Accelerating neural architecture search for LSTM with a training-free method based on swarm intelligence and Neural Tangent Kernel

Real-time route guidance during disruptions in urban rail transit systems requires rapidly providing effective strategies that simultaneously alleviate congestion and account for passengers' travel time. This study proposes an optimization framework that considers travel time, congestion perception time, and information costs, incorporating a Logit choice model with information bias to reflect passengers' behavioral responses under disruptions. A bi-level simulation evaluation mechanism is employed to rapidly evaluate the objective functions under different guidance strategies, where a Physically Consistent Incremental Simulator, based on differential computation, achieves a 599-fold speedup while maintaining high fidelity with full-scale simulations (Pearson correlation > 0.96). A hybrid algorithm combining the Gray Wolf Optimizer and Adaptive Large Neighborhood Search is developed to solve the origin-destination level route guidance optimization problem. The algorithm embeds domain knowledge-based "destroy and repair" operators with a sequential repair mechanism to enable fast global search and precise local refinement. Case study results demonstrate that the framework reduces severely congested sections by 36%, shortens average travel time by 7.16 min, and improves solution quality by 12-30% over baseline algorithms. These findings confirm the practical applicability of integrating intelligent optimization with high-efficiency simulation for emergency route guidance in large-scale metro networks.

The periodic orbits are central to the transition state theory and the calculation of chemical reaction rates. Since this computation is often highly challenging, we introduce two methods based on Particle Swarm Optimization (PSO), an algorithm that belongs to the field of swarm intelligence (a subfield of artificial intelligence), to efficiently locate these objects and overcome the associated difficulties.

Swarm intelligence results from a collective behaviour of swarm network, which harnesses distributed and simple rules of swarm systems to address complex problems without a central controller. One potential approach to transform such swarm networks into valuable and practical computational resources is by applying the reservoir computing framework. However, technical challenges, such as permutation symmetry and instability, could emerge in these networks during the process, which significantly hinder the computational performance. In this paper, we explore the potential of mobile swarm networks in a reservoir computing framework to perform machine learning tasks. We propose an observation layer using Gaussian kernel density estimation to be inserted into the reservoir computing framework. Our approach not only addresses permutation symmetry but also stabilises swarm behaviours, resulting in a scalable swarm network. We explore variations in computational capacity across different swarm sizes and combinations with four benchmark computations. We prove the effectiveness of our observation layer in addressing permutation symmetry and discovered the improvement in performance in combining different swarm networks in parallel. We found that the best ratio between ants and birds reservoir is 8:2. The performance achieves a covariance of approximately 0.20 with a swarm size of 20, comparable to that of echo-state-network (ESN) with 16 nodes. As the swarm size increases to 60, the covariance value reaches around 0.21, matching the performance of ESN with 18 nodes. This indicates that our swarm network has a reasonable amount of memory and nonlinearly capacity in performing computation tasks. We also validate our method's effectiveness on a handwriting classification task, further highlighting its practical applicability. Our findings delve into the impacts of the swarm networks' computational abilities, offering insights into mechanisms in this alternative means of swarm intelligence and application to AI.

Optimizing ethylene combustion mechanisms in scramjet: A machine learning approach combining swarm intelligence and deep learning

Optimizing diverse team formation with swarm intelligence algorithms for enhancing organizational performance

In today's fast-changing digital environment, cyber-physical systems face escalating security challenges due to increasingly sophisticated cyberattacks. Artificial Intelligence (AI) has emerged as a powerful enabler of modern cyberattack detection, offering scalable, accurate, and adaptive solutions to counter dynamic threats. This paper provides a comprehensive review of recent advancements in AI-based cyberattack detection, focusing on Machine Learning (ML), Deep Learning (DL), Reinforcement Learning (RL), Federated Learning (FL), and emerging techniques such as generative AI, neuro-symbolic AI, swarm intelligence, lightweight AI, and quantum Computing. We evaluate the strengths and limitations of these approaches, highlighting their performance on benchmark datasets. The review discusses traditional signature-based Intrusion Detection Systems (IDS) and their limitations against novel attack patterns, contrasted with AI-driven anomaly-based and hybrid detection methods that improve detection rates for unknown and zero-day attacks. Key challenges, including computational costs, data quality, privacy concerns, and model interpretability, are analysed alongside the role of Explainable AI (XAI) in enhancing trust and transparency. The impact of computational resources, dataset representativeness, and evaluation metrics on AI model performance is also explored. Furthermore, we investigate the potential of lightweight AI for resource-constrained environments like IoT and edge devices, and quantum computing's role in advancing detection efficiency and cryptographic security. The paper also draws attention to future research directions, particularly the development of up-to-date datasets, integration of hybrid quantum-classical models, and optimisation of asynchronous FL protocols to address evolving cybersecurity challenges. This study aims to inspire innovation in AI-driven cyberattack detection, fostering robust, interpretable, and efficient solutions for securing complex digital environments.

