Search

Malicious social bots pose a serious threat to social network security by spreading false information and guiding bad opinions in social networks. The singularity and scarcity of single organization data and the high cost of labeling social bots have given rise to the construction of federated models that combine federated learning with social bot detection. In this paper, we first combine the federated learning framework with the Relational Graph Convolutional Neural Network (RGCN) model to achieve federated social bot detection. A class-level cross entropy loss function is applied in the local model training to mitigate the effects of the class imbalance problem in local data. To address the data heterogeneity issue from multiple participants, we optimize the classical federated learning algorithm by applying knowledge distillation methods. Specifically, we adjust the client-side and server-side models separately: training a global generator to generate pseudo-samples based on the local data distribution knowledge to correct the optimization direction of client-side classification models, and integrating client-side classification models’ knowledge on the server side to guide the training of the global classification model. We conduct extensive experiments on widely used datasets, and the results demonstrate the effectiveness of our approach in social bot detection in heterogeneous data scenarios. Compared to baseline methods, our approach achieves a nearly 3–10% improvement in detection accuracy when the data heterogeneity is larger. Additionally, our method achieves the specified accuracy with minimal communication rounds.

The interfacial bonding capacity between cement emulsified asphalt composite binder (CEACB) and reclaimed asphalt pavement (RAP) plays a critical role in improving the pavement performance of cold recycled asphalt emulsion mixtures (CRAEMs). This study aims to investigate the formation and development of the interfacial bonding capacity between CEACB and RAP. First, the dynamic wettability and the spreading behaviors of CEACB on RAP surfaces were explored according to the surface free energy theory. Second, digital image processing (DIP) technology was employed to recognize interfacial failure patterns. Lastly, the influence of internal and external factors on the interfacial bonding capacity between CEACB and RAP during the curing process was analyzed via grey relational analysis (GRA). The results indicate that a moderate cement content with a mass ratio of asphalt to cement equivalent to 1.0 can significantly enhance the wettability of CEACB on RAP surfaces. By appropriately prolonging the curing time and controlling the curing temperature, it is possible to increase the bonding strength between CEACB and RAP. Additionally, a strong correlation exists between initial wettability and ultimate bonding capacity during the bonding strength curing process. The good wettability that developed in the initial stage of interfacial strength formation relates to the decreased spalling rate of CEACB on the RAP surface. This study is not only devoted to understanding the mechanisms that can enhance CRAEM performance but also provides important guidance for practical engineering applications of cold recycled asphalt pavements.

Vehicular Ad-hoc Network (VANET) is a significant component of intelligent transportation system, which facilitates vehicles to share sensitive information and corporate with others. However, due to its unique characteristics, such as openness, dynamic topology and high mobility, VANET suffers from various attacks. This paper proposes an anti-attack trust management scheme in VANET called AATMS to evaluate the trustworthiness of vehicles. With the help of AATMS, vehicles in VANET can avoid malicious vehicles and cooperate with trusted vehicles. The idea of AATMS is mainly inspired by TrustRank algorithm, which is used to combat web spams. In this paper, we calculate local trust and global trust, which indicate the local and global trust relationships among vehicles. First, Bayesian inference is adopted to calculate local trust of vehicles based on historical interactions. Then we select a small set of seed vehicles according to local trust and some social factors. Once we identify the reputable seed vehicles, we use the local trust link structure of vehicles to evaluate the global trust of all vehicles. The simulation results show that AATMS can efficiently identify trustworthy and untrustworthy vehicles in VANET even under malicious attacks.

