Infrared unmanned aerial vehicle (UAV) target images often suffer from motion blur degradation caused by rapid sensor movement, significantly reducing contrast between target and background. Generally, detection performance heavily depends on the discriminative feature representation between target and background. Existing methods typically treat deblurring as a preprocessing step focused on visual quality, while neglecting the enhancement of task-relevant features crucial for detection. Improving feature representation for detection under blur conditions remains challenging. In this paper, we propose a novel Joint Feature-Domain Deblurring and Detection end-to-end framework, dubbed JFD³. We design a dual-branch architecture with shared weights, where the clear branch guides the blurred branch to enhance discriminative feature representation. Specifically, we first introduce a lightweight feature restoration network, where features from the clear branch serve as feature-level supervision to guide the blurred branch, thereby enhancing its distinctive capability for detection. We then propose a frequency structure guidance module that refines the structure prior from the restoration network and integrates it into shallow detection layers to enrich target structural information. Finally, a feature consistency self-supervised loss is imposed between the dual-branch detection backbones, driving the blurred branch to approximate the feature representations of the clear one. We also construct a benchmark, named IRBlurUAV, containing 30,000 simulated and 4,118 real infrared UAV target images with diverse motion blur. Extensive experiments on IRBlurUAV demonstrate that JFD³ achieves superior detection performance while maintaining real-time efficiency.

Knowledge-aware recommendation leverages rich item-related factual information in Knowledge Graphs (KGs) to enhance recommendation systems. However, most existing methods focus on developing complex models to extract information from a given KG. They essentially follow a model-centric paradigm, overlooking data quality problems. In practice, KG data exhibits two principal quality problems, namely the noisy knowledge problem and the incomplete knowledge problem, which severely impair the performance of downstream models. To address these problems, we adopt a data-centric paradigm to improve the quality of KG data. Inspired by diffusion models' superior denoising and generation ability by fitting true data distributions, we propose a novel spectral heterogeneous diffusion framework for knowledge-aware recommendation. This framework tailors a diffusion model to capture the recommendation-oriented heterogeneous distribution in the original KG and then converts the fitted distribution into a high-quality KG. Specifically, we design a spectral heterogeneous diffusion model that integrates recommendation prior knowledge to capture task-relevant distribution and aligns its diffusion process with the features of heterogeneous graphs to model heterogeneity. Furthermore, we propose a continuous-discrete mode adapter that transforms the learned continuous distribution into a high-quality discrete KG. The resulting KG is denoised and enriched with task-relevant triples, mitigating noisy and incomplete knowledge problems. Experiments show that our plug-and-play framework can be integrated with any knowledge-aware recommendation model and boost their performance by improving KG quality. The code and theoretical analyses are available at https://github.com/xiangmli/SHGD.

• Plang: A language blending natural prompt with control flow for precise LLM guidance. • Meta-prompt programming: Enables LLM to self-modify prompt programs during execution. • 40.75%-89.55% conciseness gain: Outperforms methods like LangChain in prompt coding. • Open-source solution: Enables multi-agent collaboration & tool use with minimal code. The advent of instruction-following large language models (LLMs), exemplified by ChatGPT, has significantly enhanced the performance of generative autoregressive natural language models on general tasks, marking a crucial milestone toward artificial general intelligence. While research has shown that LLM performance critically depends on prompt effectiveness, existing prompt construction approaches - including prompt string templates, prompt programming frameworks, and prompt programming syntactic sugar - suffer from limitations in imprecise generation control and deviation from natural language syntax, thereby impeding prompt engineering advancement. To address these challenges, we introduce PromptLanguage (Plang), a string-first programming language designed specifically for LLM prompt engineering. Our key innovation lies in utilizing font styles as syntax keywords to seamlessly integrate natural language prompt text with control flow code, enabling precise intervention in the generation process while maintaining the natural language affinity of prompt programs. Notably, Plang pioneers the concept of meta-prompt programming. Extensive experimental results across prompt engineering cases and quantitative analyses demonstrate that Plang-written prompt programs offer superior read/writability, intervention precision and up to 89.55% efficiency improvements. The Plang implementation is available as open-source software at https://github.com/HJZ-XDU/plang .

In recent years, edge networks have been deployed broadly at large scale, hosting a wide variety of services. Among these, microservices have emerged as one of the predominant service paradigms. Typically, microservices run on edge servers with varying configurations, while new microservice instances are usually online generated and join in edge networks due to the dynamic attributes of requests and networks. In those cases, an effective microservice online deployment solution are expected to be vital to system performance. So it becomes a critical issue to design online deployment solutions for microservices in edge. Existing research has always focused on offline deployment of microservices. However, edge networks are characterized by dynamics, real time, and concurrency, and when the environments or requests change, traditional offline deployment solutions usually cannot handle the deployment task in such cases. To address these issues, we carry out a comprehensive investigation on those potential influencing factors in edge, fully covering deployment cost, load balance, packet loss, and network delay. We further develop an innovative holistic online deployment solution that encompasses a system model, constraint analysis, multiobjective optimization, and a deep reinforcement learning algorithm. We conducted extensive experiments and evaluated our solution over a set of metrics using a real-world microservice prototype system. The results show that our online deployment solution produces superior performance, for example, reducing deployment cost by an average of 70.14% compared to all baselines. We also evaluated our solution under varying volumes of requests and gave analysis for performance stability and parameter sensitivity. We have released the code on GitHub.

