mohammad rahmanimanesh

Message-passing graph neural network (MPNN) is a highly e effective learning tool for graph-structured data but is particularly vulnerable to adversarial attacks on information ow, where poisoned data from attacked nodes can propagate to neighboring nodes and be aggregated, amplifying the negative impact across the graph. The malicious perturbations to node features or graph structures can degrade model performance in defensive decision making, which requires a clean ow. We propose a creative defense mechanism to robustly enhance the resilience of MPGNNs against adversarial perturbations by leveraging a creative knowledge-based fuzzy approach for node classification that is implemented by a rule-based fuzzy model. Our study assigns each neighboring node a membership function, which quantifies its relevance to the target node based on knowledge extracted from structural information, node attributes, and community relationships to the target node. By dynamically adjusting the influence of neighboring nodes, the proposed model mitigates the effects of adversarial noise and reduces the impact of irrelevant or poisoned data, ensuring more robust message aggregation. Furthermore, the fuzzy approach helps prevent over-squashing, a common issue when increasing the number of GNN layers, by maintaining important information through controlled message passing and is followed by picking an opinion leader agent from each egonet to recycle the graph. However, our model does not require prior knowledge of the graph or attack types, making it adaptable to various datasets and attack scenarios. Extensive experiments on benchmark datasets, including Cora, PubMed, and Citeseer, demonstrate greater robustness and better accuracy when compared to state-of-the-art defense strategies with reasonable complexity.

Image classification tasks often involve the challenge of acquiring a sufficient number of labeled training samples, a process that is not only expensive but also time-consuming. In response to this issue, researchers have focused on transfer learning algorithms, which capitalize on prior knowledge to enhance a model training. While numerous existing transfer learning methods concentrate on knowledge transfer between a single-source domain and a single-target domain, the complexity of real-world scenarios is often underestimated. Limited studies have delved into domain adaptation within multi-source environments, where transferring knowledge from multiple sources introduces ambiguity and uncertainty into the learning process. To address this challenge, this study proposes the application of fuzzy rule-based transfer learning, leveraging the inherent ability of fuzzy rules to effectively handle uncertainty. One notable aspect of fuzzy transfer learning, and transfer learning in general, is the unresolved question of effectively combining and utilizing knowledge when multiple source domains are available. This issue is particularly pertinent in scenarios involving diverse datasets from various sources. Consequently, the present study introduces a novel approach to multi-source transfer learning anchored in fuzzy rules. By integrating fuzzy logic, the proposed method aims to provide a robust solution to the challenges posed by knowledge transfer in scenarios with multiple source domains. This research contributes to advancing transfer learning methodologies, offering a nuanced perspective on handling uncertainty in multi-source environments by applying fuzzy rule-based techniques. In conclusion, the significance of transfer learning in image classification tasks is underscored by the inherent challenges of acquiring labeled training data. The conventional focus on single-source to single-target domain transfer has limitations, prompting a shift towards addressing the more realistic and challenging scenarios of multi-source domain adaptation. This study introduces a pioneering approach to multi-source transfer learning, utilizing fuzzy rule-based techniques to effectively navigate the complexities introduced by knowledge transfer from multiple sources. Through this contribution, the research aims to propel advancements in transfer learning methodologies and foster a more comprehensive understanding of handling uncertainty in multi-source environments.

Analyzing user sentiments in marketing and enhancing customer experiences are essential for developing effective marketing strategies. This analysis is crucial for assessing the performance of social media platforms as communication tools. This research was practical in nature and cross-sectional in time, while utilizing both quantitative and qualitative variables within a descriptive research design. The study categorized user tweets without relying on prior knowledge or labeled data, employing fuzzy systems and an unsupervised approach. This advancement in sentiment analysis enabled researchers and practitioners to extract valuable insights from user opinions and emotions within their respective domains and platforms, thereby facilitating informed business decisions aimed at maximizing profitability. As a case study, this empirical research examined user experiences with Samsung and Apple mobile phones from 2022 to the present, classifying sentiments into positive, negative, and neutral categories. Three sentiment analysis tools—SentiWordNet, AFINN, and VADER—were employed to determine the polarity of the tweets. The classification results revealed a higher level of user satisfaction with Samsung mobile phones compared to Apple.

