Enhancing phishing detection with dynamic optimization and character-level deep learning in cloud environments.

The classification of clinical notes into specific diagnostic categories is critical in healthcare, especially for mental health conditions like anxiety and adjustment disorder. In this study, we compare the performance of various artificial intelligence models, including both traditional machine learning approaches (random forest, support vector machine, K-nearest neighbors, decision tree, and eXtreme Gradient Boost) and deep learning models (DistilBERT and SciBERT), to classify clinical notes into these two diagnoses. Additionally, we implemented three oversampling strategies: No Oversampling, Random Oversampling, and Synthetic Minority Over-sampling Technique (SMOTE), to assess their impact on model performance. Hyperparameter tuning was also applied to optimize model accuracy. Our results indicate that oversampling techniques had minimal impact on model performance overall. The only exception was SMOTE, which showed a positive effect specifically with Bidirectional Encoder Representations from Transformers (BERT)-based models. However, hyperparameter optimization significantly improved accuracy across the models, enhancing their ability to generalize and perform on the dataset. The decision tree and eXtreme Gradient Boost models achieved the highest accuracy among machine learning approaches, both reaching 96%, while the DistilBERT and SciBERT models also attained 96% accuracy in the deep learning category. These findings underscore the importance of hyperparameter tuning in maximizing model performance. This study contributes to the ongoing research on AI-assisted diagnostic tools in mental health by providing insights into the efficacy of different model architectures and data balancing methods.

—Driven by deep learning advances, speech processing systems such as automatic speech recognition (ASR), source segregation, noise suppression have achieved significant performance improvements. However, traditional training strategies, particularly static mini-batch selection, often overlook the dynamic variations in data complexity and model convergence behavior, resulting in ineffective training efficiency and limited model accuracy. To tackle this limitation, we introduce a novel training paradigm called Dynamic Micro-block Optimization (DMBO). The method introduces a fine-grained sampling mechanism by partitioning the training set into smaller units called “micro-blocks,” which are dynamically updated during training based on real-time characteristics such as sample loss, gradient diversity, and utterance complexity. Four sampling strategies—loss-weighted, gradient-diversity, gender-based, and accent-based—are designed to self-adjust the composition of training data. The DMBO framework is implemented using Connectionist Temporal Classification (CTC) and Long Short-term Memory (LSTM) networks for end-to-end speech recognition. Experimental evaluations on the VCTK datasets demonstrate that the proposed method significantly accelerates convergence and improves model accuracy. Specifically, the gender-homogeneous strategy reduces the Label Error Rate (LER) by 9.0% compared to standard mini-batch training, while the accent-heterogeneous strategy achieves a 9.2% absolute LER reduction. These results confirm that dynamic optimization at the micro-block level enhances the efficacy of deep learning models in speech processing tasks, and the experimental outcomes are consistent with theoretical expectations, validating the effectiveness and correctness of the proposed approach.

This study uniquely integrates Long Short-Term Memory networks (LSTM) and Graph Convolutional Networks (GCN) with a multi-head attention mechanism to address dynamic carbon pricing optimization in green transportation supply chains, overcoming the limitations of traditional static models. As global climate change issues become increasingly severe, the design of carbon pricing mechanisms for green transportation supply chains has become a key factor in promoting sustainable development. We construct a hybrid deep learning model that simultaneously captures temporal dependencies in carbon emission data and spatial relationships in supply chain network structures. Traditional carbon pricing methods often rely on static models and simplified assumptions, making it difficult to adapt to complex and dynamic supply chain environments. Experimental results show that the proposed deep learning method improves carbon price prediction accuracy by 23.7% compared to traditional methods and achieves 18.5% improvement in supply chain cost optimization. Furthermore, the method achieved an average 21.6% carbon emission reduction and 15.5% cost reduction in three real green transportation supply chain cases, demonstrating its effectiveness in practical applications. The multi-objective optimization framework successfully balances the trade-off between economic and environmental benefits through organic integration of genetic algorithms and deep learning models. Ablation experiments validated the importance of each model component, and sensitivity analysis confirmed the rationality of parameter settings. This method provides strong technical support for formulating more precise and dynamic carbon pricing policies, offering significant theoretical value and practical significance for promoting sustainable development of green transportation supply chains.OPEN ACCESS Received: 17/09/2025 Accepted: 24/11/2025 Published: 20/03/2026