Infrared unmanned aerial vehicle (UAV) images captured using thermal detectors are often affected by temperature-dependent low-frequency nonuniformity, which significantly reduces the contrast of the images. Detecting UAV targets under nonuniform conditions is crucial in UAV surveillance applications. Existing methods typically treat infrared nonuniformity correction (NUC) as a preprocessing step for detection, which leads to suboptimal performance. Balancing the two tasks while enhancing detection-beneficial information remains challenging. In this paper, we present a detection-friendly union framework, termed UniCD, that simultaneously addresses both infrared NUC and UAV target detection tasks in an end-to-end manner. We first model NUC as a small number of parameter estimation problem jointly driven by priors and data to generate detection-conducive images. Then, we incorporate a new auxiliary loss with target mask supervision into the backbone of the infrared UAV target detection network to strengthen target features while suppressing the background. To better balance correction and detection, we introduce a detection-guided self-supervised loss to reduce feature discrepancies between the two tasks, thereby enhancing detection robustness to varying nonuniformity levels. Additionally, we construct a new benchmark composed of 50,000 infrared images in various nonuniformity types, multi-scale UAV targets and rich backgrounds with target annotations, called IRBFD. Extensive experiments on IRBFD demonstrate that our UniCD is a robust union framework for NUC and UAV target detection while achieving real-time processing capabilities. Dataset can be available at https://github.com/IVPLaboratory/UniCD.

Skeleton sequences for action recognition are with complex temporal dynamics due to various factors such as speed variation and different activities. It is crucial and essential to model variation changes in the temporal dimension. In recent years, skeleton sequence is always modeled as a graph structure, and Graph Convolution Network (GCN) is employed to extract spatial and temporal features of actions. Though GCN has obtained great achievements, they typically employ fixed-size temporal kernels for temporal modeling, which ignore the complex temporal dynamic of actions, especially for long-term as well as short-term modeling. To capture this complex motion pattern effectively, we propose a Dual Multi-Scale Graph Convolutional Network (DMS-GCN), which is mainly composed of a Deformable Temporal Kernel (DTK) block and a dual multi-scale strategy. Specifically, the DTK block is proposed to flexibly capture complex temporal information of the skeleton sequence. And the dual multi-scale strategy is used to simultaneously accommodate long-term and short-term dynamic information at different scales globally as well as locally. The effectiveness of our proposed method is verified through experiments conducted on two widely used datasets, NTU-RGB+D 60 and NTU-RGB+D 120.

Microservice architecture is used by leading companies to develop their large-scale software systems. These systems comprise numerous nodes, diverse service types and instances, and substantial volumes of data. Current research usually requires a central node to collect massive data from the system to build an anomaly detection model, encountering two significant limitations: 1) Most research trains a model for the entire system, ignoring the unique characteristics of individual nodes. Additionally, processing vast system-wide data in a single node imposes significant resource demands. 2) Microservice systems change frequently, and the historical data distribution differs significantly from the real data distribution, resulting in concept drift. Thus, we proposes DistriAD, a distributed anomaly detection method specifically designed for large-scale microservice systems. DistriAD involves a lightweight anomaly detection model deployed on each distributed node for precise anomaly detection, thus enhancing its accuracy. Furthermore, DistriAD utilizes a federated learning framework and a continuous updating method incorporating human feedback to update model parameters and address concept drift. Experimental validation on public datasets, e.g., TrainTicket-based and GAIA, and a proprietary test system dataset demonstrate that DistriAD outperforms baseline methods, improving F1-score up to 39.2 %. We believe that this work can provide insights into distributed anomaly detection in large-scale microservice systems, thereby improving their performance.

Microservice resource management aims to ensure stable service instance loads and improve overall resource utilization through load balancing, elastic scaling, and container orchestration while meeting system service quality requirements.Existing research often focuses on localized solutions, addressing only single aspects of elastic scaling or load balancing without recognizing the systemic nature of microservice resource management.The complexity and dynamism of service dependencies make it challenging to quantify interactions between resource strategies and service loads.Additionally, microservice systems experience dynamic load variations influenced by user behavior, business activities, and external factors.The dynamic nature of load changes and the selection of load features significantly increase the difficulty of resource forecasting, further complicating microservice resource management.To address this problem, this paper proposes MDRM (MAPEbased Dynamic Resource Management), a dynamic optimization method that integrates load balancing and elastic scaling to overcome the limitations of isolated strategies.MDRM models system load based on business characteristics and service invocation relationships, accurately capturing dynamic variations.A composite model, combining parallel multi-layer CNNs and LSTMs, extracts spatiotemporal microservice features, enhancing resource forecasting accuracy.Additionally, MDRM formulates a comprehensive load balancing optimization function that synergizes with resource utilization and service response time objectives to generate optimal management strategies.Experimental results demonstrate that, compared to default resource management strategies in Docker Swarm and Kubernetes, MDRM significantly improves system throughput (approximately 1000 RPS) and reduces response time (approximately 30-40 ms), proving its effectiveness.

