arXiv subjects

Jitao Sang

Publications and source records attributed to Jitao Sang.

At least 37 records · Page 2Linked to original sources

arXiv2025

ReInAgent: A Context-Aware GUI Agent Enabling Human-in-the-Loop Mobile Task Navigation

Mobile GUI agents exhibit substantial potential to facilitate and automate the execution of user tasks on mobile phones. However, exist mobile GUI agents predominantly privilege autonomous operation and neglect the necessity of active user engagement during task execution. This omission undermines their adaptability to information dilemmas including ambiguous, dynamically evolving, and conflicting task scenarios, leading to execution outcomes that deviate from genuine user requirements and preferences. To address these shortcomings, we propose ReInAgent, a context-aware multi-agent framework that leverages dynamic information management to enable human-in-the-loop mobile task navigation. ReInAgent integrates three specialized agents around a shared memory module: an information-managing agent for slot-based information management and proactive interaction with the user, a decision-making agent for conflict-aware planning, and a reflecting agent for task reflection and information consistency validation. Through continuous contextual information analysis and sustained user-agent collaboration, ReInAgent overcomes the limitation of existing approaches that rely on clear and static task assumptions. Consequently, it enables more adaptive and reliable mobile task navigation in complex, real-world scenarios. Experimental results demonstrate that ReInAgent effectively resolves information dilemmas and produces outcomes that are more closely aligned with genuine user preferences. Notably, on complex tasks involving information dilemmas, ReInAgent achieves a 25% higher success rate than Mobile-Agent-v2.

cs.AI↗

arXiv2025

Understanding the Information Cocoon: A Multidimensional Assessment and Analysis of News Recommendation Systems

Personalized news recommendation systems inadvertently create information cocoons--homogeneous information bubbles that reinforce user biases and amplify societal polarization. To address the lack of comprehensive assessment frameworks in prior research, we propose a multidimensional analysis that evaluates cocoons through dual perspectives: (1) Individual homogenization via topic diversity (including the number of topic categories and category information entropy) and click repetition; (2) Group polarization via network density and community openness. Through multi-round experiments on real-world datasets, we benchmark seven algorithms and reveal critical insights. Furthermore, we design five lightweight mitigation strategies. This work establishes the first unified metric framework for information cocoons and delivers deployable solutions for ethical recommendation systems.

cs.IR↗

arXiv2025

Efficient Video-to-Audio Generation via Multiple Foundation Models Mapper

Recent Video-to-Audio (V2A) generation relies on extracting semantic and temporal features from video to condition generative models. Training these models from scratch is resource intensive. Consequently, leveraging foundation models (FMs) has gained traction due to their cross-modal knowledge transfer and generalization capabilities. One prior work has explored fine-tuning a lightweight mapper network to connect a pre-trained visual encoder with a text-to-audio generation model for V2A. Inspired by this, we introduce the Multiple Foundation Model Mapper (MFM-Mapper). Compared to the previous mapper approach, MFM-Mapper benefits from richer semantic and temporal information by fusing features from dual visual encoders. Furthermore, by replacing a linear mapper with GPT-2, MFM-Mapper improves feature alignment, drawing parallels between cross-modal features mapping and autoregressive translation tasks. Our MFM-Mapper exhibits remarkable training efficiency. It achieves better performance in semantic and temporal consistency with fewer training consuming, requiring only 16\% of the training scale compared to previous mapper-based work, yet achieves competitive performance with models trained on a much larger scale.

cs.CV↗

arXiv2025

Investigating and Enhancing Vision-Audio Capability in Omnimodal Large Language Models

Omnimodal Large Language Models (OLLMs) have shown significant progress in integrating vision and text, but still struggle with integrating vision and audio, often exhibiting suboptimal performance when processing audio queries compared to text queries. This disparity is primarily due to insufficient alignment between vision and audio modalities during training, leading to inadequate attention to visual information when using audio queries. To mitigate this issue, we propose a Self-Knowledge Distillation (Self-KD) training method where the vision-text component of the OLLM serves as the teacher and the vision-audio component as the student. This enables the model to process audio in a manner analogous to its text processing. Our experimental results demonstrate that Self-KD is an effective method for enhancing the vision-audio capabilities of OLLMs by learning from the vision-text components, which subsequently improves the interaction between audio and images and results in improved performance on multimodal tasks.

