In spite of achieving significant progress in recent years, Large Vision-Language Models (LVLMs) are proven to be vulnerable to adversarial examples. Therefore, there is an urgent need for an effective adversarial att...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
In spite of achieving significant progress in recent years, Large Vision-Language Models (LVLMs) are proven to be vulnerable to adversarial examples. Therefore, there is an urgent need for an effective adversarial attack to identify the deficiencies of LVLMs in security-sensitive applications. However, existing LVLM attackers generally optimize adversarial samples against a specific textual prompt with a certain LVLM model, tending to overfit the target prompt/network and hardly remain malicious once they are transferred to attack a different prompt/model. To this end, in this paper, we propose a novel Imperceptible Transfer Attack (ITA) against LVLMs to generate prompt/model-agnostic adversarial samples to enhance such adversarial transferability while further improving the imperceptibility. Specifically, we learn to apply appropriate visual transformations on image inputs to create diverse input patterns by selecting the optimal combination of operations from a pool of candidates, consequently improving adversarial transferability. We conceptualize the selection of optimal transformation combinations as an adversarial learning problem and employ a gradient approximation strategy with noise budget constraints to effectively generate imperceptible transferable samples. Extensive experiments on three LVLM models and two widely used datasets with three tasks demonstrate the superior performance of our ITA.
In the field of video image processing, moving target detection remains a hot topic. To address the limitations of existing methods in complex environments, This paper proposes a novel TRPCA model based on Tensor Sing...
In the field of video image processing, moving target detection remains a hot topic. To address the limitations of existing methods in complex environments, This paper proposes a novel TRPCA model based on Tensor Singular Value Decomposition (T-SVD), incorporating the advantages of side information. Firstly, by imposing $$\gamma $$ -norm constraints, the method incorporates feature side information into the background component processing, addresses the over-penalization issue caused by the nuclear norm in traditional RPCA. Secondly, for the foreground part, $$L_{1,1,2}$$ norm and tensor total variation (TTV) regularization constraints are applied to enhance the model’s sensitivity to tubal sparsity and spatiotemporal continuity, effectively reducing the interference of dynamic backgrounds on foreground extraction. To solve this model, we employ the Alternating Direction Method of Multipliers (ADMM). Extensive experiments on the datasets CDnet2014 and LASIESTA demonstrate that the proposed method achieves optimal or near-optimal performance in terms of F-measure for the majority of cases, highlighting its superiority in foreground detection precision.
Vertical Federated Learning (VFL) enables the construction of models by combining clients with different features without compromising privacy. Existing VFL methods exhibit tightly coupled participant parameters, resu...
详细信息
Model merging offers an effective way to integrate the capabilities of multiple fine-tuned models. However, the performance degradation of the merged model remains a challenge, particularly when none or few data are a...
详细信息
The rapid advancement of AI-generated image (AGI) models has introduced significant challenges in evaluating their quality, which requires considering multiple dimensions such as perceptual quality, prompt corresponde...
详细信息
Mobile Edge Computing (MEC) offers computational services near data sources to meet numerous real-time data processing demands of end devices. Scheduling dependent tasks in resource-constrained environments is a key r...
详细信息
ISBN:
(数字)9798350368369
ISBN:
(纸本)9798350368376
Mobile Edge Computing (MEC) offers computational services near data sources to meet numerous real-time data processing demands of end devices. Scheduling dependent tasks in resource-constrained environments is a key research focus in MEC, aimed at enhancing the completion rates of applications within their deadlines. However, most existing works overlook the resource competition among requests arriving at different times, leading to decreased application completion rates. In this paper, we propose a dependent task online scheduling approach for multiple applications to optimize application completion rates. For multiple applications with deadline constraints, where tasks within each application may have dependencies, we propose a multi-priority task sequencing algorithm to determine the execution order of tasks. To accommodate scenarios where requests arrive at different times, we introduce a priority-based queue to dynamically adjust task execution order based on urgency and resource demands. Finally, by comparing with baseline approaches, experimental results demonstrate that our approach can improve the application completion rate by approximately 41.02%, demonstrating its effectiveness.
Previous research on option strategies has primarily focused on their behavior near expiration, with limited attention to the transient value process of the portfolio. In this paper, we formulate Iron Condor portfolio...
详细信息
Compared to single-turn dialogue, multi-turn dialogue involving multiple images better aligns with the needs of real-world human-AI interactions. Additionally, as training data, it provides richer contextual reasoning...
One of the biggest issues facing most municipalities worldwide is garbage management. Garbage forecasting, that includes dividing garbage into discrete categories for efficient recycling or disposal, is the main eleme...
详细信息
暂无评论