Code review is essential for maintenance of pull request-based software systems. Recommending suitable reviewers for code changes can enhance defect detection and knowledge dissemination. Despite extensive research, the inherent complexity of pull requests (PRs) and reviewer profiles continues to cause challenge for accurate matching them together. Furthermore, existing methods often amplify gender and racial/ethnic disparities due to the lack of attention to biases present in historical review records. To address these issues, we first collected a dataset from 4 large-scale open-source projects involving 50 -month revision history, reaching up to 30 attributes. This dataset includes gender and racial/ethnic information, which was inferred, validated, and incorporated to enable comprehensive data bias analysis in reviewer recommendation tasks. Additionally, we introduce a fairness-aware and accurate approach: CoReBM, which leverages the advanced semantic understanding capabilities of Large Language Models (LLMs) to comprehensively capture the nuanced textual context of both PRs and reviewers, utilizing the robust planning, collaborative, and decision-making abilities of multi-agent systems. CoReBM integrates diverse factors to improve recommendation performance while mitigating bias effects through the incorporation of candidates' gender and racial/ethnic attributes. We evaluate the effectiveness of our approach on this dataset, and the results demonstrate that CoReBM outperforms state-of-the-art methods in both accuracy and fairness in recommendation.

Modern enterprises are increasingly adopting microservice architectures to enhance system flexibility and scalability. However, in the face of ever-changing business requirements, the relationships between system components have become increasingly complex, resulting in significant challenges in maintaining system robustness. In recent years, multimodal data-driven approaches based on graph neural networks have emerged as a predominant solution for root cause localization in microservice systems. Our detailed analysis of architectural characteristics and existing research reveals two critical limitations. First, simple graph is insufficient to represent the one-to-many relationships inherent in microservice component interactions, such as deployment, subordinate, and dependency. Second, the current multimodal data-based method has difficulty in performing localization on faults occurring on hosts, services, and instances at the same time.To address these challenges, we propose HyperRCA, a novel multi-granular root cause analysis approach based on hypergraph neural networks. Our approach models system states during faults via a hypergraph with instances as graph nodes, explicitly capturing heterogeneous relationships through three innovative hyperedge designs: deployment hyperedges for infrastructure relationships, subordinate hyperedges for service hierarchies, and dependency hyperedges for inter-component interactions. We used hypergraph neural networks and multi-layer perceptrons to train a root cause localization model based on hyperedge features to achieve multi-granularity root cause localization. Experimental evaluations demonstrate significant performance improvements over state-of-the-art approaches. HyperRCA achieves a maximum HR@5 improvement of 112.62% on single-granularity datasets and 466.43% in multi-granularity scenarios.

Adaptive learning, which requires the in-depth understanding of students' learning processes and rational planning of learning resources, plays a crucial role in intelligent education. However, how to effectively model these two processes and seamlessly integrate them poses significant implementation challenges for adaptive learning. As core learning resources, exercises have the potential to diagnose students' knowledge states during the learning processes and provide personalized learning recommendations to strengthen students' knowledge, thereby serving as a bridge to boost student-oriented adaptive learning. Therefore, we introduce a novel task called Knowledge-aware Exercise Generative Recommendation (KEGR). It aims to dynamically infer students' knowledge states from their past exercise responses and customizably generate new exercises. To achieve KEGR, we propose an adaptive multi-agent cooperation framework, called ExeGen, inspired by the excellent reasoning and generative capabilities of LLM-based AI agents. Specifically, ExeGen coordinates four specialized agents for supervision, knowledge state perception, exercise generation, and quality refinement through an adaptive loop workflow pipeline. More importantly, we devise two enhancement mechanisms in ExeGen: 1) A human-simulated knowledge perception mechanism mimics students' cognitive processes and generates interpretable knowledge state descriptions via demonstration-based In-Context Learning (ICL). In this mechanism, a dual-matching strategy is further designed to retrieve highly relevant demonstrations for reliable ICL reasoning. 2) An exercise generation-adversarial mechanism collaboratively refines exercise generation leveraging a group of quality evaluation expert agents via iterative adversarial feedback. Finally, a comprehensive evaluation protocol is carefully designed to assess ExeGen. Extensive experiments on real-world educational datasets and a practical deployment in college education demonstrate the effectiveness and superiority of ExeGen. The code is available at https://github.com/dsz532/exeGen.

Recently, the real-world applications of reinforcement learning (RL) have seen the problem of taking actions in an out-of-distribution (OOD) state. However, most existing research is limited to take actions to narrow the visited training distribution and OOD, and does not consider the efficiency to choose such actions. In this paper, we propose a novel approach, called Self-Supervised Reinforcement Learning for OOD recovery via Auxiliary Reward (SRL-AR), to address this issue. By leveraging cumulative reward, we force the representations to discriminate state-action pairs with different returns as auxiliary task. Then, the auxiliary reward calculated from the auxiliary loss is used to generate a new policy that can effectively handle OOD situations. Moreover, we show that our method outperforms prior works in terms of asymptotic performance and sample efficiency on MuJoCo tasks.