Cable manufacturing is a typical hybrid production system characterized by the deep coupling of continuous processes and discrete logic. However, the unique “Reel-Splitting Constraint”—where continuous cables must be segmented into strictly sequenced sub-reels—along with high energy consumption and frequent dynamic disturbances render traditional Hybrid Flexible Flow Shop scheduling models ineffective in this context. To address these challenges, this paper proposes a novel Multi-Objective Dynamic Scheduling Framework tailored for the cable industry. First, a mathematical model is constructed that explicitly formalizes the rigid logic of sub-reel sequencing and continuous material flow, aiming to simultaneously minimize total energy consumption, makespan, and changeover times. Unlike generic models, this formulation introduces a constraint-handling mechanism to ensure the physical continuity of sub-reels during optimization. Second, a two-stage hybrid swarm intelligence algorithm is developed to solve this NP-hard problem. An improved Ant Colony Optimization (ACO) algorithm is employed for “population seeding” to generate feasible initial schedules and avoid deadlocks, while a Variable Neighborhood Search (VNS) executes deep evolutionary operations—such as setup reduction and critical operation insertion—to escape local optima. Case studies based on real-world industrial data demonstrate the superior performance of the proposed method. The hybrid strategy reduces the makespan by approximately 9.8% compared to traditional approaches and effectively mitigates energy waste in bottleneck processes. Furthermore, the proposed event-driven dynamic rescheduling mechanism exhibits exceptional responsiveness, reducing rescheduling time for unexpected equipment breakdowns from 18 h to 0.83 h, thereby enabling within-shift decision-making and robust operation in volatile manufacturing environments.

Wireless Sensor Networks (WSNs) are characterized by limited energy, and energy efficiency is one of thekey design issues for routing protocols. This research aims to enhance the routing efficiency of dragonflyswarm routing by optimizing route and cluster head selection through the integration of the latest SwarmIntelligence (SI) algorithm, specifically the Dragonfly Algorithm (DA). The proposed method wassystematically compared with the traditional Particle Swarm Optimization (PSO) by measuring energyefficiency, execution time, and packet delivery ratio. Simulation results showed that the DragonflyAlgorithm reduces energy consumption and prolongs network lifetime for classical methods. It exhibitsstrong adaptability to time-varying network topologies and is less likely to be trapped in a local optimum.These results illustrate that SI is a promising technique to help improve the quality of routing protocolsin WSN applied to critical scenarios and offer possibilities for future integration with, for example,machine learning techniques for achieving higher performance.

The field of artificial intelligence (AI), machine learning and intelligent automation has become pervasive in our modern digital world. It extends from business and public services to environmental management and even into people's daily experiences with technology. In this special issue "Intelligent Systems and the Next Wave of Digital Innovation," published in the Journal of Informatics and Web Engineering, reviews several studies to explore the increasing role of intelligent systems in current society and their importance. Some of the more significant areas of discussion are how to formalize the expectations for explainable AI, evaluating face recognition models in the real world, how trust and transparency of AI models are evaluated and more. It also highlights a promising and emerging frontier of intelligent automation — from swarm intelligence and optimization within manufacturing to the ubiquity of multimodal interfaces, such as sign language chatbots. Furthermore, smart environmental analytics techniques such as neuro-intelligent techniques for drought prediction and IoT-generated flood intelligence systems help communities to plan for disaster events are also being studied. All of these contributions in turn reinforce the notion that intelligent systems can be developed more responsively and contextually through data-driven architectures. These also reflect a wider digital innovation trend: an era when decision-support tools and algorithmic intelligence and real-time data and other technologies converge toward reliability, efficiency, inclusivity, and resilience in increasingly complex social and technical ecosystems.