cs.CV↗

arXiv2025

Mitigating Filter Bubble from the Perspective of Community Detection: A Universal Framework

In recent years, recommender systems have primarily focused on improving accuracy at the expense of diversity, which exacerbates the well-known filter bubble effect. This paper proposes a universal framework called CD-CGCN to address the filter bubble issue in recommender systems from a community detection perspective. By analyzing user-item interaction histories with a community detection algorithm, we reveal that state-of-the-art recommendations often focus on intra-community items, worsening the filter bubble effect. CD-CGCN, a model-agnostic framework, integrates a Conditional Discriminator and a Community-reweighted Graph Convolutional Network which can be plugged into most recommender models. Using adversarial learning based on community labels, it counteracts the extracted community attributes and incorporates an inference strategy tailored to the user's specific filter bubble state. Extensive experiments on real-world datasets with multiple base models validate its effectiveness in mitigating filter bubbles while preserving recommendation quality. Additionally, by applying community debiasing to the original test set to construct an unbiased test set, we observe that CD-CGCN demonstrates superior performance in capturing users' inter-community preferences.

cs.IR↗

arXiv2025

When Algorithms Mirror Minds: A Confirmation-Aware Social Dynamic Model of Echo Chamber and Homogenization Traps

Recommender systems increasingly suffer from echo chambers and user homogenization, systemic distortions arising from the dynamic interplay between algorithmic recommendations and human behavior. While prior work has studied these phenomena through the lens of algorithmic bias or social network structure, we argue that the psychological mechanisms of users and the closed-loop interaction between users and recommenders are critical yet understudied drivers of these emergent effects. To bridge this gap, we propose the Confirmation-Aware Social Dynamic Model which incorporates user psychology and social relationships to simulate the actual user and recommender interaction process. Our theoretical analysis proves that echo chambers and homogenization traps, defined respectively as reduced recommendation diversity and homogenized user representations, will inevitably occur. We also conduct extensive empirical simulations on two real-world datasets and one synthetic dataset with five well-designed metrics, exploring the root factors influencing the aforementioned phenomena from three level perspectives: the stochasticity and social integration degree of recommender (system-level), the psychological mechanisms of users (user-level), and the dataset scale (platform-level). Furthermore, we demonstrate four practical mitigation strategies that help alleviate echo chambers and user homogenization at the cost of some recommendation accuracy. Our findings provide both theoretical and empirical insights into the emergence and drivers of echo chambers and user homogenization, as well as actionable guidelines for human-centered recommender design.

cs.SI↗

arXiv2025

WebSynthesis: World-Model-Guided MCTS for Efficient WebUI-Trajectory Synthesis

Recent advancements in large language models (LLMs) have significantly improved the capabilities of web agents. However, effectively navigating complex and dynamic web environments still requires more advanced trajectory-level planning and execution. Prior studies have addressed self-improving agents by collecting extensive GUI trajectories from real-environment interactions. Despite their effectiveness, these approaches encounter two critical challenges: (1) Uncontrollable environment states, where real or sandboxed web environments often yield unstable and non-deterministic feedback, complicating the reproduction and debugging of agent behaviors; and (2) High API costs, as generating even a single interaction trajectory can involve hundreds of queries, leading to considerable API usage and computational expenses. To address these limitations and enable scalable self-improvement for agents, we propose WebSynthesis, a novel framework for trajectory synthesis and training. WebSynthesis leverages a learned world model to simulate virtual web environments, allowing a policy agent to perform efficient and reversible tree-based planning. This approach supports the large-scale generation of diverse and high-quality trajectories, which are subsequently utilized to refine the agent's policy. Experimental results demonstrate that an agent trained using WebSynthesis on a small-scale synthetic dataset achieves performance comparable to or even surpassing that of models trained on large-scale real-world data.