Excessively large classes that encapsulate multiple responsibilities are challenging to comprehend and maintain. Addressing this issue, several Extract Class refactoring tools have been proposed, employing a two-phase process: identifying suitable fields or methods for extraction, and implementing the mechanics of refactoring. These tools traditionally generate an intra-class dependency graph to analyze the class structure, applying hard-coded rules based on this graph to unearth refactoring opportunities. Yet, the graph-based approach predominantly illuminates direct, “one-to-one” relationship between pairwise entities. Such a perspective is restrictive as it overlooks the complex, “one-to-many” dependencies among multiple entities that are prevalent in real-world classes. This narrow focus can lead to refactoring suggestions that may diverge from developers’ actual needs, given their multifaceted nature. To bridge this gap, our paper leverages the concept of intra-class dependency hypergraph to model one-to-many dependency relationship and proposes a hypergraph learning-based approach to suggest Extract Class refactoring opportunities named HECS. For each target class, we first construct its intra-class dependency hypergraph and assign attributes to nodes with a pre-trained code model. All the attributed hypergraphs are fed into an enhanced hypergraph neural network for training. Utilizing this trained neural network alongside a large language model (LLM), we construct a refactoring suggestion system. We trained HECS on a large-scale dataset and evaluated it on two real-world datasets. The results show that demonstrates an increase of 38.5% in precision, 9.7% in recall, and 44.4% in f1-measure compared to 3 state-of-the-art refactoring tools including JDeodorant, SSECS, and LLMRefactor, which is more useful for 64% of participants. The results also unveil practical suggestions and new insights that benefit existing extract-related refactoring techniques.

HECS is an advanced tool designed for Extract Class refactoring by leveraging hypergraph learning to model complex dependencies within large classes. Unlike traditional tools that rely on direct one-to-one dependency graphs, HECS uses intra-class dependency hypergraphs to capture one-to-many relationships. This allows HECS to provide more accurate and relevant refactoring suggestions. The tool constructs hypergraphs for each target class, attributes nodes using a pre-trained code model, and trains an enhanced hypergraph neural network. Coupled with a large language model, HECS delivers practical refactoring suggestions. In evaluations on large-scale and real-world datasets, HECS achieved a 38.5% increase in precision, 9.7% in recall, and 44.4% in f1-measure compared to JDeodorant, SSECS, and LLMRefactor. These improvements make HECS a valuable tool for developers, offering practical insights and enhancing existing refactoring techniques.

Methods implemented in incorrect classes will cause excessive reliance on other classes than their own, known as a typical code smell symptom: feature envy, which makes it difficult to maintain increased coupling between classes. Addressing this issue, several Move Method refactoring tools have been proposed, employing a two-phase process: identifying misplaced methods to move and appropriate classes to receive, and implementing the mechanics of refactoring. These tools traditionally use hard-coded metrics to measure correlations between movable methods and target classes and apply heuristic thresholds or trained classifiers to unearth refactoring opportunities. Yet, these approaches predominantly illuminate pairwise correlations between methods and classes while overlooking the complex and complicated dependencies binding multiple code entities within these methods/classes that are prevalent in real-world cases. This narrow focus can lead to refactoring suggestions that may diverge from developers' actual needs. To bridge this gap, our paper leverages the concept of inter-class code entity dependency hypergraph to model complicated dependency relationships involving multiple code entities within various methods/classes and proposes a hypergraph learning-based approach to suggest Move Method refactoring opportunities named HMove. We first construct inter-class code entity dependency hypergraphs from training samples and assign attributes to entities with a pre-trained code model. All the attributed hypergraphs are fed into a hybrid hypergraph neural network for training. Utilizing this trained neural network alongside a large language model, we construct a refactoring suggestion system. We trained HMove on a large-scale dataset and evaluated it on two real-world datasets. The results show that demonstrates an increase of 27.8% in precision, 2.5% in recall, and 18.5% in f1-measure compared to 9 state-of-the-art refactoring tools, which is more useful for 68% of participants. The results also unveil practical suggestions and new insights that benefit existing feature envy-related refactoring techniques.

Assigning pull requests to appropriate code reviewers can accelerate the review process and help uncover potential bugs. However, the inherent complexities in pull requests and code reviewers present challenges in making suitable matches between them. Prior studies focus on mining rich semantic information from pull requests or profile information from code reviewers to improve efficiency. These approaches often overlook the intrinsic relationships between pull requests and code reviewers, which can be represented by a combination of multiple factors and strategies, resulting in suboptimal recommendation accuracy.