In recent years, machine learning-based intrusion detection systems (IDSs) have proven to be effective; especially, deep neural networks improve the detection rates of intrusion detection models. However, as models become more and more complex, people can hardly get the explanations behind their decisions. At the same time, most of the works about model interpretation focuses on other fields like computer vision, natural language processing, and biology. This leads to the fact that in practical use, cybersecurity experts can hardly optimize their decisions according to the judgments of the model. To solve these issues, a framework is proposed in this paper to give an explanation for IDSs. This framework uses SHapley Additive exPlanations (SHAP), and combines local and global explanations to improve the interpretation of IDSs. The local explanations give the reasons why the model makes certain decisions on the specific input. The global explanations give the important features extracted from IDSs, present the relationships between the feature values and different types of attacks. At the same time, the interpretations between two different classifiers, one-vs-all classifier and multiclass classifier, are compared. NSL-KDD dataset is used to test the feasibility of the framework. The framework proposed in this paper leads to improve the transparency of any IDS, and helps the cybersecurity staff have a better understanding of IDSs' judgments. Furthermore, the different interpretations between different kinds of classifiers can also help security experts better design the structures of the IDSs. More importantly, this work is unique in the intrusion detection field, presenting the first use of the SHAP method to give explanations for IDSs.

The detection and removal of malicious social bots in social networks has become an area of interest in industry and academia. The widely used bot detection method based on machine learning leads to an imbalance in the number of samples in different categories. Classifier bias leads to a low detection rate of minority samples. Therefore, we propose an improved conditional generative adversarial network (improved CGAN) to extend imbalanced data sets before applying training classifiers to improve the detection accuracy of social bots. To generate an auxiliary condition, we propose a modified clustering algorithm, namely, the Gaussian kernel density peak clustering algorithm (GKDPCA), which avoids the generation of data-augmentation noise and eliminates imbalances between and within social bot class distributions. Furthermore, we improve the CGAN convergence judgment condition by introducing the Wasserstein distance with a gradient penalty, which addresses the model collapse and gradient disappearance in the traditional CGAN. Three common oversampling algorithms are compared in experiments. The effects of the imbalance degree and the expansion ratio of the original data on oversampling are studied, and the improved CGAN performs better than the others. Experimental results comparing with three common oversampling algorithms show that the improved CGAN achieves the higher evaluation scores in terms of F1-score, G-mean and AUC.

To explore the advantages of adversarial learning and deep learning, we propose a novel network intrusion detection model called SAVAER-DNN, which can not only detect known and unknown attacks but also improve the detection rate of low-frequent attacks. SAVAER is a supervised variational auto-encoder with regularization, which uses WGAN-GP instead of the vanilla GAN to learn the latent distribution of the original data. SAVAER's decoder is used to synthesize samples of low-frequent and unknown attacks, thereby increasing the diversity of training samples and balancing the training data set. SAVAER's encoder is used to initialize the weights of the hidden layers of the DNN and explore high-level feature representations of the original samples. The benchmark NSL-KDD (KDDTest+), NSL-KDD (KDDTest-21) and UNSW-NB15 datasets are used to evaluate the performance of the proposed model. The experimental results show that the proposed SAVAER-DNN is more suitable for data augmentation than the other three well-known data oversampling methods. Moreover, the proposed SAVAER-DNN outperforms eight well-known classification models in detection performance and is more effective in detecting low-frequent and unknown attacks. Furthermore, compared with other state-of-the-art intrusion detection models reported in the IDS literature, the proposed SAVAER-DNN offers better performance in terms of overall accuracy, detection rate, F1 score, and false positive rate.

Aiming at the problems existing in the application of machine learning algorithm,an optimization system of the machine learning model based on the heuristic algorithm was constructed.Firstly,the existing types of heuristic algorithms and the modeling process of heuristic algorithms were introduced.Then,the advantages of the heuristic algorithm were illustrated from its applications in machine learning,including the parameter and structure optimization of neural network and other machine learning algorithms,feature optimization,ensemble pruning,prototype optimization,weighted voting ensemble and kernel function learning.Finally,the heuristic algorithms and their development directions in the field of machine learning were given according to the actual needs.

