Enhancing software reliability mitigates failures and reduces maintenance costs. In the LLVM compiler, different option sequences applied to the intermediate representation (IR) produce binaries with varying reliability levels, making the search for an optimal sequence in reliability-oriented compilation a key challenge. Although reinforcement learning (RL) has been employed to automate this process, existing methods suffer from two major limitations: they rely on structure-based static IR embeddings that overlook the dynamic behaviors induced by compilation transformations, leading to low-fidelity state representations, and construct action spaces that fail to preserve critical inter-option dependencies while maintaining spatial compactness, thereby impairing training efficiency and constraining optimization gains. This article presents Optimizing Software Reliability at Compile-time using Reinforcement Learning (OSRC-RL), which integrates two novel components: a Compilation Behavior-Aware State Representation (CBA-SR) and an Action Space Construction via Option Dependency Awareness and space minimization (ASC-ODA). CBA-SR jointly encodes program semantics and compilation dynamics, using activated options as multi-label supervisory signals to guide an expressive Graph Isomorphism Network in learning IR subgraph patterns correlated with option activations, yielding high-fidelity states that enable more reliable policy learning and faster convergence. ASC-ODA reconciles dependency preservation with spatial compactness by constructing Option Dependency Graph, performing dependency-aware hierarchical clustering and recursive intra-cluster path analysis, and applying a preference-aware selection to generate a compact yet dependency-preserving action space that enhances sample efficiency and policy stability. Evaluations across 20 benchmarks show that OSRC-RL achieves the highest average reliability gain ( \(I_{rg}{=}0.6694\) ) and competitive convergence speed compared with five baselines. Ablations attribute these gains to its core components: CBA-SR (up to 11.34% improvement via IR-graph learning) and ASC-ODA (up to 15.22% improvement via dependency-faithful actions). On industrial instances, it attains a 0.1818 average gain, surpassing the next-best method by 33.77%. Bounded overheads are mitigated by Lightweight Post-Processing (LPP), ensuring practical feasibility.

With the rapid development of 5G communication technology, the data in the Internet of Medical Things (IoMT) application systems exhibits complex characteristics such as large volume, high dimensionality, nonlinearity, and diversity, which significantly affect the efficiency and detection performance of anomaly detection tasks. How to efficiently extract nonlinear features from high-dimensional data in the context of the IoMT while minimizing information distortion in data objects are challenging problems in recent academic research. A novel adaptive nonlinear feature extraction method via fruit fly olfactory neural network (Fly dimension expansion projection and remain main components by PCA, FDEPCA) is proposed, where 1) the data are mean-centered; 2) a binary sparse random projection matrix is used for dimension expansion projection; and 3) PCA is used to extract principal component information. The proposed method overcomes the problems of present nonlinear feature extraction in the face of high-dimensional outliers where the intrinsic geometric structure of the data is severely distorted and computationally expensive. The dataset after nonlinear feature extraction by the FDEPCA algorithm is applied to specific anomaly detection models, using ROC curves and AUC as evaluation metrics for classification performance. Extensive comparison experiments are conducted on eight publicly available datasets, and experimental results show that compared with the popular nonlinear feature extraction algorithms, the FDEPCA algorithm has better classification performance and projection time advantage. When applied to proximity-based, probability-based, and ensemble-based different anomaly detection models respectively, the FDEPCA algorithm exhibits strong applicability in different types of anomaly detection classifiers.

With recent advancements in Industrial Internet of Things (IIoT), the stream data generated in IIoT applications presents new characteristics: huge amount of data, ultrahigh computational complexity and large memory consumption, and the existence of concept drift leading to the ineffective distinction between real drift and anomalies. It is difficult for the current mainstream methods to cope with the above problems effectively. In this article, we propose an efficient Gaussian kernel microcluster real-time-clustering method for IIoT data streams (GKMC).The method uses a microcluster sketch structure instead of individual data sample points to participate in clustering directly, which solves the problem of not being able to store unlimited data in limited memory; it uses a Gaussian kernel function to calculate the local density of microclusters to enhance the detection of anomalies; in addition, using the gravity energy function recursively to update the microcluster online and using the relearning strategy to improve the detection ability of whether the outdated microcluster belongs to abnormal microcluster or has real concept drift, ensuring that the current microcluster is always the latest microcluster most closely related to the cluster. The theoretical analysis and sufficient comparison experiments on three data sets show that the proposed algorithm has a better clustering effect than the current mainstream stream clustering algorithms.