IgA nephropathy (IgAN) is a leading cause of renal failure, characterized by significant clinical and pathological heterogeneity. Accurate subtype classification remains challenging due to overlapping clinical manifestations and the multidimensional nature of data. Traditional methods often fail to fully capture IgAN's complexity, limiting their clinical applicability. This study introduces MAL-Net, a deep learning framework for multi-label classification of IgAN subtypes, leveraging multidimensional clinical data and incorporating sensor-based inputs such as laboratory indices and symptom tracking. MAL-Net integrates Long Short-Term Memory (LSTM) networks with Multi-Head Attention (MHA) mechanisms to effectively capture sequential and contextual dependencies in clinical data. A memory network module extracts features from clinical sensors and records, while the MHA module emphasizes critical features and mitigates class imbalance. The model was trained and validated on clinical data from 500 IgAN patients, incorporating demographic, laboratory, and symptomatic variables. Performance was evaluated against six baseline models, including traditional machine learning and deep learning approaches. MAL-Net outperformed all baseline models, achieving 91% accuracy and an AUC of 0.97. The integration of MHA significantly enhanced classification performance, particularly for underrepresented subtypes. The F1-score for the Ni-du subtype improved by 0.8, demonstrating the model's ability to address class imbalance and improve precision. MAL-Net provides a robust solution for multi-label IgAN subtype classification, tackling challenges such as data heterogeneity, class imbalance, and feature interdependencies. By integrating clinical sensor data, MAL-Net enhances IgAN subtype prediction, supporting early diagnosis, personalized treatment, and improved prognosis evaluation.

The rise in complexity of software systems and the rise in sophistication of cyber threats have led to cybersecurity becoming a key issue in software engineering. The conventional security systems are usually unable to capture and react to the current attacks in real time, especially in a large-scale and dynamic environment. The development of Artificial Intelligence (AI) has become a revolutionary solution due to its ability to detect threats intelligently, evaluate vulnerabilities automatically, perform predictive analytics, detect anomalies, and implement adaptive defense mechanisms. This paper will discuss AI-driven cybersecurity in the software engineering field, where machine learning, deep learning, natural language processing, and reinforcement learning can be applied throughout the software development lifecycle to provide increased security. The study assesses the efficacy of AI-powered technologies in secure coding, threat modeling, penetration testing, reviewing code, malware detection, and responding to an incident. It also examines issues like adversarial attacks, data privacy issues, bias in algorithms, explainability, and integration issues in current development pipelines. Based on a systematic review and comparative framework, the paper shows that AI-based cybersecurity can greatly enhance the ability to proactively defend against threats, decrease reaction time, and increase resilience in the face of new threats. The results highlight the importance of ethical governance, human control, and ongoing updates of models to maximize the advantages of AI in safe software engineering activities. The work can be regarded as a valuable roadmap to researchers and developers, as well as organizations aiming to deploy intelligent cybersecurity approaches to the contemporary software environment.

Dynamic multimodal optimization problems (DMMOPs) demand algorithms capable of swiftly locating and tracking multiple optimal solutions over time. The primary challenge lies in controlling the population diversity to facilitate effective exploration, all within the limitation of computational resources between consecutive environmental changes. In this paper, we study the utilization of density information derived from both current and historical populations to enhance exploration. First, for each active sub-population, we construct a density landscape based on the distribution of concurrently active sub-populations, and establish dominance relationships between candidate solutions in the sub-population based on density and fitness values, directing this sub-population towards exploring low-density promising areas. Then, for each converged sub-population, we construct a density landscape based on the distribution of sub-populations that have historically become extinct, guiding the restart of this sub-population in low-density unexploited areas. Finally, we develop a comprehensive framework of density-assisted evolutionary algorithm (DAEA), which encompasses density-assisted search and restart, also combined with initialization. Moreover, we employ prediction and memory strategies to enhance the performance of DAEA in dynamic environments. Notably, the algorithm relies on an external monitor to detect environmental changes and trigger the dynamic response strategy. DAEA is tested on the CEC’2022 dynamic multimodal optimization benchmark suite, and is compared against several state-of-the-art dynamic multimodal optimization algorithms. The experimental results demonstrate the competitiveness of DAEA in handling DMMOPs. Additionally, experimental results from the berth allocation problem further confirm the applicability of DAEA to real-world dynamic multimodal optimization tasks.