Various machine-learning methods have been applied to anomaly intrusion detection. However, the Intrusion Detection System still faces challenges in improving Detection Rate and reducing False Positive Rate. In this paper, a Class-Level Soft-Voting Ensemble (CLSVE) scheme based on the Chaos Bat Algorithm (CBA), called CBA-CLSVE, is proposed for intrusion detection. The Support Vector Machine (SVM), K-Nearest Neighbor (KNN) and Decision Tree (DT) are selected as the base learners of the ensemble. The Chaos Bat Algorithm is used to generate class-level weights to create the weighted voting ensemble. A weighted fitness function considering the tradeoff between maximizing Detection Rate and minimizing False Positive Rate is proposed. In the experiments, the NSL-KDD, UNSW-NB15 and CICIDS2017 datasets are used to verify the scheme. The experimental results show that the class-level weights generated by CBA can be used to improve the combinative performance. They also show that the same ensemble performance can be achieved using about half the total number of features or fewer.

In order to improve user authentication accuracy based on keystroke dynamics and mouse dynamics in hybrid scenes and to consider the user operation changes in different scenes that aggravate user status changes and make it difficult to simulate user behaviors, we present a user authentication method entitled SIURUA. SIURUA uses scene-irrelated features and user-related features for user identification. First, features are extracted based on keystroke data and mouse movement data. Next, scene-irrelated features that have a low correlation with scenes are obtained. Finally, scene-irrelated features are fused with user-related features to ensure the integrity of the features. Experimental results show that the proposed method has the advantage of improving user authentication accuracy in hybrid scenes, with an accuracy of 84% obtained in the experiment.

Keystroke biometrics is a well-investigated dynamic behavioral methodology that utilizes the unique behavioral patterns of users to verify their identity when tapping keys. However, the performance of keystroke biometrics is unreliable due to its high error rate and low robustness. In this paper, we propose differential evolution and adversarial noise-based user authentication (DEANUA), which is a verification scheme for enhancing reliability by reducing the error rate and improving robustness. We investigate the current mainstream features and build a more comprehensive feature set that composed of 146 features. Then, we use a differential evolution method to select an optimized feature set. With the support vector regression method on this feature set, we achieve an equal error rate (EER) of 0.12660% and also a 31.25% energy consumption reduction rate. In this paper, the model is trained with the training samples collected from one situation, but the model is used in various situations. Thus, the robustness of the model is inadequate. We constructed the adversarial noise samples to simulate users' behavioral characteristics in different situational contexts. We use the adversarial noise samples to test the models in a strict experimental environment, which raises the EER by 83.59%, to 10.9299%. Then, we enhance the model with adversarial noise samples to obtain an EER of 8.70932%, which is a reduction of 20.32%.

Social engineering has been increasingly used during the past few years. Social engineering attacks have resulted in great financial losses. Research on social engineering models and frameworks is still in its elementary stage. An appropriate social engineering framework can interpret all the attack components and their relationships clearly, which will contribute to the defense of social engineering attacks. In this tutorial paper, existing social engineering models and frameworks are summarized and a new social engineering framework is proposed involving the concept of the session and dialogue. An entire social engineering attack is defined as a social engineering session (SES). A social engineering dialogue (SED) refers to a specific attack phase, which is included in a SES. A SES contains several well-organized SEDs. Then, the attack graph is used to formalize the proposed social engineering framework. The SED is treated as an atomic attack during the whole SES. The human weaknesses that an attacker can exploit are described as vulnerabilities, the information, and trust that an attacker owns as permissions. Finally, three real-world social engineering cases are analyzed using the proposed framework and attack graph. The analyses illustrate the usability of the proposed framework and provide a better understanding of various social engineering attacks.

In a real network environment, multiple types of attacks can occur. The more important the service or network, the more attacks it may suffer simultaneously. Moving target defense (MTD) technology is a revolutionary game-changing cyberspace technology that has found various applications in recent years. However, the existing strategies are targeted at defending against specific types of attacks and do not meet the security requirements for multiple attacks. Therefore, we propose a joint defense strategy based on the MTD that can select one or multiple mutant elements to defend against different types of attacks. In addition, we use the analytic hierarchy process (AHP) to quantify the factors affecting the attack and defense costs. After comprehensively analyzing the effects of the different MTD technologies against different attacks, we propose an efficient strategy selection algorithm based on joint defense. Finally, we conduct experiments to evaluate the selection of a joint defense strategy under multiple attacks. The experimental results demonstrate the feasibility and effectiveness of the proposed joint defense strategy selection approach.