Effective and accurate management of dynamic workloads in wireless sensor networks (WSNs) is critical to ensure energy efficiency, network longevity, and reliable information delivery in heterogeneous environments. Traditional heterogeneous WSN solutions often suffer from unbalanced energy usage, poor node placement and load distribution optimization, leading to poor performance and high operating costs. This proposes a new energy-aware load balancing architecture for heterogeneous WSNs that incorporates a multi-level optimization approach. First, Energy Weighted Probabilistic Deployment (EWPD) takes energy efficiency into account, guaranteeing optimal node placement to maximize network coverage and extend lifetime. Second, network initialization and node classification based on Sliding Time Windows (STW) predict task execution patterns and allow adaptive allocation of nodes according to task performance. Third, Expected Time to Task Completion (ETTC) scheduling with cluster formation and path discovery can predict workload times to maximize routing selection. Fourth, workload analysis is a tool based on Spider Swarm Intelligence (SSI) that measures the computational needs of nodes and finds the most efficient allocation of workload resources. Finally, the Minimum-Maximum Priority Energy-Aware Self-Scheduling Mechanism (DMMPS-EASSM) is a dynamic operation and migration mechanism designed to balance energy consumption and provide efficient data transmission. Experimental results demonstrate that the proposed framework can significantly improve network performance, reduce energy wastage, and improve resource utilization compared with traditional approaches based on heterogeneous WSNs.

Artificial bee colony (ABC) algorithm is one representative of many wellknown swarm intelligence methods for continuous optimization problems. However, it cannot directly solve discrete optimization problems without using complex transfer functions. Furthermore, the solutions quality and deviations obtained by many famous intelligent algorithms are still to be enhanced for solving uncapacitated facility location problems (UFLP). To this end, a continuous ABC called cABC is proposed for UFLP. In cABC, a chaotic initialization technique is employed to produce a good initial population in the range of [0,1), which enables cABC to evolve in continuous space. Then, a common probability discretizing mechanism is used to convert a continuous individual to a 0-1 vector, which enables cABC to solve UFLP. In addition, for infeasible solutions, a dynamic repair strategy is presented. Next, to enhance search performance of ABC, a random guiding mechanism is proposed. Subsequently, a time varying perturbation scheme is presented to share much more information between current individual and guiding individual. Next, a modified probability choice mechanism with random character is employed before entering onlooker bees phase. Last, an opposition based learning technique is employed to improve continuous nonupdating individual at the scout bees phase. To test effectiveness of cABC, it is first compared with traditional ABC on famous CAP dataset consisting of fifteen instances. To further validate superiority of cABC, it is compared with other eleven famous approaches on CAP dataset and M* dataset with twenty instances. Experimental results show that cABC surpasses other state-of-the-art methods in terms of solution accuracy and robustness.

The efficacy of swarm intelligence algorithms in navigating high-dimensional, non-convex landscapes depends on the dynamic balance between global exploration and local exploitation. Drawing inspiration from the intricate social dynamics of Pyrrhura molinae, this study proposes a novel bio-inspired metaheuristic, the Comprehensive Learning Parrot Optimizer (CL-PO). While the original Parrot Optimizer (PO) simulates collective foraging and communication, it often suffers from population homogenization and entrapment in local optima due to its reliance on single-source social learning. To address these limitations, CL-PO incorporates a dimension-wise multi-exemplar social learning mechanism analogous to the cross-individual knowledge transfer observed in avian colonies. This strategy enables stagnant individuals to reconstruct their search trajectories by learning from multiple superior peers, thereby sustaining population diversity and facilitating adaptive exploration. Rigorous benchmarking on 29 test functions from the CEC 2017 suite reveals that CL-PO achieves statistically superior performance compared to nine state-of-the-art algorithms, securing a top-tier average Friedman rank of 1.28. Furthermore, the practical utility of CL-PO is substantiated through a complex reservoir production optimization task using the Egg benchmark model, where it consistently maximizes the net present value (NPV), reaching 9.625×108 USD. These findings demonstrate that CL-PO is a powerful and reliable solver for addressing large-scale engineering optimization problems under complex constraints.

