Large Language Model (LLM) Agents are advancing quickly, with the increasing leveraging of LLM Agents to assist in development tasks such as code generation. While LLM Agents accelerate code generation, studies indicate they may introduce adverse effects on development. However, existing metrics solely measure pass rates, failing to reflect impacts on long-term maintainability and readability, and failing to capture human intuitive evaluations of PR. To increase the comprehensiveness of this problem, we investigate and evaluate the characteristics of LLM to know the pull requests’ characteristics beyond the pass rate. We observe the code quality and maintainability within PRs based on code metrics to evaluate objective characteristics and developers’ reactions to the pull requests from both humans and LLM’s generation. Evaluation results indicate that LLM Agents frequently disregard code reuse opportunities, resulting in higher levels of redundancy compared to human developers. In contrast to the quality issues, our emotions analysis reveals that reviewers tend to express more neutral or positive emotions towards AI-generated contributions than human ones. This disconnect suggests that the surface-level plausibility of AI code masks redundancy, leading to the silent accumulation of technical debt in real-world development environments. Our research provides insights for improving human-AI collaboration.

Current parameter-efficient fine-tuning methods for adapting pre-trained language models to downstream tasks are susceptible to interference from noisy data. Conventional noise-handling approaches either rely on laborious data pre-processing or employ model architecture modifications prone to error accumulation. In contrast to existing noise-process paradigms, we propose a noise-robust adaptation method via asymmetric LoRA poisoning experts (LoPE), a novel framework that enhances model robustness to noise only with generated noisy data. Drawing inspiration from the mixture-of-experts architecture, LoPE strategically integrates a dedicated poisoning expert in an asymmetric LoRA configuration. Through a two-stage paradigm, LoPE performs noise injection on the poisoning expert during fine-tuning to enhance its noise discrimination and processing ability. During inference, we selectively mask the dedicated poisoning expert to leverage purified knowledge acquired by normal experts for noise-robust output. Extensive experiments demonstrate that LoPE achieves strong performance and robustness purely through the low-cost noise injection, which completely eliminates the requirement of data cleaning.

This letter introduces a pioneering, training-free and explainable framework for High-Resolution Range Profile (HRRP) automatic target recognition (ATR) utilizing large-scale pre-trained Large Language Models (LLMs). Diverging from conventional methods requiring extensive task-specific training or fine-tuning, our approach converts one-dimensional HRRP signals into textual scattering center representations. Prompts are designed to align LLMs’ semantic space for ATR via few-shot in-context learning, effectively leveraging its vast pre-existing knowledge without any parameter update.

Matrix inversion is critical in mathematics and scientific applications. Large-scale dense matrix inversion is especially challenging for modern computers due to its heavy dependency of matrix elements and the poor temporal data locality. In this paper, we propose a novel accelerator termed MI2D, which converts matrix inversion into regular matrix multiplications using 2-dimensional cross-tile operations and novel algorithms for efficient data reuse and computations. Our evaluations show that MI2D can be easily integrated with existing matrix engines in modern high-end CPU and NPU, and effectively improves matrix inversion with 2.7× speedup against Intel Skylake CPU, and 24× against NVIDIA RTX 2080 Ti.

Tags have been used in different social medias, such as Delicious, Flickr, LinkedIn and Weibo. In previous work, considerable efforts have been made to make use of tags without identification of their different types. In this study, we argue that tags in user profile indicate three different types of information, say the basics (age, status, locality, etc), interests and specialty of a person. Based on this novel user tag taxonomy, we propose a tag classification approach in Weibo to conduct a clearer image of user profiles, which makes use of three categories of features: general statistics feature (including user links with followers and followings), content feature and syntax feature. Furthermore, different from many previous studies on tag which concentrate on user specialties, such as expert finding, we find that valuable information can be discovered with the basics and interests user tags. We show some interesting findings in two scenarios, including user profiling with people come from different generations and area profiling with mass appeal, with large scale tag clustering and mining in over 6 million identical tags with 13 million users in Weibo data.

Cloud computing is a cost-effective solution to the current expensive IT infrastructure for healthcare but it has not been adopted due to the due to the concerns that current Cloud technologies are inadequate to protect sensitive information contents of health records in the outsourced environment. This paper focuses on how electronic health records can be protected in the Cloud environments deploying our proposed trust-oriented data protection framework. In particular, it introduces a technique for transforming raw electronic health records to active data cubes and for encoding data for storage and fast access. Implementation of the framework is also presented.