The main challenge of user personality recognition is low accuracy resulting from small sample size and severe sample distribution imbalance. This paper analyzes the impact of imbalanced data distribution and positive and negative sample overlap on the machine learning classification model. The classification model is based on the data resampling technique, which can improve the classification accuracy. These problems can be solved once the data are effectively resampled. We present a personality prediction method based on particle swarm optimization (PSO) and synthetic minority oversampling technique+Tomek Link (SMOTETomek)resampling (PSO-SMOTETomek), which, apart from effective SMOTETomek resampling of data samples, is able to execute PSO feature optimization for each set of feature combinations. Validated by simulation, our analysis reveals that the PSO-SMOTETomek method is efficient under a small dataset, and the accuracy of personality recognition is improved by up to around 10%. The results are better than those of previous similar studies. The average accuracies of the plain text dataset and the non-plain text dataset are 75.34% and 78.78%, respectively. The average accuracies of the short text dataset and the long text dataset are 75.34% and 64.25%, respectively. From the experimental results, we found that short text has a better classification effect than long text. Plain text data can still have high personality discrimination accuracy, but there is no relevant external information. The proposed model is able to facilitate the design and implementation of a personality recognition system, and the model significantly outperforms existing state-of-the-art models.

Software-defined network (SDN) provides a solution for the scalable network framework with decoupled control and data plane. Migrating switches can balance the resource utilization of controllers and improve network performance. Switch migration problem has to date been formulated as a resource utilization maximization problem to address the scalability of the control plane. However, this problem is NP-hard with high-computational complexities and without addressing the security challenges of the control plane. In this paper, we propose a switch migration method, which interprets switch migration as a signature matching problem and is formulated as a 3-D earth mover's distance model to protect strategically important controllers in the network. Considering the scalability, we further propose a heuristic method which is time-efficient and suitable to large-scale networks. Simulation results show that our proposed methods can disguise strategically important controllers by diminishing the difference of traffic load between controllers. Moreover, our proposed methods can significantly relieve the traffic pressure of controllers and prevent saturation attacks.

Phishing has become one of the biggest and most effective cyber threats, causing hundreds of millions of dollars in losses and millions of data breaches every year. Currently, anti-phishing techniques require experts to extract phishing sites features and use third-party services to detect phishing sites. These techniques have some limitations, one of which is that extracting phishing features requires expertise and is time-consuming. Second, the use of third-party services delays the detection of phishing sites. Hence, this paper proposes an integrated phishing website detection method based on convolutional neural networks (CNN) and random forest (RF). The method can predict the legitimacy of URLs without accessing the web content or using third-party services. The proposed technique uses character embedding techniques to convert URLs into fixed-size matrices, extract features at different levels using CNN models, classify multi-level features using multiple RF classifiers, and, finally, output prediction results using a winner-take-all approach. On our dataset, a 99.35% accuracy rate was achieved using the proposed model. An accuracy rate of 99.26% was achieved on the benchmark data, much higher than that of the existing extreme model.

Based on the openness and accessibility of user data, personality recognition is widely used in personalized recommendation, intelligent medicine, natural language processing, and so on. Existing approaches usually adopt a single deep learning mechanism to extract personality information from user data, which leads to semantic loss to some extent. In addition, researchers encode scattered user posts in a sequential or hierarchical manner, ignoring the connection between posts and the unequal value of different posts to classification tasks. We propose a hierarchical hybrid model based on a self-attention mechanism, namely HMAttn-ECBiL, to fully excavate deep semantic information horizontally and vertically. Multiple modules composed of convolutional neural network and bi-directional long short-term memory encode different types of personality representations in a hierarchical and partitioned manner, which pays attention to the contribution of different words in posts and different posts to personality information and captures the dependencies between scattered posts. Moreover, the addition of a word embedding module effectively makes up for the original semantics filtered by a deep neural network. We verified the hybrid model on the MyPersonality dataset. The experimental results showed that the classification performance of the hybrid model exceeds the different model architectures and baseline models, and the average accuracy reached 72.01%.