cs.AI↗

arXiv2025

ODE: Open-Set Evaluation of Hallucinations in Multimodal Large Language Models

Hallucination poses a persistent challenge for multimodal large language models (MLLMs). However, existing benchmarks for evaluating hallucinations are generally static, which may overlook the potential risk of data contamination. To address this issue, we propose ODE, an open-set, dynamic protocol designed to evaluate object hallucinations in MLLMs at both the existence and attribute levels. ODE employs a graph-based structure to represent real-world object concepts, their attributes, and the distributional associations between them. This structure facilitates the extraction of concept combinations based on diverse distributional criteria, generating varied samples for structured queries that evaluate hallucinations in both generative and discriminative tasks. Through the generation of new samples, dynamic concept combinations, and varied distribution frequencies, ODE mitigates the risk of data contamination and broadens the scope of evaluation. This protocol is applicable to both general and specialized scenarios, including those with limited data. Experimental results demonstrate the effectiveness of our protocol, revealing that MLLMs exhibit higher hallucination rates when evaluated with ODE-generated samples, which indicates potential data contamination. Furthermore, these generated samples aid in analyzing hallucination patterns and fine-tuning models, offering an effective approach to mitigating hallucinations in MLLMs.

cs.CL↗

arXiv2025

Towards Robust Recommendation: A Review and an Adversarial Robustness Evaluation Library

Recently, recommender system has achieved significant success. However, due to the openness of recommender systems, they remain vulnerable to malicious attacks. Additionally, natural noise in training data and issues such as data sparsity can also degrade the performance of recommender systems. Therefore, enhancing the robustness of recommender systems has become an increasingly important research topic. In this survey, we provide a comprehensive overview of the robustness of recommender systems. Based on our investigation, we categorize the robustness of recommender systems into adversarial robustness and non-adversarial robustness. In the adversarial robustness, we introduce the fundamental principles and classical methods of recommender system adversarial attacks and defenses. In the non-adversarial robustness, we analyze non-adversarial robustness from the perspectives of data sparsity, natural noise, and data imbalance. Additionally, we summarize commonly used datasets and evaluation metrics for evaluating the robustness of recommender systems. Finally, we also discuss the current challenges in the field of recommender system robustness and potential future research directions. Additionally, to facilitate fair and efficient evaluation of attack and defense methods in adversarial robustness, we propose an adversarial robustness evaluation library--ShillingREC, and we conduct evaluations of basic attack models and recommendation models. ShillingREC project is released at https://github.com/chengleileilei/ShillingREC.

cs.IR↗

arXiv2025

Prescribing the Right Remedy: Mitigating Hallucinations in Large Vision-Language Models via Targeted Instruction Tuning

Despite achieving outstanding performance on various cross-modal tasks, current large vision-language models (LVLMs) still suffer from hallucination issues, manifesting as inconsistencies between their generated responses and the corresponding images. Prior research has implicated that the low quality of instruction data, particularly the skewed balance between positive and negative samples, is a significant contributor to model hallucinations. Recently, researchers have proposed high-quality instruction datasets, such as LRV-Instruction, to mitigate model hallucination. Nonetheless, our investigation reveals that hallucinatory concepts from different LVLMs exhibit specificity, i.e. the distribution of hallucinatory concepts varies significantly across models. Existing datasets did not consider the hallucination specificity of different models in the design processes, thereby diminishing their efficacy in mitigating model hallucination. In this paper, we propose a targeted instruction data generation framework named DFTG that tailored to the hallucination specificity of different models. Concretely, DFTG consists of two stages: hallucination diagnosis, which extracts the necessary information from the model's responses and images for hallucination diagnosis; and targeted data generation, which generates targeted instruction data based on diagnostic results. The experimental results on hallucination benchmarks demonstrate that the targeted instruction data generated by our method are more effective in mitigating hallucinations compared to previous datasets.

