrates the learning of anisotropic Wasserstein metrics with downstream control tasks in a closed-loop manner, thus enabling ambiguity sets to be systematically adjusted along performance-critical directions and yielding more effective control policies. This framework is formulated as a bilevel program: the inner level characterizes dynamical system evolution under DRC, while the outer level refines the anisotropic metric leveraging control-performance feedback across a range of initial conditions. To solve this program efficiently, we develop a stochastic augmented Lagrangian algorithm tailored to the bilevel structure. Theoretically, we prove that the learned ambiguity sets preserve statistical finite-sample guarantees under a novel radius adjustment mechanism, and we establish the well-posedness of the bilevel formulation by demonstrating its continuity with respect to the learnable metric. Furthermore, we show that the algorithm converges to stationary points of the outer level problem, which are statistically consistent with the optimal metric at a non-asymptotic convergence rate. Experiments on both numerical and inventory control tasks verify that the proposed framework achieves superior closed-loop performance and robustness compared against state-of-the-art methods.

math.OC↗

arXiv2025

PC-UNet: An Enforcing Poisson Statistics U-Net for Positron Emission Tomography Denoising

Positron Emission Tomography (PET) is crucial in medicine, but its clinical use is limited due to high signal-to-noise ratio doses increasing radiation exposure. Lowering doses increases Poisson noise, which current denoising methods fail to handle, causing distortions and artifacts. We propose a Poisson Consistent U-Net (PC-UNet) model with a new Poisson Variance and Mean Consistency Loss (PVMC-Loss) that incorporates physical data to improve image fidelity. PVMC-Loss is statistically unbiased in variance and gradient adaptation, acting as a Generalized Method of Moments implementation, offering robustness to minor data mismatches. Tests on PET datasets show PC-UNet improves physical consistency and image fidelity, proving its ability to integrate physical information effectively.

cs.CV↗

arXiv2025

Image-Based Visual Servoing for Enhanced Cooperation of Dual-Arm Manipulation

The cooperation of a pair of robot manipulators is required to manipulate a target object without any fixtures. The conventional control methods coordinate the end-effector pose of each manipulator with that of the other using their kinematics and joint coordinate measurements. Yet, the manipulators' inaccurate kinematics and joint coordinate measurements can cause significant pose synchronization errors in practice. This paper thus proposes an image-based visual servoing approach for enhancing the cooperation of a dual-arm manipulation system. On top of the classical control, the visual servoing controller lets each manipulator use its carried camera to measure the image features of the other's marker and adapt its end-effector pose with the counterpart on the move. Because visual measurements are robust to kinematic errors, the proposed control can reduce the end-effector pose synchronization errors and the fluctuations of the interaction forces of the pair of manipulators on the move. Theoretical analyses have rigorously proven the stability of the closed-loop system. Comparative experiments on real robots have substantiated the effectiveness of the proposed control.

cs.RO↗

arXiv2025

VidBridge-R1: Bridging QA and Captioning for RL-based Video Understanding Models with Intermediate Proxy Tasks

The "Reason-Then-Respond" paradigm, enhanced by Reinforcement Learning, has shown great promise in advancing Multimodal Large Language Models. However, its application to the video domain has led to specialized models that excel at either question answering (QA) or captioning tasks, but struggle to master both. Naively combining reward signals from these tasks results in mutual performance degradation, which we attribute to a conflict between their opposing task natures. To address this challenge, we propose a novel training framework built upon two intermediate proxy tasks: DarkEventInfer, which presents videos with masked event segments, requiring models to infer the obscured content based on contextual video cues; and MixVidQA, which presents interleaved video sequences composed of two distinct clips, challenging models to isolate and reason about one while disregarding the other. These proxy tasks compel the model to simultaneously develop both holistic, divergent understanding and precise, convergent reasoning capabilities. Embodying this framework, we present VidBridge-R1, the first versatile video reasoning model that effectively bridges the paradigm conflict. Extensive experiments show that VidBridge-R1 achieves significant performance gains on both QA and captioning within one model, demonstrating the efficacy of our approach in fostering more generalizable and powerful video understanding models.

