Recent advancements in Audio-Video Large Language Models (AV-LLMs) have enhanced their capabilities in tasks like audio-visual question answering and multimodal dialog systems. Video and audio introduce an extended temporal dimension, resulting in a larger key-value (KV) cache compared to static image embedding. A naive optimization strategy is to selectively focus on and retain KV caches of audio or video based on task. However, in the experiment, we observed that the attention of AV-LLMs to various modalities in the high layers is not strictly dependent on the task. In higher layers, the attention of AV-LLMs shifts more towards the video modality. In addition, we also found that directly integrating temporal KV of audio and spatial-temporal KV of video may lead to information confusion and significant performance degradation of AV-LLMs. If audio and video are processed indiscriminately, it may also lead to excessive compression or reservation of a certain modality, thereby disrupting the alignment between modalities. To address these challenges, we propose AccKV, an Adaptive-Focusing and Cross-Calibration KV cache optimization framework designed specifically for efficient AV-LLMs inference. Our method is based on layer adaptive focusing technology, selectively focusing on key modalities according to the characteristics of different layers, and enhances the recognition of heavy hitter tokens through attention redistribution. In addition, we propose a Cross-Calibration technique that first integrates inefficient KV caches within the audio and video modalities, and then aligns low-priority modality with high-priority modality to selectively evict KV cache of low-priority modality. The experimental results show that AccKV can significantly improve the computational efficiency of AV-LLMs while maintaining accuracy.

Xueyu Hu, Tao Xiong, Biao Yi, Zishu Wei, Ruixuan Xiao, Yurun Chen, Jiasheng Ye, Meiling Tao, Xiangxin Zhou, Ziyu Zhao, Yuhuai Li, Shengze Xu, Shenzhi Wang, Xinchen Xu, Shuofei Qiao, Zhaokai Wang, Kun Kuang, Tieyong Zeng, Liang Wang, Jiwei Li, Yuchen Eleanor Jiang, Wangchunshu Zhou, Guoyin Wang, Keting Yin, Zhou Zhao, Hongxia Yang, Fan Wu, Shengyu Zhang, Fei Wu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Federated Learning (FL) is popular as a privacy-preserving machine learning paradigm for generating a single model on decentralized data. However, statistical heterogeneity poses a significant challenge for FL. As a subfield of FL, personalized FL (pFL) has attracted attention for its ability to achieve personalized models that perform well on non-independent and identically distributed (Non-IID) data. However, existing pFL methods are limited in terms of leveraging the global model’s knowledge to enhance generalization while achieving personalization on local data. To address this, we proposed a new method personalized Federated learning with Adaptive Feature Aggregation and Knowledge Transfer (FedAFK), to train better feature extractors while balancing generalization and personalization for participating clients, which improves the performance of personalized models on Non-IID data. We conduct extensive experiments on three datasets in two widely-used heterogeneous settings and show the superior performance of our proposed method over thirteen state-of-the-art baselines.

As researchers continue to optimize AI agents for more effective task execution within operating systems, they often overlook a critical security concern: the ability of these agents to detect ''impostors'' within their environment. Through an analysis of the agents' operational context, we identify a significant threat-attackers can disguise malicious attacks as environmental elements, injecting active disturbances into the agents' execution processes to manipulate their decision-making. We define this novel threat as the Active Environment Injection Attack (AEIA). Focusing on the interaction mechanisms of the Android OS, we conduct a risk assessment of AEIA and identify two critical security vulnerabilities: (1) Adversarial content injection in multimodal interaction interfaces, where attackers embed adversarial instructions within environmental elements to mislead agent decision-making; and (2) Reasoning gap vulnerabilities in the agent's task execution process, which increase susceptibility to AEIA attacks during reasoning. To evaluate the impact of these vulnerabilities, we propose AEIA-MN, an attack scheme that exploits interaction vulnerabilities in mobile operating systems to assess the robustness of MLLM-based agents. Experimental results show that even advanced MLLMs are highly vulnerable to this attack, achieving a maximum attack success rate of 93% on the AndroidWorld benchmark by combining two vulnerabilities.

Multiparty private set intersection (mPSI) protocol is capable of finding the intersection of multiple sets securely without revealing any other information. However, its limitation lies in processing only those elements present in every participant's set, which proves inadequate in scenarios where certain elements are common to several, but not all, sets.