Recently cloud computing has been increasingly gaining attention and spreading in various industries. This novel paradigm also has received significant research focus on the data security and privacy issues due to its outsourced nature. However, few of the research efforts have been focused on data mobility issues which existed intrinsically in highly complex and dynamic cloud services. To address this challenge, this paper discusses full mobility management of cloud data from two aspects: data mobility management and user mobility management. The work is based on our previously proposed active data-centric framework. Taking advantage of this framework, the active data can establish an active binding relationship with its user. Any data mobility-related behavior could inform to its bundled user instantly and safely. The paper also discusses user mobility management via the self-adaptable heterogeneous network handoff mechanism to ensure the ubiquitous informing service in the mobile cloud environment. The experimental outcomes demonstrate the feasibility, proactivity, and efficiency by the full mobility management scheme.

We propose a new, active scheme for fast and reliable detection of P2P malware by exploiting the enemies' strength against them. Our new scheme works in two phases: host-level dynamic binary analysis to automatically extract built-in remotely-accessible/controllable mechanisms (referred to as Malware Control Birthmarks or MCB) in P2P malware, followed by network-level informed probing for detection. Our new design demonstrates a novel combination of the strengths from both host-based and network-based approaches. Compared with existing detection solutions, it is fast, reliable, and scalable in its detection scope. Furthermore, it can be applicable to more than just P2P malware, more broadly any malware that opens a service port for network communications (e.g., many Trojans/backdoors). We develop a prototype system, PeerPress, and evaluate it on many representative real-world P2P malware (including Storm, Conficker, and more recent Sality). The results show that it can effectively detect the existence of malware when MCBs are extracted, and the detection occurs in an early stage during which other tools (e.g., BotHunter) typically do not have sufficient information to detect. We further discuss its limitations and implications, and we believe it is a great complement to existing passive detection solutions.

Deploying state-of-the-art technologies is vital and inevitable in healthcare industry to cope with emerging services such as healthcare resource sharing and integration, collaborative consultation, and electronic health record. Cloud computing allows simple and easy user access, coping with users' dynamic and elastic demands, providing metered usage for its resources and hence is increasingly being adopted by individual users as well as enterprise users. The Cloud is being considered as appropriate technology for future healthcare infrastructure. However, in order to use of Cloud services effectively, users' data and/or resource have to be transferred to the cloud side and this inevitably raises several serious issues concerning losing control of users' resource, data privacy protection, data ownership and security. This paper addresses security and privacy challenges in healthcare cloud by deploying a novel framework with CPRBAC (Cloud-based Privacy-aware Role Based Access Control) model for controllability, traceability of data and authorized access to system resources. Furthermore, the work seeks to develop a unique active auditing service that is capable of tracing, tracking, and triggering an alarm on any operation, data or policy violations in the Cloud environment.

Despite cloud computing has been widely adopted by most industries, the healthcare industry still reveals a slow development in cloud-based solution due to the raising of user fear that their confidential health data or privacy would leak out in the cloud. To allay users' concern of data control, data ownership, security and privacy, we propose a robust data protection framework which is surrounded by a chain of protection schemes from access control, monitoring, to active auditing. The framework includes three key components which are Cloud-based Privacy-aware Role Based Access Control (CPRBAC) model, Triggerable Data File Structure (TDFS), and Active Auditing Scheme (AAS) respectively. Our schemes address controllability, traceability of data and authorize access to healthcare system resource. Data violation against access control policies can be proactively triggered to perform corresponding defense mechanisms. Our goal is to bring benefits of cloud computing to healthcare industries to assist them improve quality of service and reduce the cost of overall healthcare.

Deploying state-of-the-art technologies is vital and inevitable in assistive healthcare to cope with emerging services such as remote monitoring, collaborative consultation, and electronic health record. Grid computing has succeeded somewhat in enabling the sharing of resources across organizations but has not been deployed widely due to its complex implementation and interface. Cloud computing overcomes this aspect by allowing simple and easy user access, coping with users' dynamic and elastic demands, providing metered usage for its resources and hence is increasingly being adopted by individual users as well as enterprise users. The Cloud may just be the right technology for healthcare infrastructure. However, several serious issues concerning security, data protection and ownership, quality of services, and mobility need to be resolved before Cloud computing can be widely adopted. This paper proposes Mobile Cloud for Assistive Healthcare (MoCAsH) as an infrastructure for assistive healthcare. Besides inheriting the advantages of Cloud computing, MoCAsH embraces important concepts of mobile sensing, active sensor records, and collaborative planning by deploying intelligent mobile agents, context-aware middleware, and collaborative protocol for efficient resource sharing and planning. MoCAsH addresses security and privacy issues by deploying selective and federated P2P Cloud to protect data, preserve data ownership and strengthen aspects of security. It also addresses various quality-of-service issues concerning critical responses and energy consumption.