cs.CV↗

arXiv2025

BaseReward: A Strong Baseline for Multimodal Reward Model

The rapid advancement of Multimodal Large Language Models (MLLMs) has made aligning them with human preferences a critical challenge. Reward Models (RMs) are a core technology for achieving this goal, but a systematic guide for building state-of-the-art Multimodal Reward Models (MRMs) is currently lacking in both academia and industry. Through exhaustive experimental analysis, this paper aims to provide a clear ``recipe'' for constructing high-performance MRMs. We systematically investigate every crucial component in the MRM development pipeline, including \textit{reward modeling paradigms} (e.g., Naive-RM, Critic-based RM, and Generative RM), \textit{reward head architecture}, \textit{training strategies}, \textit{data curation} (covering over ten multimodal and text-only preference datasets), \textit{backbone model} and \textit{model scale}, and \textit{ensemble methods}. Based on these experimental insights, we introduce \textbf{BaseReward}, a powerful and efficient baseline for multimodal reward modeling. BaseReward adopts a simple yet effective architecture, built upon a {Qwen2.5-VL} backbone, featuring an optimized two-layer reward head, and is trained on a carefully curated mixture of high-quality multimodal and text-only preference data. Our results show that BaseReward establishes a new SOTA on major benchmarks such as MM-RLHF-Reward Bench, VL-Reward Bench, and Multimodal Reward Bench, outperforming previous models. Furthermore, to validate its practical utility beyond static benchmarks, we integrate BaseReward into a real-world reinforcement learning pipeline, successfully enhancing an MLLM's performance across various perception, reasoning, and conversational tasks. This work not only delivers a top-tier MRM but, more importantly, provides the community with a clear, empirically-backed guide for developing robust reward models for the next generation of MLLMs.

cs.CV↗

arXiv2025

Resilient Control for Networked Switched Systems With/Without ACK: An Active Quantized Framework

This paper deals with the quantized control problem for switched systems under denial-of-service (DoS) attack. Considering the system's defensive capability and the computational resources of quantizers and controllers, four control strategies are proposed. These strategies incorporate different combinations of controllers (active and passive), quantizers (centered on the origin or custom-designed), and network configurations (with or without ACK signals). For each strategy, specific update laws for the encoder and decoder are designed to avoid quantization saturation. Furthermore, the uniformity of encoder and decoder operations is maintained by transmitting additional information to the decoder. To achieve asymptotic stability, sufficient conditions concerning the switching signal and DoS attack constraints are derived by taking into account the asynchronous behaviors. The proposed active quantization strategy with the ACK signal leverages the system model information to compute the control signal in real-time, allowing for possible convergence of the system state despite DoS attack. Additionally, a well-designed switching signal is suggested to further mitigate the impact of DoS attack. A passive quantization strategy with ACK signal is also developed as a simplified version of the active quantized control strategy, providing the foundation for a strategy without ACK signal. Inspired by time-triggered and event-triggered mechanisms, the passive quantization strategy without ACK signal is investigated, with two feasible update laws for the quantizer. Finally, two simulations are conducted to validate the effectiveness of the proposed strategies.

eess.SY↗

arXiv2025

Debiasing Multimodal Large Language Models via Penalization of Language Priors

In the realms of computer vision and natural language processing, Multimodal Large Language Models (MLLMs) have become indispensable tools, proficient in generating textual responses based on visual inputs. Despite their advancements, our investigation reveals a noteworthy bias: the generated content is often driven more by the inherent priors of the underlying Large Language Models (LLMs) than by the input image. Empirical experiments underscore the persistence of this bias, as MLLMs often provide confident answers even in the absence of relevant images or given incongruent visual inputs. To rectify these biases and redirect the model's focus toward visual information, we propose two simple, training-free strategies. First, for tasks such as classification or multi-choice question answering, we introduce a "Post-Hoc Debias" method using an affine calibration step to adjust the output distribution. This approach ensures uniform answer scores when the image is absent, acting as an effective regularization technique to alleviate the influence of LLM priors. For more intricate open-ended generation tasks, we extend this method to "Visual Debias Decoding", which mitigates bias by contrasting token log-probabilities conditioned on a correct image versus a meaningless one. Additionally, our investigation sheds light on the instability of MLLMs across various decoding configurations. Through systematic exploration of different settings, we achieve significant performance improvements--surpassing previously reported results--and raise concerns about the fairness of current evaluation practices. Comprehensive experiments substantiate the effectiveness of our proposed strategies in mitigating biases. These strategies not only prove beneficial in minimizing hallucinations but also contribute to the generation of more helpful and precise illustrations.

cs.CV↗

arXiv2025

Information-triggered Learning with Application to Learning-based Predictive Control