Abstract Locality‐sensitive hashing (LSH) is widely used in the context of nearest neighbor search of large‐scale high‐dimensions. However, there are serious imbalance problems between the efficiency of data index structure construction and the query accuracy of LSH methods. In this article, a novel higher‐entropy‐hyperplane clusters LSH (HEHC‐LSH) algorithm is proposed, which we improve vector quantization to preprocess the data and greatly shortens the preprocessing time; We innovatively integrate the maximum entropy principle into the distribution estimation algorithm to construct a novel hash function cluster method, also incorporate bootstrap aggregating of ensemble learning, and adopt the parallel index dictionary to improve the generalization performance of the index structure. And in the query stage, we realize the comprehensive filtering of index set using integrated learning idea, which not only avoids a lot of distance calculation, but also improves the quality of query results. We also analyze the rationality and effectiveness of the proposed method. Finally, extensive experiment results show that HEHC‐LSH can achieve more higher precision and efficiency simultaneously comparing to current methods, and reflect the strong robustness on different datasets.

With the advent of the 5G era, wireless multimedia sensor networks (WMSNs) will be more widely used in security monitoring, environmental monitoring, intelligent transportation, and other applications of the Internet of Things (IoT). Data collected by WMSNs, part of IoT systems, are high-dimensional, multilevel, unstructured, and large-scale complex data. Effective nearest neighbor searching faces the “dimension curse” problem. To this end, by incorporating the neighborhood rough set (NRS) approach, this article proposes a novel and effective locality-sensitive hashing (LSH) method for high-dimensional WMSN data based on the neighborhood (NLSH). This method innovatively combines the neighborhood knowledge representation method with the LSH mechanism and extends the indexing ability of LSH. Moreover, this method does not require any prior knowledge and has good universality. The indexing method of neighborhood bucket building can reduce the number of searches and improve the response speed of the WMSN system. Extensive experiments have been carried out on several real-world multimedia data sets. The results show that a large-scale high-dimensional NLSH search based on the NRS approach can efficiently query the target point and produce very accurate results. The NLSH algorithm based on the NRS approach outperforms other advanced mainstream LSH algorithms.

Wireless multimedia sensor networks (WMSNs) have been widely used in environmental monitoring, intelligent transportation, and other scenarios; however, their data have high-dimensional, large-scale, and multitype properties, as well as other characteristics. It is technically difficult to construct an appropriate index structure and query strategy while we perform a highly accurate search. This paper proposes a novel locality-sensitive hashing (LSH) fast Johnson-Lindenstrauss transform (FJLT)-fly locality-sensitive hashing (FLSH) algorithm for WMSN Internet of Things search. In this method, the projection method of FJLT and the winner-takes-all feature selection strategy in the fruity FLSH are considered. The method provides a new solution for the nearest neighbor search of high-dimensional data. We also discuss the distance-keeping property of our algorithm, and prove theoretically that the method proposed in this paper has better distance keeping performance than the traditional dimensionality reduction method. The experimental results show that the proposed algorithm has better generalization, accuracy of the search results, and time efficiency when using the Drosophila olfactory nerve to simulate the LSH process. This method effectively solves the problem of the approximate neighbor query of high-dimensional big data and can be effectively applied to search application on WMSN system.

Architecture-based software performance optimization can not only significantly save time but also reduce cost. A few rule-based performance optimization approaches at software architecture (SA) level have been proposed in recent years. However, in these approaches, the number of rules being used and the order of application of each rule are uncertain in the optimization process and these uncertainties have not been fully considered so far. As a result, the search space for performance improvement is limited, possibly excluding optimal solutions. Aiming to solve this problem, we propose an evolutionary algorithm for rule-based performance optimization at SA level named EA4PO. First, the rule-based software performance optimization at SA level is abstracted into a mathematical model called RPOM. RPOM can precisely characterize the mathematical relation between the usage of rules and the optimal solution in the performance improvement space. Then, a framework named RSEF is designed to support the execution of rule sequences. Based on RPOM and RSEF, EA4PO is proposed to find the optimal performance improvement solution. In EA4PO, an adaptive mutation operator is designed to guide the search direction by fully considering heuristic information of rule usage during the evolution. Finally, the effectiveness of EA4PO is validated by comparing EA4PO with a typical rule-based approach. The results show that EA4PO can explore a relatively larger space and get better solutions.