Abstract Bug triaging is a vital process in software maintenance, involving assigning bug reports to developers in the issue tracking system. Current studies predominantly treat automatic bug triaging as a classification task, categorizing bug reports using developers as labels. However, this approach deviates from the essence of triaging, which is establishing bug–developer correlations. These correlations should be explicitly leveraged, offering a more comprehensive and promising paradigm. Our bug triaging model utilizes graph collaborative filtering (GCF), a method known for handling correlations. However, GCF encounters two challenges in bug triaging: data sparsity in bug fixing records and semantic deficiency in exploiting input data. To address them, we propose PCG, an innovative framework that integrates prototype augmentation and contrastive learning with GCF. With bug triaging modeled as predicting links on the bipartite graph of bug–developer correlations, we introduce prototype clustering‐based augmentation to mitigate data sparsity and devise a semantic contrastive learning task to overcome semantic deficiency. Extensive experiments against competitive baselines validate the superiority of PCG. This work may open new avenues for investigating correlations in bug triaging and related scenarios.

Extract Method is one of the most frequently performed refactoring operations for the decomposition of large and complex methods, which can also be combined with other refactoring operations to remove a variety of design flaws. Several Extract Method refactoring tools have been proposed based on the quantification of extraction criteria. To the best of our knowledge, state-of-the-art related techniques can be broadly divided into two categories: the first line is non-machine-learning-based approaches built on heuristics, and the second line is machine learning-based approaches built on historical data. Most of these approaches characterize the extraction criteria by deriving software metrics from fine-grained code properties. However, in most cases, these metrics can be challenging to concretize, and their selections and thresholds also largely rely on expert knowledge. Thus, in this paper, we propose an approach to automatically recommend Extract Method refactoring opportunities named REMS via mining multi-view representations from code property graph. We fuse various representations together using compact bilinear pooling and further train machine learning classifiers to guide the extraction of suitable lines of code as new method. We evaluate our approach on two publicly available datasets. The results show that our approach outperforms five state-of-the-art refactoring tools including GEMS, JExtract, SEMI, JDeodorant, and Segmentation in effectiveness and usefulness. Our approach demonstrates an increase of 29% in precision, 15% in recall, and 23% in f1-measure. The results also unveil practical suggestions and provide new insights that benefit additional extract-related refactoring techniques.

Multi-scale infrared unmanned aerial vehicle (UAV) targets (IRUTs) detection under dynamic scenarios remains a challenging task due to weak target features, varying shapes and poses, and complex background interference. Current detection methods find it difficult to address the above issues accurately and efficiently. In this paper, we design a dynamic attentive network (DANet) incorporating a scale-adaptive feature enhancement mechanism (SaFEM) and an attention-guided cross-weighting feature aggregator (ACFA). The SaFEM adaptively adjusts the network's receptive fields at hierarchical network levels leveraging separable deformable convolution (SDC), which enhances the network's multi-scale IRUT awareness. The ACFA, modulated by two crossing attention mechanisms, strengthens structural and semantic properties on neighboring levels for the accurate representation of multi-scale IRUT features from different levels. A plug-and-play anti-distractor contrastive regularization (ADCR) is also imposed on our DANet, which enforces similarity on features of targets and distractors from a new uncompressed feature projector (UFP) to increase the network's anti-distractor ability in complex backgrounds. To further increase the multi-scale UAV detection performance of DANet while maintaining its efficiency superiority, we propose a novel scale-specific knowledge distiller (SSKD) based on a divide-and-conquer strategy. For the "divide'' stage, we intendedly construct three task-oriented teachers to learn tailored knowledge for small-, medium-, and large-scale IRUTs. For the "conquer'' stage, we propose a novel element-wise attentive distillation module (EADM), where we employ a pixel-wise attention mechanism to highlight teacher and student IRUT features, and incorporate IRUT-associated prior knowledge for the collaborative transfer of refined multi-scale IRUT features to our DANet. Extensive experiments on real infrared UAV datasets demonstrate that our DANet is able to detect multi-scale UAVs with a satisfactory balance between accuracy and efficiency.

Owing to the high complexity of chip architecture and assay protocol, considerable effort has been directed toward the design automation of continuous-flow microfluidics over the past decade. Existing methods, however, perform the corresponding design tasks, including binding, scheduling, placement, and routing separately, leading to serious gaps between different steps and potentially even cause design failure. To overcome these drawbacks, in this article, we propose a one-pass design paradigm for continuous-flow microfluidic lab-on-a-chip systems, integrating all the design steps into an “organic whole,” which has never been considered in prior work. With the proposed paradigm, all the design tasks can be synchronized seamlessly and performed in a combined manner, thereby eliminating the gaps between design steps. Consequently, optimized biochip architectures can be generated without any design adjustments and modifications. The experimental results demonstrate the effectiveness of the proposed automation flows.