Learning-based control has attracted significant attention in recent years, especially for plants that are difficult to model based on first-principles. A key issue in learning-based control is how to make efficient use of data as the abundance of data becomes overwhelming. To address this issue, this work proposes an information-triggered learning framework and a corresponding learning-based controller design approach with guaranteed stability. Specifically, we consider a linear time-invariant system with unknown dynamics. A set-membership approach is introduced to learn a parametric uncertainty set for the unknown dynamics. Then, a data selection mechanism is proposed by evaluating the incremental information in a data sample, where the incremental information is quantified by its effects on shrinking the parametric uncertainty set. Next, after introducing a stability criterion using the set-membership estimate of the system dynamics, a robust learning-based predictive controller (LPC) is designed by minimizing a worst-case cost function. The closed-loop stability of the LPC equipped with the information-triggered learning protocol is discussed within a high-probability framework. Finally, comparative numerical experiments are performed to verify the validity of the proposed approach.

eess.SY↗

arXiv2025

SwiftPrune: Hessian-Free Weight Pruning for Large Language Models

Post-training pruning, as one of the key techniques for compressing large language models, plays a vital role in lightweight model deployment and model sparsity. However, current mainstream pruning methods dependent on the Hessian matrix face significant limitations in both pruning speed and practical effectiveness due to the computationally intensive nature of second-order derivative calculations. This paper presents SwiftPrune, a novel Hessian-free weight pruning method that achieves hardware-efficient model compression through two key innovations: 1) SwiftPrune eliminates the need for computationally intensive Hessian matrix calculations by introducing a contribution-based weight metric, which evaluates the importance of weights without relying on second-order derivatives. 2) we employ the Exponentially Weighted Moving Average (EWMA) technique to bypass weight sorting, enabling the selection of weights that contribute most to LLM accuracy and further reducing time complexity. Our approach is extended to support structured sparsity pruning, facilitating efficient execution on modern hardware accelerators. We validate the SwiftPrune on three LLMs (namely LLaMA2, LLaMA3, and Pythia), demonstrating that it significantly enhances compression performance. The experimental findings reveal that SwiftPrune completes the pruning process within seconds, achieving an average speedup of 12.29x (up to 56.02x) over existing SOTA approaches.

cs.LG↗

arXiv2025

MV-Crafter: An Intelligent System for Music-guided Video Generation

Music videos, as a prevalent form of multimedia entertainment, deliver engaging audio-visual experiences to audiences and have gained immense popularity among singers and fans. Creators can express their interpretations of music naturally through visual elements. However, the creation process of music video demands proficiency in script design, video shooting, and music-video synchronization, posing significant challenges for non-professionals. Previous work has designed automated music video generation frameworks. However, they suffer from complexity in input and poor output quality. In response, we present MV-Crafter, a system capable of producing high-quality music videos with synchronized music-video rhythm and style. Our approach involves three technical modules that simulate the human creation process: the script generation module, video generation module, and music-video synchronization module. MV-Crafter leverages a large language model to generate scripts considering the musical semantics. To address the challenge of synchronizing short video clips with music of varying lengths, we propose a dynamic beat matching algorithm and visual envelope-induced warping method to ensure precise, monotonic music-video synchronization. Besides, we design a user-friendly interface to simplify the creation process with intuitive editing features. Extensive experiments have demonstrated that MV-Crafter provides an effective solution for improving the quality of generated music videos.

cs.HC↗

arXiv2025

EmbodiedEval: Evaluate Multimodal LLMs as Embodied Agents

Multimodal Large Language Models (MLLMs) have shown significant advancements, providing a promising future for embodied agents. Existing benchmarks for evaluating MLLMs primarily utilize static images or videos, limiting assessments to non-interactive scenarios. Meanwhile, existing embodied AI benchmarks are task-specific and not diverse enough, which do not adequately evaluate the embodied capabilities of MLLMs. To address this, we propose EmbodiedEval, a comprehensive and interactive evaluation benchmark for MLLMs with embodied tasks. EmbodiedEval features 328 distinct tasks within 125 varied 3D scenes, each of which is rigorously selected and annotated. It covers a broad spectrum of existing embodied AI tasks with significantly enhanced diversity, all within a unified simulation and evaluation framework tailored for MLLMs. The tasks are organized into five categories: navigation, object interaction, social interaction, attribute question answering, and spatial question answering to assess different capabilities of the agents. We evaluated the state-of-the-art MLLMs on EmbodiedEval and found that they have a significant shortfall compared to human level on embodied tasks. Our analysis demonstrates the limitations of existing MLLMs in embodied capabilities, providing insights for their future development. We open-source all evaluation data and simulation framework at https://github.com/thunlp/EmbodiedEval.