Many problems in the real-world have more than one objective, with at least two objectives in conflict with one another. In addition, at least one objective changes over time. These kinds of problems are called dynamic multi-objective optimisation problems (DMOOPs). Studies have shown that both the quantum particle swarm optimisation (QPSO) and charged particle swarm optimisation (CPSO) algorithms perform well in dynamic environments, since they maintain swarm diversity. Therefore, this paper investigates the effect of using either QPSOs or CPSOs in the sub-swarms of the dynamic vector-evaluated particle swarm optimisation (DVEPSO) algorithm. These DVEPSO variations are then compared against the default DVEPSO algorithm that uses gbest PSOs and DVEPSO using heterogeneous PSOs that contain both charged and quantum particles. Furthermore, all of the aforementioned DVEPSO configurations are compared against the dynamic multi-objective optimisation (DMOPSO) algorithm that was the winning algorithm of a comprehensive comparative study of dynamic multi-objective optimisation algorithms. The results indicate that charged and quantum particles improve the performance of DVEPSO, especially for DMOOPs with a deceptive POF and DMOOPs with a non-linear POS.

Composite particle optimization with hyper-reflection scheme in dynamic environments

The convergence of vehicular technology, artificial intelligence (AI), and distributed computing has catalyzed the emergence of the Internet of Vehicles (IoV) as a cornerstone of next-generation intelligent transportation systems (ITS). By enabling vehicle-to-everything (V2X) communication, IoV supports cooperative perception, real-time decision-making, and autonomous driving. However, the reliance on large-scale, data-driven intelligence in IoV exposes systems to critical challenges, including adversarial poisoning, privacy leakage, identity forgery, and the fragility of centralized learning architectures. Federated Learning (FL) has been proposed as a promising paradigm to alleviate some of these issues by enabling distributed model training without centralizing sensitive vehicular data. Nonetheless, conventional FL remains vulnerable to security and trust limitations, particularly in dynamic vehicular environments. This thesis addresses these challenges by designing secure, privacy-preserving, and scalable FL frameworks that leverage distributed ledger technologies and cutting-edge security mechanisms.The thesis advances knowledge through four interconnected contributions. First, two novel optimization-driven poisoning attack models are introduced: PA-PSOSA and PAPSOGA, which combine particle swarm optimization with simulated annealing and genetic algorithms, respectively. These models demonstrate that even a small poisoning budget can substantially degrade global model utility under black-box and clean-label constraints, highlighting the urgency of robust defenses in vehicular FL. Second, a permissioned blockchain-enabled FL (BCFL) framework is proposed, in which consortium edge nodes running Practical Byzantine Fault Tolerance (PBFT) consensus replace the central aggregator. With blockchain integration and data validation mechanisms, this design ensures identity authentication, verifiable audit trails, and improved resilience against poisoning and Sybil attacks, while maintaining high model accuracy under adversarial conditions. Third, the framework is further enhanced to achieve inference-resistance by integrating secure aggregation (SecAgg) and differential privacy (DP), and lightweight with off-chain commitments. This design significantly reduces ledger storage requirements, increases system throughput, and mitigates inference-based privacy risks. Finally, to overcome the scalability limitations of PBFT-based BCFL, a DAG-enabled FL (DFL) framework is developed. By leveraging parallel validation, utility-score-based tip selection, and reputation-weighted aggregation, this framework significantly improves scalability, reduces communication complexity, and enhances robustness in asynchronous vehicular environments.Together, these contributions articulate a coherent progression from exposing vulnerabilities in vehicular FL to constructing secure, privacy-preserving, and scalable frameworks tailored for IoV ecosystems. The findings demonstrate that interdisciplinary integration of optimization theory, cryptography, differential privacy, and distributed ledger technologies is indispensable for trustworthy vehicular intelligence. Beyond theoretical significance, the proposed frameworks offer practical designs for deployment in safety-critical IoV environments. Future research directions include the integration of zero-knowledge proofs (ZKP) for verifiable privacy, adaptive defenses against evolving adversarial strategies, and experimental validation in real-world vehicular testbeds. Collectively, this thesis establishes a foundation for secure federated intelligence in IoV, contributing to the reliability, efficiency, and trustworthiness of next-generation ITS.