Malicious social media bots are disseminators of malicious information on social networks and seriously affect information security and the network environment. Efficient and reliable classification of social media bots is crucial for detecting information manipulation in social networks. Aiming to correct the defects of high-cost labeling and unbalanced positive and negative samples in the existing methods of social media bot detection, and to reduce the training of abnormal samples in the model, we propose an anomaly detection framework based on a combination of a Variational AutoEncoder and an anomaly detection algorithm. The purpose is to use Variational AutoEncoder to automatically encode and decode sample features. The normal sample features are more similar to the initial features after decoding; however, there is a difference between the abnormal samples and the initial features. The decoding representation and the original features are combined, and then the anomaly detection method is used for detection. The results show that the area under the curve of the proposed model for identifying social media bots reaches 98% through the experiments on public datasets, which can effectively distinguish bots from common users and further verify the performance of the proposed model.

Cross-domain relation extraction has become an essential approach when target domain lacking labeled data. Most existing works adapted relation extraction models from the source domain to target domain through aligning sequential features, but failed to transfer non-local and non-sequential features such as word co-occurrence which are also critical for cross-domain relation extraction. To address this issue, in this paper, we propose a novel tripartite graph architecture to adapt non-local features when there is no labeled data in the target domain. The graph uses domain words as nodes to model the co-occurrence relation between domain-specific words and domain-independent words. Through graph convolutions on the tripartite graph, the information of domain-specific words is propagated so that the word representation can be fine-tuned to align domain-specific features. In addition, unlike the traditional graph structure, the weights of edges innovatively combine fixed weight and dynamic weight, to capture the global non-local features and avoid introducing noise to word representation. Experiments on three domains of ACE2005 datasets show that our method outperforms the state-of-the-art models by a big margin.

While there have been significant advances in detecting emotions in text, in the field of utterance-level emotion recognition (ULER), there are still many problems to be solved. In this paper, we address some challenges in ULER in dialog systems. (1) The same utterance can deliver different emotions when it is in different contexts. (2) Long-range contextual information is hard to effectively capture. (3) Unlike the traditional text classification problem, for most datasets of this task, they contain inadequate conversations or speech. (4) To better model the emotional interaction between speakers, speaker information is necessary. To address the problems of (1) and (2), we propose a hierarchical transformer framework (apart from the description of other studies, the “transformer” in this paper usually refers to the encoder part of the transformer) with a lower-level transformer to model the word-level input and an upper-level transformer to capture the context of utterance-level embeddings. For problem (3), we use bidirectional encoder representations from transformers (BERT), a pretrained language model, as the lower-level transformer, which is equivalent to introducing external data into the model and solves the problem of data shortage to some extent. For problem (4), we add speaker embeddings to the model for the first time, which enables our model to capture the interaction between speakers. Experiments on three dialog emotion datasets, Friends, EmotionPush, and EmoryNLP, demonstrate that our proposed hierarchical transformer network models obtain competitive results compared with the state-of-the-art methods in terms of the macro-averaged F1-score (macro-F1).