Background: Many safety-critical industrial applications have turned to deep learning systems as a fundamental component. Most of these systems rely on deep learning libraries, and bugs of such libraries can have irreparable consequences. Aims: Over the years, dependency structure has shown to be a practical indicator of software quality, widely used in numerous bug prediction techniques. The problem is that when analyzing bugs in deep learning libraries, researchers are unclear whether dependency structures still have a high correlation and which forms of dependency structures perform the best. Method: In this paper, we present a systematic investigation of the above question and implement a dependency structure-centric bug analysis tool: Depend4BL, capturing the interaction between dependency structures and bug locations in deep learning libraries. Results: We employ Depend4BL to analyze the top 5 open-source deep learning libraries on Github in terms of stars and forks, with 279,788 revision commits and 8,715 bug fixes. The results demonstrate the significant differences among syntactic, history, and semantic structures, and their vastly different impacts on bug locations. Their combinations have the potential to further improve bug prediction for deep learning libraries. Conclusions: In summary, our work provides a new perspective regarding to the correlation between dependency structures and bug locations in deep learning libraries. We release a large set of benchmarks and a prototype toolkit to automatically detect various forms of dependency structures for deep learning libraries. Our study also unveils useful findings based on quantitative and qualitative analysis that benefit bug prediction techniques for deep learning libraries.

Incorrect placement of methods within classes is a typical code smell called Feature Envy, which causes additional maintenance and cost during evolution. To remove this design flaw, several Move Method refactoring tools have been proposed. To the best of our knowledge, state-of-the-art related techniques can be broadly divided into two categories: the first line is non-machine-learning-based approaches built on software measurement, while the selection and thresholds of software metrics heavily rely on expert knowledge. The second line is machine learning-based approaches, which suggest Move Method refactoring by learning to extract features from code information. However, most approaches in this line treat different forms of code information identically, disregarding their significant variation on data analysis. In this paper, we propose an approach to recommend Move Method refactoring named RMove by automatically learning structural and semantic representation from code fragment respectively. We concatenate these representations together and further train the machine learning classifiers to guide the movement of method to suitable classes. We evaluate our approach on two publicly available datasets. The results show that our approach outperforms three state-of-the-art refactoring tools including PathMove, JDeodorant, and JMove in effectiveness and usefulness. The results also unveil useful findings and provide new insights that benefit other types of feature envy refactoring techniques.

Recent advances in continuous-flow microfluidics have enabled highly integrated lab-on-a-chip biochips. These chips can execute complex biochemical applications precisely and efficiently within a tiny area, but they require a large number of control ports and the corresponding control logic to generate required pressure patterns for flow control, which, consequently, offset their advantages and prevent their wide adoption. In this article, we propose the first flow-control layer co-synthesis flow called MiniControl, for continuous-flow microfluidic biochips under strict constraints for control ports, incorporating high-level synthesis, physical design, and control system design simultaneously, which has never been considered in previous work. With the maximum number of allowed control ports specified in advance, this synthesis flow aims to generate biochip architectures with high execution efficiency and the corresponding control systems with optimized timing performance. Besides, the overall cost of a biochip can be reduced and the tradeoff between a control system and execution efficiency of biochemical applications can be evaluated for the first time. The experimental results demonstrate that MiniControl leads to high execution efficiency, low platform cost, as well as excellent timing performance, while strictly satisfying the given control-port constraints.

Logs that record system abnormal states (anomaly logs) can be regarded as outliers, and the k-Nearest Neighbor (kNN) algorithm has relatively high accuracy in outlier detection methods. Therefore, we use the kNN algorithm to detect anomalies in the log data. However, there are some problems when using the kNN algorithm to detect anomalies, three of which are: excessive vector dimension leads to inefficient kNN algorithm, unlabeled log data cannot support the kNN algorithm, and the imbalance of the number of log data distorts the classification decision of kNN algorithm. In order to solve these three problems, we propose an efficient log anomaly detection method based on an improved kNN algorithm with an automatically labeled sample set. This method first proposes a log parsing method based on N-gram and frequent pattern mining (FPM) method, which reduces the dimension of the log vector converted with Term frequency.Inverse Document Frequency (TF-IDF) technology. Then we use clustering and self-training method to get labeled log data sample set from historical logs automatically. Finally, we improve the kNN algorithm using average weighting technology, which improves the accuracy of the kNN algorithm on unbalanced samples. The method in this article is validated on six log datasets with different types.

In recent years, many decentralized applications based on blockchain (DApp) have been developed. Some development tools provide testing functions, but only for developers to write unit tests for smart contracts rather than test DApp as a whole. Moreover, due to the difficulty for testers to understand the implementation details of smart contracts, insufficient functional testing causes some DApps not to meet functional design expectations. The inherent complexity of DApp, inconvenient pre-state setting, and not-so-readable logs make DApp testing challenging. In this paper, we propose Kaya, a testing framework for DApps to bridge these gaps. Firstly, Kaya formulate automatically executed test cases that cover both front-end behaviors and back-end logics with simple setting. Secondly, Kaya provides a flexible and convenient way for test engineers to set the blockchain pre-states. Thirdly, Kaya transforms incomprehensible addresses into readable variables for easier comprehension. Besides, to fit the various application environments, we provide both GUI and CLI for test engineers to use Kaya. Our case study and preliminary human study demonstrates the potential of Kaya in helping test engineers to test DApps more easily. A demo video is at https://youtu.be/7DyI_EpVZFw.