Distributed Denial of Service (DDoS) attack is the widespread sending of valid or invalid packets to a server on the Internet, occupying its bandwidth and preventing execute legitimate requests of other users. The best approach to secure the network from such attacks is to exploit security controls such as intrusion detection and prevention systems. Cyber security researchers have significantly focused on identifying and counteracting this attack and have increased the accuracy and performance of security systems by providing various artificial intelligence solutions. The purpose of this paper is also to provide a solution for detecting DDoS attack, where, decision tree, multi-layer perceptron and random forest classifiers have been utilized in an ensemble method to mitigate the over-fitting problem. Also, two approache, i.e., batch learning and active learning have been implemented and evaluated in the classification phase of the proposed method. The evaluation results show that the mean value of accuracy in DDoS attack detection is 99.81%.

Due to the increasing amount of video data, a lot of research has been done in the field of retrieving and categorizing this type of data. On the other hand, with the growing popularity of football and the increasing number of its audiences, the importance of automatic and real-time extraction of statistics and information about soccer matches has increased. One of the critical and challenging tasks in soccer video analysis is the detection of players’ blobs and regions, along with identifying the teams related to the players. This task encounters many challenges, including grass loss in the playfield, the presence of playfield lines and players' shadows, the overlapping of players with objects and other players, and different shapes of players in different situations. This paper proposes a framework for detecting players and their related teams. For this purpose, an object-sieve-based method is used to detect players’ blobs, and a genetic algorithm is used to identify their related teams. Each chromosome of the genetic algorithm is a window that lies on one blob whose fitness function shows how much its color and shape characteristics fit with the uniforms of each of the two teams. The proposed method was evaluated by 50 different frames of broadcast soccer videos, including 563 players, and 40 different sub-images, including 84 players. The results show 98% and 91.6% precision for player detection and labeling, respectively.

With the explosive growth of threats to Internet security, malware visualization in malware classification has become a promising study area in security and machine learning. This paper proposes a visualization method for malware analysis based on word embedding features of byte sequences.Based on some assistant information such as word embedding, the basic to a strong malware classification approach is to transfer the learned information from the malware domain to the image domain, which needs correlation modeling between these domains. However, most current methods neglect to model the relationships in an embedding way, ensue in low performance of malware classification. To catch this challenge, we consider the Word Embeddings duty as a Semantic Information Extraction. Our Proposed method aims to learn effective representations of malware families, which takes as input a set of embedded vectors corresponding to the malware. Word embedding is designed to generate features of a malware sample by leveraging its malware semantics. Our results show that visual models in the domain of images can be used for efficient malware classification. We evaluated our method on the kaggle dataset of Windows PE file instances, obtaining an average classification accuracy of 0.9896%.

The world is now in the throes of a public health crisis caused by the Covid virus, which has affected people in almost every aspect of daily life and left the world economy in a state of uncertainty. Also, the epidemic of the virus in the payment industry and supply chains has caused many changes, so the present study with an interpretive approach has identified the variables of electronic supply chain stability from the perspective of blockchain activists and university professors. The research approach to the applied method is qualitative methodology and the data method of the foundation has been used to conduct research operations. The data of this study were collected using semi-structured interviews. The sampling method used in this study is purposeful sampling and using the theoretical saturation index, 10 prominent university professors and activists in the field of blockchain were interviewed as a research sample. Their attitudes and views on the research topic were examined. Research findings User relationship flow, service flow, information flow, financial flow, environmental flow affecting the core platform of identifying and managing suppliers, demand management, technology and information management, environmental management as causal and environmental conditions Technological, social environment, economic environment, legal environment as background conditions and political factors, determining international market factors as intervening conditions from the perspective of experts. Finally, policy and planning, infrastructure, awareness and training of appropriate solutions. Which will lead to improved quality of goods and services, improved performance, improved social space, improved financial processes, improved environment.