In previous work, imbalanced datasets composed of more benign samples (the majority class) than the malicious one (the minority class) have been widely adopted in Android malware detection. These imbalanced datasets bias learning toward the majority class, so that the minority class examples are more likely to be misclassified. To solve the problem, we propose a new oversampling method called fuzzy–synthetic minority oversampling technique, which is based on fuzzy set theory and the synthetic minority oversampling technique method. As the sample size of the majority class increases relative to that of the minority class, fuzzy–synthetic minority oversampling technique generates more synthetic examples for each minority class examples in the fuzzy region, where the minority examples have a low degree of membership to the minority class and are more likely to be misclassified. Using the new synthetic examples, the classifiers build larger decision regions that contain more minority examples, and they are no longer biased to the majority class. Compared with synthetic minority oversampling technique and Borderline–synthetic minority oversampling technique methods, fuzzy–synthetic minority oversampling technique achieves higher accuracy on both the minority class and the entire datasets.

Intrusion detection systems play an important role in preventing security threats and protecting networks from attacks. However, with the emergence of unknown attacks and imbalanced samples, traditional machine learning methods suffer from lower detection rates and higher false positive rates. We propose a novel intrusion detection model that combines an improved conditional variational AutoEncoder (ICVAE) with a deep neural network (DNN), namely ICVAE-DNN. ICVAE is used to learn and explore potential sparse representations between network data features and classes. The trained ICVAE decoder generates new attack samples according to the specified intrusion categories to balance the training data and increase the diversity of training samples, thereby improving the detection rate of the imbalanced attacks. The trained ICVAE encoder is not only used to automatically reduce data dimension, but also to initialize the weight of DNN hidden layers, so that DNN can easily achieve global optimization through back propagation and fine tuning. The NSL-KDD and UNSW-NB15 datasets are used to evaluate the performance of the ICVAE-DNN. The ICVAE-DNN is superior to the three well-known oversampling methods in data augmentation. Moreover, the ICVAE-DNN outperforms six well-known models in detection performance, and is more effective in detecting minority attacks and unknown attacks. In addition, the ICVAE-DNN also shows better overall accuracy, detection rate and false positive rate than the nine state-of-the-art intrusion detection methods.

Machine learning plays an important role in building intrusion detection systems. However, with the increase of data capacity and data dimension, the ability of shallow machine learning is becoming more limited. In this paper, we propose a fuzzy aggregation approach using the modified density peak clustering algorithm (MDPCA) and deep belief networks (DBNs). To reduce the size of the training set and the imbalance of the samples, MDPCA is used to divide the training set into several subsets with similar sets of attributes. Each subset is used to train its own sub-DBNs classifier. These sub-DBN classifiers can learn and explore high-level abstract features, automatically reduce data dimensions, and perform classification well. According to the nearest neighbor criterion, the fuzzy membership weights of each test sample in each sub-DBNs classifier are calculated. The output of all sub-DBNs classifiers is aggregated based on fuzzy membership weights. Experimental results on the NSL-KDD and UNSW-NB15 datasets show that our proposed model has higher overall accuracy, recall, precision and F1-score than other well-known classification methods. Furthermore, the proposed model achieves better performance in terms of accuracy, detection rate and false positive rate compared to the state-of-the-art intrusion detection methods.

Considering trust issues in cloud computing, we analyze the feasibility of adopting ant colony optimization algorithm to simulate trust relationships between entities in the cloud and then propose a novel behavior trust model: ACO-BTM. Trust relationships between entities in cloud computing are dynamic, uncertain and hard to quantify. ACO-BTM introduces the conception of ‘pheromone’ and transition probability to represent behavior trust. Then, it focuses on the research of dynamic trust evaluation, time constraint and some other issues. Furthermore, a detailed algorithm process of behavior trust evaluation is given in this context. Finally, ACO-BTM is applied to cloud computing platform to simulate the establishment of behavior trust relationships. The simulation experiment verifies that trust degree change with time varies and the frequency of interactions. Compared with the other model, ACO-BTM can provide better trust recommendation services and protect against attacks of malicious nodes effectively in cloud computing environment. It is proved that ACO-BTM has good flexibility, accuracy and robustness.