Smart contracts are the back-end programs of blockchain-based applications and the execution results are deterministic and publicly visible. Developers are unwilling to release source code of some smart contracts to generate randomness or for security reasons, however, attackers still can use reverse engineering tools to decompile and analyze the code. In this paper, we propose EShield, an automated security enhancement tool for protecting smart contracts against reverse engineering. EShield replaces original instructions of operating jump addresses with anti-patterns to interfere with control flow recovery from bytecode. We have implemented four methods in EShield and conducted an experiment on over 20k smart contracts. The evaluation results show that all the protected smart contracts are resistant to three different reverse engineering tools with little extra gas cost.

Self-Adaptive Systems (SASs) can dynamically adjust themselves to adapt to the changes, by planning some strategies to guide the adjustments. However, many uncertainties in runtime affect the efficiency of strategies and the ability to attain goals for SASs. So, it is necessary to validate strategies before they are executed. At present, most of the validation methods ignore the fact that less data can be observed at runtime, so it is difficult to describe the state accurately with these data. Most methods do not support uncertain state reasoning, but state transition is uncertain. This will result in the verification method not dealing well with the uncertainties. This paper proposes a strategy validation method based on Partially Observable Markov Decision Process, and makes several key contributions: (1) a uncertain state model that not only support description of states through partial information, but also describe the uncertain transitions of states, (2) a validation method to validate whether the strategies meets the requirements at runtime, (3) a strategy correction method to get effective strategies as quickly as possible.

Blockchain-based decentralized applications (DApp) have been widely adopted in different areas and trusted by more and more users due to the fact that the back end code of a DApp is publicly run on the blockchain and cannot be modified implicitly. However, there are few effective methods and tools for testing DApps and bugs can be easily introduced by inexperienced developers. The existing testing techniques either focus on testing front-end programs or back-end code but ignore the interaction between them, which makes it difficult to apply the techniques directly on DApp. In this paper, we present an automated testing technique for DApps which works in a two-phase manner. First, we employ random events to infer an abstract relation between browser-side events and blockchain-side contracts. Second, our technique generates a set of test cases under the guidance of inferred relations and orders the test cases based on a read-write graph. We also use taint analysis to track data flow of the smart contract and feed it to the generation procedure for following test cases. We have developed a tool called Sungari to implement our approach, and evaluated it on representative real-world DApps. The preliminary evaluation results demonstrated the potential of Sungari in achieving a significant optimization compared to random testing approaches.

Self-Adaptive Systems (SASs) need to analyze software changes accurately and continuously, that is, recognize events caused by changes, and adjust structure or behavior.However, present event recognition methods frequently monitor events, resulting in waste of system resources.And most of them ignore the impact of operating environment uncertainty, causing errors in recognizing the event and directly affecting the reliability of SASs.Addressing the above problems, this paper proposes an event recognition method based on "detection-recognition" mechanism.Firstly, the Naive Bayesian Classification algorithm is used to detect the state of the system.If the system is judged to be abnormal, we will combine with rule reasoning and fuzzy reasoning to recognize events.The system does not have to monitor the occurrence of events from time to time, avoiding the waste of system resources.Moreover, the probabilistic reasoning method of Bayesian Classification and the introduction of fuzzy reasoning can cope with environmental uncertainty and improve the accuracy of event recognition.Finally, we exemplify this mechanism with the Web system, which proves the effectiveness of the methods.

Metasearch engine integrates search results from multiple underlying search engines, improving recall ratio in the big data environment. Multi-agent system is an important way to implement metasearch engine. Great progress has been made in this area, however the previous studies are still short of personalization level. To improve the precision ratio, this paper proposes a personalized metasearch engine which of Agent-based architecture. According to click-through data, the metasearch engine has the ability to schedule the appropriate search engines based on the expertness model, merge all of results into a single list by taking user interest into account, and provide personalized recommendation. Experimental results show that the proposed personalized metasearch engine performs better on precision. It is feasible to provide the required search results more effectively.

Related query suggestion is very important for search engines. Users could find required information more quickly and accurately with the help of query suggestions, which could greatly improve users' search experience. Thus, query suggestion technology has become a research hotspot in the field of the search engine. Most of existing methods focused on the query log data to mine related queries. However, some of the query log data exist relatively sparse characteristics and have some interferential noise data. Besides, the method that only focus on query log trend to fail to consider the user's initial query intention. These shortages would reduce the accuracy of the recommendation. Thus, this paper proposes a query suggestion method based on random walk and topic concepts (QuS-RWTC). The method is based on the query log data and suggestions from other mature search engines, which could make the suggestions more comprehensive and obtain a higher coverage. In addition, the paper further executes procedures of topic concepts to re-order the candidate queries, which make the suggestions more accurate, since they are more satisfied to the user's initial intention. The results prove the excellent performance of QuS-RWTC method compared with traditional methods and validate the importance of topic concepts.