Aiming at the drawbacks of slow convergence, proneness to local optima, and insufficient segmentation accuracy of traditional multi-threshold image segmentation algorithms in complex scenarios, an Improved Grey Wolf Optimizer (IGWO) integrated with opposition-based learning strategy and nonlinear dynamic convergence factor is proposed. Specifically, we use improved opposition-based learning for population initialization and iteration optimization. This step helps increase population diversity and speed up convergence. Instead of the traditional linear decreasing convergence factor, we adopt a nonlinear dynamic one. This change achieves an adaptive balance between global exploration and local exploitation of the algorithm. We take the maximum between-class variance (OTSU) as the fitness function and build a multi-threshold segmentation optimization model. We validate IGWO through 6 benchmark test functions and compare it with four advanced swarm intelligence algorithms. The results show that IGWO has obvious advantages in convergence speed, solution accuracy and stability. It also has a strong ability to avoid local optimal solutions. When applied to multi-threshold image segmentation, IGWO produces segmentation regions with clear boundaries and well-preserved details. This algorithm provides a new technical method for efficient and accurate segmentation of complex images. It can be used in fields such as computer vision and communication equipment fault detection.

This paper proposes an Enhanced Projection-Iterative-Methods-based Optimizer (EPIMO) to overcome the limitations of its predecessor, the Projection-Iterative-Methods-based Optimizer (PIMO), including deterministic parameter decay, insufficient diversity maintenance, and static exploration–exploitation balance. The enhancements incorporate three core strategies: (1) an adaptive decay strategy that introduces stochastic perturbations into the step-size evolution; (2) a mirror opposition-based learning strategy to actively inject structured population diversity; and (3) an adaptive adjustment mechanism for the Lévy flight parameter β to enable phase-sensitive optimization behavior. The effectiveness of EPIMO is validated through a multi-stage experimental framework. Systematic evaluations on the CEC 2017 and CEC 2022 benchmark suites, alongside four classical engineering optimization problems (Himmelblau function, step-cone pulley design, hydrostatic thrust bearing design, and three-bar truss design), demonstrate its comprehensive superiority. The Wilcoxon rank-sum test confirms statistically significant performance improvements over its predecessor (PIMO) and a range of state-of-the-art and classical algorithms. EPIMO exhibits exceptional performance in convergence accuracy, stability, robustness, and constraint-handling capability, establishing it as a highly reliable and efficient metaheuristic optimizer. This research contributes a systematic, adaptive enhancement framework for projection-based metaheuristics, which can be generalized to improve other swarm intelligence systems when facing complex, constrained, and high-dimensional engineering optimization tasks.

Advancements in Industry 4.0 technologies, which combine big data analytics, robotics, and intelligent decision systems to enable new ways to increase automation in the industrial sector, have undergone significant transformations. In this research, a Hybrid Attention-Gated Recurrent Unit (At-GRU) model, combined with Sand Cat Optimization (SCO), is proposed to enhance fault identification and predictive maintenance capabilities. The model utilized multivariate sensor data from cyber-physical and IoT-enabled robotic platforms to learn operational patterns and predict failures with enhanced reliability. The At-GRU provides deeper temporal feature extraction, thereby improving classification performance. The robustness of the proposed model is validated through analysis of a benchmark dataset for industrial robots, and the results demonstrate that the proposed model exhibits impressive predictive capacity, surpassing other prediction methods and predictive maintenance approaches. Additionally, the performance evaluation indicates a lower computational cost due to the lightweight gating architecture of GRU, combined with attention. The robotic motion is further optimized by the SCO algorithm, which reduces energy usage, execution delay, and trajectory deviations while ensuring smooth operation. Overall, the proposed work offers an intelligent and scalable solution for next-generation industrial automation systems. Furthermore, the proposed model demonstrates the real-world applicability and significant benefits of incorporating hybrid artificial intelligence models into real-time robot control applications for smart manufacturing environments.