cs.CV↗

arXiv2025

MME-Unify: A Comprehensive Benchmark for Unified Multimodal Understanding and Generation Models

Existing MLLM benchmarks face significant challenges in evaluating Unified MLLMs (U-MLLMs) due to: 1) lack of standardized benchmarks for traditional tasks, leading to inconsistent comparisons; 2) absence of benchmarks for mixed-modality generation, which fails to assess multimodal reasoning capabilities. We present a comprehensive evaluation framework designed to systematically assess U-MLLMs. Our benchmark includes: Standardized Traditional Task Evaluation. We sample from 12 datasets, covering 10 tasks with 30 subtasks, ensuring consistent and fair comparisons across studies." 2. Unified Task Assessment. We introduce five novel tasks testing multimodal reasoning, including image editing, commonsense QA with image generation, and geometric reasoning. 3. Comprehensive Model Benchmarking. We evaluate 12 leading U-MLLMs, such as Janus-Pro, EMU3, VILA-U, and Gemini2-flash, alongside specialized understanding (e.g., Claude-3.5-Sonnet) and generation models (e.g., DALL-E-3). Our findings reveal substantial performance gaps in existing U-MLLMs, highlighting the need for more robust models capable of handling mixed-modality tasks effectively. The code and evaluation data can be found in https://mme-unify.github.io/.

cs.CV↗

arXiv2025

What is the symmetry group of a d-${\text{P}_{\mathrm{II}}}$ discrete Painlevé equation?

The symmetry group of a (discrete) Painlevé equation provides crucial information on the properties of the equation. In this paper we argue against the commonly-held belief that the symmetry group of a given equation is solely determined by its surface type as given in the famous Sakai classification. We will dispel this misconception on a specific example of a d-${\text{P}_{\mathrm{II}}}$ equation which corresponds to a half-translation on the root lattice dual to its surface-type root lattice, but which becomes a genuine translation on a sub-lattice thereof that corresponds to its real symmetry group. The latter fact is shown in two different ways: first by a brute force calculation and second through the use of normalizer theory, which we believe to be an extremely useful tool for this purpose. We finish the paper with the analysis of a sub-case of our main example which arises in the study of gap probabilities for Freud unitary ensembles, and the symmetry group of which is even further restricted due to the appearance of a nodal curve on the surface on which the equation is regularized.

nlin.SI↗

arXiv2025

Quantized Distributed Nonconvex Optimization Algorithms with Linear Convergence under the Polyak--$Ł$ojasiewicz Condition

This paper considers distributed optimization for minimizing the average of local nonconvex cost functions, by using local information exchange over undirected communication networks. To reduce the required communication capacity, we introduce an encoder--decoder scheme. By integrating them with distributed gradient tracking and proportional integral algorithms, respectively, we then propose two quantized distributed nonconvex optimization algorithms. Assuming the global cost function satisfies the Polyak--Łojasiewicz condition, which does not require the global cost function to be convex and the global minimizer is not necessarily unique, we show that our proposed algorithms linearly converge to a global optimal point and that larger quantization level leads to faster convergence speed. Moreover, we show that a low data rate is sufficient to guarantee linear convergence when the algorithm parameters are properly chosen. The theoretical results are illustrated by numerical examples.

math.OC↗

arXiv2025

Scalable Overload-Aware Graph-Based Index Construction for 10-Billion-Scale Vector Similarity Search

Approximate Nearest Neighbor Search (ANNS) is essential for modern data-driven applications that require efficient retrieval of top-k results from massive vector databases. Although existing graph-based ANNS algorithms achieve a high recall rate on billion-scale datasets, their slow construction speed and limited scalability hinder their applicability to large-scale industrial scenarios. In this paper, we introduce SOGAIC, the first Scalable Overload-Aware Graph-Based ANNS Index Construction system tailored for ultra-large-scale vector databases: 1) We propose a dynamic data partitioning algorithm with overload constraints that adaptively introduces overlaps among subsets; 2) To enable efficient distributed subgraph construction, we employ a load-balancing task scheduling framework combined with an agglomerative merging strategy; 3) Extensive experiments on various datasets demonstrate a reduction of 47.3% in average construction time compared to existing methods. The proposed method has also been successfully deployed in a real-world industrial search engine, managing over 10 billion daily updated vectors and serving hundreds of millions of users.

cs.IR↗