Meta-search engine is a comprehensive search tool, which is build base on those member search engines. All result entities reported by member search engines are merged into one ranked list according to their quality. It is well knowing that the core problem in meta-search engine is how to merge the results and provide user with a more effective rank list. This paper dealt with a synthesized merging algorithm by utilizing five features to estimate the quality of each result entity. For a returned result entity, we first record its' position of the original result list. Secondly, count the number of duplications. Thirdly, calculate the similarity between query terms and result content. Fourthly, get the capacity of the search members which will be called later. Fifth, analyze whether the current entity is in line with user's interests. Wherein users' interests are obtained both according to users' browsing history and feedback. Finally, we use the linear fusion model to merge the results set and re-ranking the results list. Experimental results shown that the merging method we proposed in this paper improved the accuracy and satisfaction degree in some cases compared with member search engines and several current meta-search engines.

Future-generation SASs need to have the adaptive abilities to efficiently handle changes from different sources and to mitigate conflicts caused by multiple simultaneous changes. However, existing methods cannot simultaneously make Future-generation SASs have the above abilities. This paper proposes an adaptive system framework based on agent technology and search-based software engineering technology (SBSE) for developing future-generation SASs with above-mentioned abilities. The framework integrates a hybrid adaptation logic based on agents to deal with various software changes from different layers, and an adaptation planning method with search-based optimization mechanism to mitigate conflicts caused by multiple simultaneous changes.

Planning a suitable solution to adapt to software changes is the most important and fundamental ability of self-adaptive software (SAS). However, with the increasing complexities of managed resources, context and user preferences, existing self-adaptive planning approaches need to be improved to deal with the complex changes which are multiple, interrelated and evolving. Search-based optimization (SBO) is well-suited to deal with multiple and complex problems. Hence, using SBO as a new self-adaptive planning approach may be a particularly promising research trajectory. This paper proposes a multi-agent framework for SAS with SBO to deal with complex changes, reduce maintenance time and cost, and enhance software quality. This framework defines a special software architecture of SAS to choose different planning approaches, uses the SBO to plan solutions for complex changes, and supports the online planning by multi agents. In addition, a corresponding workbench is being established to develop SAS according to this framework.

Some methods and auxiliary platforms/tools have been proposed to design and develop Self-Adaptive Systems (SASs).However, these methods and tools ignore how to turn existing systems to SASs.Therefore, the paper proposes a multiagentbased toolkit for adding self-adaptive abilities to the existing systems.With the Agent Packager and Rule Designer in it, existing systems are transformed into Multi-Agent Systems (MAS), without modifying codes.Based on the Supporting Platform in MABT, MAS run normally and achieves self-adaption with both Global and Local approach.

In the supply chain, RFID tags are deployed more widely. In the life of the supply chain, the owner of the tag will change frequently. Ownership transfer protocol can achieve the purpose that the access rights of the tag are transferred from the original owner to the new owner, and protect the privacy of the original owner and the new owner. To resist cloning attack and side channel analysis attack, physical unclonable function (PUF) has been proposed to enhance the security of the tags. Since the PUF of each tag is unique and different, it is difficult to be forged. However, most of PUF-based authentication protocols need the response value previously stored in the readers. On the other hand, most of the ownership transfer protocols assume the original owner and the new owner has a secure channel. However, in an open environment, due to time and space constraints, such a channel is often unable to quickly established. In this paper, we studied the ownership transfer protocols in an open environment and proposed a PUF-based RFID ownership transfer protocols, PROTP. The new protocol is the first ownership transfer protocol based on the PUF in an open environment. The new protocol does not need to store the respond values of the PUF. To utilize the randomness of the PUF, it replaces the pseudo-random generator. Meanwhile, PROTP can protect the privacy of the original owner and the new owner. In terms of efficiency, since the protocol is designed to satisfy the requirement in an open environment, the total cost of the computation is more than others protocols. However, due to the new protocol utilizes the PUF to replace the pseudo-random generator, the each step of the authentication messages achieves a better optimization in computational cost.

This paper proposes a new monitoring path design paradigm for independent dual link failures. Specifically, the new approach exploits the sequential arrival and departure property of independent failure events to uniquely localize failed links. Such property, however, cannot be captured by the existing approach, which is built upon the notion of shared risk link groups. Consequently, we show via solution space comparison that the existing approach can result in overdesign in terms of monitoring resources required. Numerical results further indicate that the new approach outperforms the existing one in terms of monitoring cost and computational efficiency.

Introducing agent techniques into the process of system integration will solve the problem about flexible dynamic system integration. The contract net protocol (CNP) has been applied to the negotiation in multi-agent system. On the basis of the research on traditional CNP, by improving its ability to deal with dynamic environment, this paper presents a contract net model based on agent active perception (CNMAAP) to improve the efficiency of negotiation. With the introduction of a perception coefficient and a parameter for the degree of credibility to CNMAAP, the paper puts forward the tendering strategy for contract manager and the bidding strategy for contract tender, after which an analysis of the efficiency of CNMAAP is made. Finally, CNMAAP is applied to the agent-based system integration tool, and has been tested with the border and coast defense simulation system. The result demonstrates that, compared with traditional CNP, CNMAAP can effectively reduce the negotiation cost with assured quality.