A critical challenge in swarm intelligence is the effective utilization of knowledge gained during the search, a process often confounded by the risk of negative knowledge transfer. To address this, we introduce the Learning-Aided Artificial Bee Colony (LA-ABC), a novel framework guided by a Neural Knowledge Transfer mechanism for global optimization. Our framework establishes a co-evolutionary mechanism between the search process of the ABC algorithm and an online neural knowledge learning engine. LA-ABC operates on a dual-pathway architecture, probabilistically arbitrating between foundational swarm exploration and a knowledge-transfer pathway. In this second pathway, an Artificial Neural Network (ANN) learns a predictive, non-linear model from a dynamic archive of historically successful solutions. This approach enables the model to interpret the complex context of successful moves, thereby preventing the negative knowledge transfer where a beneficial pattern in one region of the search space could be detrimental in another. This learned intelligence is then operationalized through a generative operator that transfers validated positive knowledge to create high-quality candidate solutions. The process transforms the ABC from a memoryless explorer into an intelligent agent that learns to navigate the fitness landscape with high efficacy. The superiority of the LA-ABC framework is demonstrated through comprehensive benchmarking on 23 standard test functions, the competitive IEEE CEC 2019 suite, and a real-world photovoltaic parameter extraction problem. Our proposed neural knowledge transfer approach significantly outperforms 12 state-of-the-art algorithms, including ABC, L-SHADE, JSO, L-DE, L-PSO, KL-variants, and RL variants with the significance of these improvements validated by rigorous statistical tests (Wilcoxon, Bonferroni-Dunn, Friedman, and ANOVA). Ultimately, LA-ABC provides a robust new paradigm for integrating reinforcement learning and knowledge transfer within evolutionary computation.

The performance of metaheuristic algorithms in solving high-dimensional, non-convex optimization problems is intricately linked to the balance between global exploration and local exploitation. Inspired by biomimetic principles of swarm intelligence, this study evaluates the Educational Competition Optimizer (ECO), a human learning-inspired metaheuristic, and addresses its vulnerability to rapid population homogenization and premature convergence in complex landscapes. To bridge the gap between rigid hierarchical competition and flexible biological cooperation, we propose the Comprehensive Learning-Enhanced Educational Competition Optimizer (CL-ECO), which introduces a dimension-wise multi-exemplar social learning mechanism to the ECO framework. Analogous to cooperative information sharing in animal swarms, CL-ECO reconstructs search trajectories by learning from different peers across decision variables, thereby promoting population diversity and adaptive exploration. Rigorous validation on the CEC 2017 benchmark suite demonstrates that CL-ECO achieves statistically superior convergence accuracy and robustness compared to seven state-of-the-art algorithms, securing the top Friedman rank (1.5862). Furthermore, the practical utility of CL-ECO is substantiated through a complex reservoir production optimization case study, where it outperforms the baseline algorithm in NPV maximization, proving its capability in managing complex, real-world engineering constraints.

Brain tumors often require treatment and multiple biopsies. They are the third most common cancer among young adults in both incidence and mortality. The expression of the O6-methylguanine-DNA methyltransferase (MGMT) gene plays an important role in predicting tumor behavior. It affects how patients respond to chemotherapy and may reduce the need for invasive procedures. Machine learning can help make accurate medical predictions, but it requires large and diverse patient datasets. These datasets are difficult to access due to privacy and legal restrictions. This article proposes a Federated Learning (FL) framework to address these challenges. FL allows different institutions to train a shared model without exchanging raw data. A hybrid deep learning model combining recurrent neural networks (RNNs) and convolutional neural networks (CNNs) is developed to analyze magnetic resonance imaging (MRI) scans from the BraTS 2021 dataset. The model aims to detect glioblastoma and predict MGMT gene expression. Two swarm intelligence algorithms, the Bayesian Search Optimization Algorithm and the Sparrow Search Optimization Algorithm, are used to optimize the model’s hyperparameters. The FL system was tested across ten universities. It performed similarly to models trained on centralized data. The proposed model, BrainGeneDeepNet, achieved high performance: 0.9758 accuracy, 0.0769 loss, 0.9980 AUC, 0.9770 recall, and 0.9782 precision. These results show that federated learning is a secure and effective approach for medical imaging and biomarker prediction.

Adaptive small-family population-guided swarm intelligence optimization algorithm