Phishing is the ubiquitous cyber threat from social engineering and web-deception to compromise user security, which may lead to credential theft, financial loss and data breaches. Originally blacklist-based detection methods cannot be used since the nature of phishing attacks varies and has changed frequently. We present our novel deep learning methodology for phishing URL detection based upon the publicly available PhiUSIIL dataset of large-scale lexical, host-based, content-based URL features in this paper. The new model employs sophisticated feature engineering, correlation analysis, and deep learning techniques to estimate discriminative patterns between benign and malicious URLs. This tool was implemented by a multi-layered deep neural network optimized with Adam optimizer and binary cross-entropy loss with dropout regularization to avoid overfitting. There was an extensive study of the model on the other model classesifiers of machine learning called Logistic Regression, Random Forest and Support Vector Machines. The results show that the deep learning model stands apart from the base models as well with 99.4% accuracy, 98.1% precision, 97.9% recall, 98.0% F1-score, and a strong generalization in absence of test samples. This confirms that the intelligent deep learning model proposed provides ascalable, adaptive, and robust method for phishing URL detection. This framework can be applied to any security applications in real-world such as browser extensions, email filters, and intrusion detection systems to actively defend against phishing threats. For future extensions, hybrid deep learning models will be studied, explained AI processes, and deployment in real-time detection.

In online courses, discussion forums play a key role in enhancing student interaction with peers and instructors. Due to large enrolment sizes, instructors often struggle to respond to students in a timely manner. To address this problem, both traditional Machine Learning (ML) (e.g., Random Forest) and Deep Learning (DL) approaches have been applied to classify educational forum posts (e.g., those required urgent responses vs. that did not). However, there lacks an in-depth comparison between these two kinds of approaches. To better guide people to select an appropriate model, we aimed at providing a comparative study on the effectiveness of six frequently-used traditional ML and DL models across a total of seven different classification tasks centering around two datasets of educational forum posts. Through extensive evaluation, we showed that (i) the up-to-date DL approaches did not necessarily outperform traditional ML approaches; (ii) the performance gap between the two kinds of approaches can be up to 3.68% (measured in F1 score); (iii) the traditional ML approaches should be equipped with carefully-designed features, especially those of common importance across different classification tasks. Based on the derived findings, we further provided insights to help instructors and educators construct effective classifiers for characterizing educational forum discussions, which, ultimately, would enable them to provide students with timely and personalized learning support.

The substantial growth of Internet-connected devices within the Internet of Things (IoT) has given rise to significant security challenges. Among the various threats confronting these interconnected devices, Denial of Service (DoS)/Distributed Denial of Service (DDoS) attacks emerge as significant concerns. The attacks, which seek to disrupt IoT services by flooding networks with unnecessary traffic, there is a critical need for robust security measures. Intrusion Detection Systems (IDS) are vital in identifying suspicious activities, yet many existing systems lack real-time capabilities to address evolving attack strategies. This study investigates the vulnerabilities of IoT networks and the pressing need to detect and classify DoS/DDoS attacks in real time. Traditional IDS, while effective in recognizing known attack patterns, fall short in identifying new attack types due to their reliance on historical data. To bridge this gap, this research focuses on developing an enhanced hybrid network intrusion detection and classification system. This study seeks to make contribution to the advancement of resilient security measures within IoT environments. The framework uses the Kth Nearest Neighbor (KNN) algorithm and dense neural networks to efficiently detect and categorize DoS/DDoS attacks in real time. To achieve this, a simulation model implementing the proposed hybrid algorithm will be developed. The framework utilized the MQTT-IoT-IDS2020 dataset, and a comparison was presented with recent approaches found in the literature. The results demonstrated enhanced performance in terms of true positive rate, accuracy, speed of detection.

As one of the core tasks of natural language processing technology, text classification methods general-ly face the problems of insufficient global semantic capture and limited feature focusing ability when processing long texts or complex semantics. To address this issue, a deep learning model that integrates improved convolutional neural networks, unbalanced bidirectional long short-term memory networks, and multi-head attention mechanisms is proposed. Utilizing an improved bidirectional long short-term memory network to capture global semantic information, while dynamically focusing on key features through a multi head attention mechanism to enhance the model's adaptability to classification tasks. The performance of the model is validated through experiments on AG News (short text) and IMDb (long text) datasets. The results show that in short text classification, the proposed method has an accu-racy rate of 96% and a classification error rate of only 1.46%. In the task of long text classification, the method proposed in the study has a product under the curve of 0.98. In adversarial attack testing, the accuracy rates of adversarial samples generated by different methods are 92.85% and 90.63%, respec-tively, with the lowest robustness degradation rates of 3.72% and 5.49%, respectively. In cross domain generalization testing, it shows the least classification errors and superior cross domain adaptability. These results validate the high performance, robustness, and wide applicability of the method. The re-search indicates that this approach can validly improve the performance of text classification and pro-vide new solutions for natural language processing related tasks in long text and multi-category scenar-ios.