User identity linkage (UIL) refers to identifying user accounts belonging to the same identity across different social media platforms. Most of the current research is based on text analysis, which fails to fully explore the rich image resources generated by users, and the existing attempts touch on the multimodal domain, but still face the challenge of semantic differences between text and images. Given this, we investigate the UIL task across different social media platforms based on multimodal user-generated contents (UGCs). We innovatively introduce the efficient user identity linkage via aligned multi-modal features and temporal correlation (EUIL) approach. The method first generates captions for user-posted images with the BLIP model, alleviating the problem of missing textual information. Subsequently, we extract aligned text and image features with the CLIP model, which closely aligns the two modalities and significantly reduces the semantic gap. Accordingly, we construct a set of adapter modules to integrate the multimodal features. Furthermore, we design a temporal weight assignment mechanism to incorporate the temporal dimension of user behavior. We evaluate the proposed scheme on the real-world social dataset TWIN, and the results show that our method reaches 86.39% accuracy, which demonstrates the excellence in handling multimodal data, and provides strong algorithmic support for UIL. [ABSTRACT FROM AUTHOR]

Rapid advancement of intelligent transportation systems (ITS) and autonomous driving (AD) have shown the importance of accurate and efficient detection of traffic signs. However, certain drawbacks, such as balancing accuracy and real-time performance, hinder the deployment of traffic sign detection algorithms in ITS and AD domains. In this study, a novel traffic sign detection algorithm was proposed based on the bidirectional Res2Net architecture to achieve an improved balance between accuracy and speed. An enhanced backbone network module, called C2Net, which uses an upgraded bidirectional Res2Net, was introduced to mitigate information loss in the feature extraction process and to achieve information complementarity. Furthermore, a squeeze-and-excitation attention mechanism was incorporated within the channel attention of the architecture to perform channel-level feature correction on the input feature map, which effectively retains valuable features while removing non-essential features. A series of ablation experiments were conducted to validate the efficacy of the proposed methodology. The performance was evaluated using two distinct datasets: the Tsinghua-Tencent 100K and the CSUST Chinese traffic sign detection benchmark 2021. On the TT100K dataset, the method achieves precision, recall, andMap0.5 scores of 83.3%, 79.3%, and 84.2%, respectively. Similarly, on the CCTSDB 2021 dataset, the method achieves precision, recall, and Map0.5 scores of 91.49%, 73.79%, and 81.03%, respectively. Experimental results revealed that the proposed method had superior performance compared to conventional models, which includes the faster region-based convolutional neural network, single shot multibox detector, and you only look once version 5. [ABSTRACT FROM AUTHOR]

Social engineering attacks are considered one of themost hazardous cyberattacks in cybersecurity, as human vulnerabilities are often the weakest link in the entire network. Such vulnerabilities are becoming increasingly susceptible to network security risks. Addressing the social engineering attack defense problem has been the focus of many studies. However, two main challenges hinder its successful resolution. Firstly, the vulnerabilities in social engineering attacks are unique due to multistage attacks, leading to incorrect social engineering defense strategies. Secondly, social engineering attacks are real-time, and the defense strategy algorithms based on gaming or reinforcement learning are too complex to make rapid decisions. This paper proposes a multiattribute quantitative incentive method based on human vulnerability and an improved Q-learning (IQL) reinforcement learning method on human vulnerability attributes. The proposed algorithm aims to address the two main challenges in social engineering attack defense by using a multiattribute incentive method based on human vulnerability to determine the optimal defense strategy. Furthermore, the IQL reinforcement learning method facilitates rapid decision-making during real-time attacks. The experimental results demonstrate that the proposed algorithm outperforms the traditional Qlearning (QL) and deep Q-network (DQN) approaches in terms of time efficiency, taking 9.1% and 19.4% less time, respectively. Moreover, the proposed algorithm effectively addresses the non-uniformity of vulnerabilities in social engineering attacks and provides a reliable defense strategy based on human vulnerability attributes. This study contributes to advancing social engineering attack defense by introducing an effective and efficient method for addressing the vulnerabilities of human factors in the cybersecurity domain. [ABSTRACT FROM AUTHOR]