Blockchain is an emerging technology that experts hope its implementation and integration with the supply chain could improve the transparency, safety, and accuracy of supply chain processes. Reviewing the related literature and consulting/interviewing with 12 relevant experts in the electronic payment services industry who have been selected through purposive judgment sampling, due to the immaturity of the issue of its application and its facilitation in achieving supply chain sustainability - especially in the services industries - this study tried to identify the main and secondary variables of sustainable supply chain performance which are mostly affected by the implementation of this technology, as well as the know-how of using blockchain in each aspect and its solutions. Finally, using the Fuzzy Delphi Method, the most important variables were prioritized based on the degree of this technology's impact on them. The results show that the greatest impact of blockchain implementation in the electronic payment supply chain is on cost, reliability in relationships and transactions, and accountability, respectively; other variables are in the next priorities.

Despite all blockchain's advantages, it has experienced some challenges and disadvantages. Identifying these challenges and minimizing them are important steps in advancing the goals of this technology. The purpose of this study is to investigate and identify the challenges of using blockchain technology in international markets with regard to the risks of such newfound technologies. This is an applied research in terms of purpose and a mixed research in terms of methodology. The effective factors in using blockchain technology have been discovered by the meta-synthesis method at first; then the ranking of these factors has been done by the fuzzy pairwise comparison. A systematic search was performed in the database of Science Direct, Emerald, and Google Scholar. Also, the seven-step method of Sandlowski and Barroso (2007) was applied. All the related articles and the challenges of using blockchain in international markets were identified and coded. The comparison method between the two coders was used to adjust the reliability and the kappa coefficient was used to calculate the correlation coefficient between the two codes. The findings of the extracted sources were finally proposed and 41 challenges were classified into 6 categories of technical, educational, structural, market, infrastructure and legal which divided in both internal and external areas.

Today, the manufacturing industry, with the expansion of the physical constraints of trade worldwide, has adopted modern information technologies to optimize the business process and achieve integration with geographically dispersed supply chain partners. Traditional supply chain models focus on optimizing physical flows. However, it is equally important to ensure that physical units can process appropriate information. This paper aims to propose a model for the optimization of information process performance in the IoT-based virtual supply chain. 

Design/methodology/approach:

 In this study, information processing performance in the closed-loop virtual supply chain has been optimized to maximize profit and information processing speed by considering costs of virtual, information security, and energy consumption. The final programming model has been optimized using meta-heuristic algorithms, the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), and the Strength Pareto Evolutionary Algorithm (SPEA-II).

Findings:

 The results indicated that there is an inverse relationship between virtual supply chain profit and information processing speed (delay). The results of model solving using NSGA-II and SPEA-II algorithms underlined the virtual supply chain profit of 9.93×106 and 4.23×106, and the data processing speed of 337.48 and 94.07, respectively. Thus, the NSGA-II algorithm contributes more to the supply chain profitability.

Research limitations/implications :

The proposed model can be used in manufacturing industries equipped with IoT. Unavailability of practical examples and insufficient data are the two main limitations of the study.

 Practical implications:

 The proposed model improves the production process and helps managers to plan better for their supply chain management and make timely decisions by sharing information across the supply chain and being aware of the flows of products and associated parts.Social implications - The Internet of Things in the virtual supply chain provides an opportunity to manage logistics systems and results in efficient online delivery with minimal cost. The information flow integrates all links and participants in the virtual supply chain. It enables each member to obtain the accurate information needed for logistics capability, reduces resource wastage, and improves customer satisfaction. 

Originality/value: 

One of the innovative aspects of this research is the use of IoT in the virtual supply chain for the integration and transparency of information in the supply chain, considering the importance of information in the virtual supply chain and examining the impact of IoT usage on closed-loop virtual supply costs and target functions. In addition to considering the physical flow costs of the closed-loop, including production costs, separation costs, repair, disposal, recycling, etc., in the cost objective function, virtual flow costs included IoT usage costs and information security costs. Energy consumption was also included in the objective function. Also, due to the virtualization of the supply chain and the significant role of information, optimization of information processing speed was considered in modeling the supply chain performance, which is another innovative aspect of research.