Along with the development of Internet of Vehicles (IoV) and wireless technology, the usage of applications that require low latency, such as autonomous driving and intelligent navigation, is increasing rapidly, and the demand for content is increasing greatly. This paper proposes an edge caching approach for the IoV based on multi-agent deep reinforcement learning (ECSMADRL) so as to resolve the problem of excessive response delay due to the large increase of data traffic in the IoV. The approach jointly considers content distribution and caching in dynamic environments. In other words, each moving vehicle in the IoV can be seen as an agent, and it can make decisions about content caching and content access adaptively according to the changing environment to minimize the delay in the process of content distribution. It is proved by experiments that compared with other methods, the proposed edge caching (EC) approach has better performance in reducing content distribution delay, improving content hit rate and success rate.

Optimization issues are often defined statically, assuming constant environmental conditions.However, in many real-world scenarios, problem environments are dynamic and continuously changing.Thus we need optimization algorithms that could solves those issues in dynamic environments as well.Dynamic optimization problems are change(s) that may occur through the time.Such environments are characterized by uncertainty, temporal changes, and structural complexities, which makes the optimization process a significant challenge.In addressing these challenges, evolutionary algorithms have emerged as one of the most effective approaches for solving dynamic optimization problems (DOPs).Among these algorithms, the Imperialist Competitive Algorithm (ICA), designed based on swarm intelligence and competition among imperialist countries, has garnered considerable attention due to its capability in solving static optimization issues.In this research, Imperialist Competitive Algorithms, inspired by the historical and political processes of colonization and assimilation, have been known as one of the efficient evolutionary algorithms.These algorithms face numerous challenges when dealing with dynamic problems , including reduced population diversity, performance degradation in conditions of rapid environmental changes, and limitations in optimal convergence.These cases indicate the need to develop improved and more adaptable versions of these algorithms.Using concepts such as memory, population clustering, and repulsion mechanisms, this algorithm has been able to maintain population diversity at all stages while increasing the speed of convergence in the face of environmental changes.The key feature of the proposed algorithm is the use of memory to store previous optimal solutions, a clustering mechanism to manage population diversity, and repulsion to prevent unnecessary accumulation of solutions in specific regions.Nevertheless, ICA exhibits poor performance in dynamic environments because it lacks mechanisms to maintain diversity, quick adaptation to environmental changes, and new optima track.This study presents an improved version of ICA aimed at overcoming these limitations.The proposed algorithm incorporates a combination of a memory mechanism and a clustering strategy to enhance its adaptability to environmental changes and preserve diversity within the population.The memory mechanism stores information about previous optima and utilizes it under appropriate conditions to accelerate the optimization process.forclustering method is used for clustering.Clustering in the proposed method ensures that diversity is maintained for the population during the execution of the algorithm.In this study, our goal is to solve problems that change the environment in a global way.That means, the fitness of all points in the environment changes.By testing just one point in the environment and comparing the fitness obtained with its previously stored value, we can detect a change in the environment.On the other hand, the clustering strategy, particularly the k-means technique, to maintain maintain population diversity and prevents the convergence of solutions to specific regions.Together, these two components create a balance between exploration and exploitation, thereby improving the algorithm's performance in dynamic environments.To evaluate the performance of the proposed algorithm, the Moving Peaks Benchmark (MPB) was used as a standard metric.Due to its capability to simulate complex and diverse changes in dynamic environmentsparticularly in Branke's second scenario-MPB is one of the most recognized tools for assessing the performance of dynamic optimization algorithms.The proposed algorithm was evaluated alongside advanced algorithms such as FTmPSO (TMO), RAmQSO-s4, RmNAFSA-s4, TFTmPSO, RFTmPSO, mQSO10 (5+5q), FMSO, CellularPSO, Multi-SwarmPSO, mCPSO, AmQSO*, FTMPSO, almPSO, and CDEPSA.Experimental results demonstrated that the proposed algorithm outperformed other methods in areas such as convergence speed, adaptability to environmental changes, and population diversity preservation.A key feature of the proposed algorithm is its ability to retain identified optima even after environmental changes.Additionally, the use of the k-means clustering technique has ensured that the algorithm effectively avoids excessive focus on specific regions and maintains population diversity while facing complex environmental changes.Another advantage of this algorithm is its scalability in handling dynamic optimization issues with high dimensions and complexities.These findings indicate that the proposed algorithm is not only effective in laboratory settings but also suitable for real-world applications with fast and dynamic changes.