cs.CV↗

arXiv2025

Mobile-Agent-V: A Video-Guided Approach for Effortless and Efficient Operational Knowledge Injection in Mobile Automation

The exponential rise in mobile device usage necessitates streamlined automation for effective task management, yet many AI frameworks fall short due to inadequate operational expertise. While manually written knowledge can bridge this gap, it is often burdensome and inefficient. We introduce Mobile-Agent-V, an innovative framework that utilizes video as a guiding tool to effortlessly and efficiently inject operational knowledge into mobile automation processes. By deriving knowledge directly from video content, Mobile-Agent-V eliminates manual intervention, significantly reducing the effort and time required for knowledge acquisition. To rigorously evaluate this approach, we propose Mobile-Knowledge, a benchmark tailored to assess the impact of external knowledge on mobile agent performance. Our experimental findings demonstrate that Mobile-Agent-V enhances performance by 36% compared to existing methods, underscoring its effortless and efficient advantages in mobile automation.

cs.CL↗

arXiv2025

Mobile-Agent-V: A Video-Guided Approach for Effortless and Efficient Operational Knowledge Injection in Mobile Automation

cs.AI↗

arXiv2025

AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models

Due to their multimodal capabilities, Vision-Language Models (VLMs) have found numerous impactful applications in real-world scenarios. However, recent studies have revealed that VLMs are vulnerable to image-based adversarial attacks. Traditional targeted adversarial attacks require specific targets and labels, limiting their real-world impact.We present AnyAttack, a self-supervised framework that transcends the limitations of conventional attacks through a novel foundation model approach. By pre-training on the massive LAION-400M dataset without label supervision, AnyAttack achieves unprecedented flexibility - enabling any image to be transformed into an attack vector targeting any desired output across different VLMs.This approach fundamentally changes the threat landscape, making adversarial capabilities accessible at an unprecedented scale. Our extensive validation across five open-source VLMs (CLIP, BLIP, BLIP2, InstructBLIP, and MiniGPT-4) demonstrates AnyAttack's effectiveness across diverse multimodal tasks. Most concerning, AnyAttack seamlessly transfers to commercial systems including Google Gemini, Claude Sonnet, Microsoft Copilot and OpenAI GPT, revealing a systemic vulnerability requiring immediate attention.

cs.LG↗

arXiv2025

MF-CLIP: Leveraging CLIP as Surrogate Models for No-box Adversarial Attacks

The vulnerability of Deep Neural Networks (DNNs) to adversarial attacks poses a significant challenge to their deployment in safety-critical applications. While extensive research has addressed various attack scenarios, the no-box attack setting where adversaries have no prior knowledge, including access to training data of the target model, remains relatively underexplored despite its practical relevance. This work presents a systematic investigation into leveraging large-scale Vision-Language Models (VLMs), particularly CLIP, as surrogate models for executing no-box attacks. Our theoretical and empirical analyses reveal a key limitation in the execution of no-box attacks stemming from insufficient discriminative capabilities for direct application of vanilla CLIP as a surrogate model. To address this limitation, we propose MF-CLIP: a novel framework that enhances CLIP's effectiveness as a surrogate model through margin-aware feature space optimization. Comprehensive evaluations across diverse architectures and datasets demonstrate that MF-CLIP substantially advances the state-of-the-art in no-box attacks, surpassing existing baselines by 15.23% on standard models and achieving a 9.52% improvement on adversarially trained models. Our code will be made publicly available to facilitate reproducibility and future research in this direction.