As blockchain technology becomes more and more popular, a typical financial scam, the Ponzi scheme, has also emerged in the blockchain platform Ethereum. This Ponzi scheme deployed through smart contracts, also known as the smart Ponzi scheme, has caused a lot of economic losses and negative impacts. Existing methods for detecting smart Ponzi schemes on Ethereum mainly rely on bytecode features, op-code features, account features, and transaction behavior features of smart contracts, and the performance of identifying schemes is insufficient. In this paper, we propose SourceP, a method to detect smart Ponzi schemes on the Ethereum platform using pre-trained models and data flow, which only requires using the source code of smart contracts as features. SourceP reduces the difficulty of data acquisition and feature extraction of existing detection methods. Specifically, we first convert the source code of a smart contract into a data flow graph and then introduce a pre-trained model based on learning code representations to build a classification model to identify Ponzi schemes in smart contracts. The experimental results show that SourceP achieves 87.2% recall and 90.7% F-score for detecting smart Ponzi schemes within Ethereum’s smart contract dataset, outperforming state-of-the-art methods in terms of performance and sustainability.

The public sharing of user information opens the door for adversaries to infer private data, leading to privacy breaches and facilitating malicious activities. While numerous studies have concentrated on privacy leakage via public user attributes, the threats associated with the exposure of user relationships, particularly through network structure, are often neglected. This study aims to fill this critical gap by advancing the understanding and protection against privacy risks emanating from network structure, moving beyond direct connections with neighbors to include the broader implications of indirect network structural patterns. To achieve this, we first investigate the problem of Graph Privacy Leakage via Structure (GPS), and introduce a novel measure, the Generalized Homophily Ratio, to quantify the various mechanisms contributing to privacy breach risks in GPS. Based on this insight, we develop a novel graph private attribute inference attack, which acts as a pivotal tool for evaluating the potential for privacy leakage through network structures under worst-case scenarios. To protect users' private data from such vulnerabilities, we propose a graph data publishing method incorporating a learnable graph sampling technique, effectively transforming the original graph into a privacy-preserving version. Extensive experiments demonstrate that our attack model poses a significant threat to user privacy, and our graph data publishing method successfully achieves the optimal privacy-utility trade-off compared to baselines.

In recent years, Non-Fungible Token (NFT) has gradually become the key application of blockchain technology. Static NFT is the most common type of NFT. Once static NFT is minted on the blockchain, its additional metadata is immutable. However, some NFTs that mark real assets, games, sports, and other types must dynamically update the metadata. Therefore, a dynamic NFT with changeable features is needed. The emergence of dynamic NFT has greatly expanded the application innovation scene, and promoted the rapid development of community ecology, but also brought new problems and challenges to anti-fraud and supervision. This paper aims to realize the classification and detection of dynamic NFT. First, define and classify dynamic NFTs from both dynamic and static perspectives. Second, a complete dataset of dynamic NFT smart contract codes on Ethereum was constructed for the first time, and analyzed from multiple perspectives. Third, a smart contract feature model of dynamic NFT is proposed, and machine learning methods are used for recognition and classification. After experimental verification, the method proposed in this article can be effectively used to detect and identify dynamic NFTs, helping NFT holders avoid risks.

In order to curb the illegal activities in power resources detection, improve the credibility and contribution rate of the industry, and promote the development of high-quality services, this paper proposes a secure and reliable trusted testing system of electric power materials based on blockchain. Firstly, a device and personal information query authorization mechanism is established to provide solutions for personnel and testing equipment authorization. It can help ensure the reliability of testing data on the premise of security. Secondly, we propose a method to deal with the difficulties of testing information management. Lastly, we introduce the case of electricity management helping the power authorities to supervise effectively and increasing the credibility of power material procurement evidence to prove the feasibility of this system.

To support the migration of software applications to the cloud environment, cloud venders have proposed different migration methodologies and guidelines. Yet, most of them require human intervention, involving manually performing repetitive tasks. This paper proposes a pattern-based transformation approach for cloud application migration. The approach automatically modifies the source code of an application before the migration, to make it cloud-ready, and then transforms the source code to the target code in the cloud environment. The approach is supported by three key elements (patterns, rules and templates) and a process that systematically applies these elements. First, a pattern matching engine based on a regular expression processing technique is used to identify the parts of the source code that require modification and to extract the essential tokens from the source code for code transformation. Next, transformation rules are invoked to change the source code into the target code using a template, designed according to the target cloud environment. The proposed approach has been demonstrated on 19 open-source projects, by migrating them to Amazon Web Services.

A well arranged task priority in dual-shore collaborative software design phase will smooth the collaboration and reduces the potential coordination lag. However, it is really difficult to determine the priority when there are many collaboration tasks and the collaborative process is a little complicated. On the basis of behavior compatibility analysis using Petri-net, this paper proposed yet a new automatic method to determine the priority. The collaborative design is modeled in Petri-net and some algorithms are presented for the automatic priority determining.