cs.LG↗

arXiv2025

Mind with Eyes: from Language Reasoning to Multimodal Reasoning

Language models have recently advanced into the realm of reasoning, yet it is through multimodal reasoning that we can fully unlock the potential to achieve more comprehensive, human-like cognitive capabilities. This survey provides a systematic overview of the recent multimodal reasoning approaches, categorizing them into two levels: language-centric multimodal reasoning and collaborative multimodal reasoning. The former encompasses one-pass visual perception and active visual perception, where vision primarily serves a supporting role in language reasoning. The latter involves action generation and state update within reasoning process, enabling a more dynamic interaction between modalities. Furthermore, we analyze the technical evolution of these methods, discuss their inherent challenges, and introduce key benchmark tasks and evaluation metrics for assessing multimodal reasoning performance. Finally, we provide insights into future research directions from the following two perspectives: (i) from visual-language reasoning to omnimodal reasoning and (ii) from multimodal reasoning to multimodal agents. This survey aims to provide a structured overview that will inspire further advancements in multimodal reasoning research.

cs.CL↗

arXiv2025

VaLiD: Mitigating the Hallucination of Large Vision Language Models by Visual Layer Fusion Contrastive Decoding

Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities in multimodal task reasoning. However, they often generate responses that appear plausible yet do not accurately reflect the visual content, a phenomenon known as hallucination. Recent approaches have introduced training-free methods to mitigate hallucinations by adjusting the decoding strategy during the inference stage, typically attributing hallucinations to the language model itself. Our analysis, however, reveals that distortions in the visual encoding process significantly affect the model's reasoning capabilities. Specifically, earlier visual layers may retain key features but gradually distort as the information propagates toward the output layer. Building on these insights, we propose a novel hallucination-mitigation method from the visual encoding perspective: \textbf{V}isu\textbf{a}l \textbf{L}ayer Fus\textbf{i}on Contrastive \textbf{D}ecoding (\textbf{VaLiD}). This method utilizes uncertainty to guide the visual layer selection, correcting distortions in the visual encoding process and thereby enhancing the reliability of the generated content. Experimental results demonstrate the effectiveness of VaLiD in mitigating hallucinations across various benchmarks, achieving state-of-the-art performance when compared to baseline methods. Codes are available at \href{https://github.com/RicardoLuL/VaLiD_LVLMs_hallucinations}{Github}.

cs.CV↗

arXiv2025

Unifying Perplexing Behaviors in Modified BP Attributions through Alignment Perspective

Attributions aim to identify input pixels that are relevant to the decision-making process. A popular approach involves using modified backpropagation (BP) rules to reverse decisions, which improves interpretability compared to the original gradients. However, these methods lack a solid theoretical foundation and exhibit perplexing behaviors, such as reduced sensitivity to parameter randomization, raising concerns about their reliability and highlighting the need for theoretical justification. In this work, we present a unified theoretical framework for methods like GBP, RectGrad, LRP, and DTD, demonstrating that they achieve input alignment by combining the weights of activated neurons. This alignment improves the visualization quality and reduces sensitivity to weight randomization. Our contributions include: (1) Providing a unified explanation for multiple behaviors, rather than focusing on just one. (2) Accurately predicting novel behaviors. (3) Offering insights into decision-making processes, including layer-wise information changes and the relationship between attributions and model decisions.

cs.LG↗

arXiv2025

Debiased Prompt Tuning in Vision-Language Model without Annotations

Prompt tuning of Vision-Language Models (VLMs) such as CLIP, has demonstrated the ability to rapidly adapt to various downstream tasks. However, recent studies indicate that tuned VLMs may suffer from the problem of spurious correlations, where the model relies on spurious features (e.g. background and gender) in the data. This may lead to the model having worse robustness in out-of-distribution data. Standard methods for eliminating spurious correlation typically require us to know the spurious attribute labels of each sample, which is hard in the real world. In this work, we explore improving the group robustness of prompt tuning in VLMs without relying on manual annotation of spurious features. We notice the zero - shot image recognition ability of VLMs and use this ability to identify spurious features, thus avoiding the cost of manual annotation. By leveraging pseudo-spurious attribute annotations, we further propose a method to automatically adjust the training weights of different groups. Extensive experiments show that our approach efficiently improves the worst-group accuracy on CelebA, Waterbirds, and MetaShift datasets, achieving the best robustness gap between the worst-group accuracy and the overall accuracy.

cs.CV↗