In a variety of real-world scenarios, techniques such as machine learning and data mining are applied. Traditional machine learning frameworks suppose that training data and testing data come from the same domain, have the same feature space, and have the same feature space distribution. This assumption, however, is capable of being applied in certain realistic machine learning cases, especially when gathering training data is prohibitively costly or impossible. As a result, high-performance learners must be developed using data that is more conveniently gathered from various domains. Transfer learning is the name given to this method; it is a learning environment based on a person's capacity to extrapolate information through activities to learn more quickly. Transfer learning tries to establish a structure for applying previous knowledge learned skills to tackle new but related issues more swiftly and efficiently. Transfer learning methodologies, in opposition to traditional machine learning technics, use data from auxiliary domains to enhance predictive modelling of distinct data patterns in the present domain. Transfer learning focuses on improving target participants' performance on target domains by passing data or knowledge from numerous but similar source domains. As a result, the reliance on a various number of target-domain available data for building target learners can be minimized. This survey paper explains transfer learning categories based on problems and solutions and explains experiment results and examples of its application and perspective related to transfer learning. Also, it provides a concise overview of the processes and methods of transfer learning, which may aid readers in better understanding the current research state and idea.

In cancer research, early detection of breast cancer is effective in prognosis and increasing the survival of patients. Artificial intelligence and data mining methods are predictive methods that can be useful in this case. The purpose of this paper is to design and evaluate an automatic computer-aided diagnosis system to predict the type of breast tumor by using a fuzzy combination of regression methods.
In this descriptive-analytic study, a group decision-making system was designed by using the combination of PCA (Principal Components Analysis), RSM (Response Surface Methodology) and SVR-Firefly (Support Vector Regression- Firefly) methods to predict the type of breast tumor (benign or malignant). To evaluate the designed system, we used the Wisconsin Breast Cancer Dataset stored in the UCI (University of California, Irvine) Machine Learning Repository. After pre-processing, the data was split into training and testing datasets, which were evaluated by each of mentioned regression methods. Finally the results of the regression methods were combined by fuzzy voting and fuzzy averaging approaches. The proposed group decision-making system was simulated by MATLAB software.
This study was conducted on 683 women with breast cancer, including 444 cases with benign tumors, as well as 239 cases with malignant tumors. For each case, nine clinical variables were used as inputs. The performance of the proposed group decision-making system in test phase is based on accuracy, precision, sensitivity and specificity. The values of these indicators were obtained, on average, 0.9832, 0.9588, 0.9900, 0.9832 in fuzzy voting method and 0.9820, 0.9524, 0.9929, 0.9804 in fuzzy averaging method while the performance indicators were obtained 1 by both methods in the best case.
The results of experiments show the effectiveness of the proposed group decision-making system in predicting the type of breast tumors (benign or malignant). It can help the physicians to choose the best treatment method and prevent the cancer progression. The developed software can also be used for training physicians.

The survival guide to competitiveness for organizations is to become closer to the customers and deliver value-added products and services in the shortest time. This, in turn, requires integration of the business processes of the enterprise, which is the stronghold of Enterprise Resource Planning (ERP). On the other hand, most ERP projects are believed to be doomed to failures. In this research, the factors leading to failure of ERP projects implementation in Iranian enterprises were identified and surveyed. Subsequently, the factors have been classified and ranked consulting the experts and using fuzzy flowsort inference system. For this purpose, the failure factors were extracted reviewing the previous related researches. After that, the experts merged the synonymous and homogeneous factors. The remaining 58 factors were classified in seven categories including: organizational, project management, human resources, managerial, vendors and consultants, processes-related, and technical factors. The factors were prioritized using the experts’ opinions and were ranked utilizing fuzzy flowsort inference system. Finally, ten criteria were determined as the most critical failure factors of ERP projects implementation in Iranian enterprises.

In this paper, an anomaly detection method in cluster-based mobile ad hoc networks with ad hoc on demand distance vector (AODV) routing protocol is proposed. In the method, the required features for describing the normal behavior of AODV are defined via step by step analysis of AODV and independent of any attack. In order to learn the normal behavior of AODV, a fuzzy averaging method is used for combining one-class support vector machine (OCSVM), mixture of Gaussians (MoG), and self-organizing maps (SOM) one-class classifiers and the combined model is utilized to partially detect the attacks in cluster members. The votes of cluster members are periodically transmitted to the cluster head and final decision on attack detection is carried out in the cluster head. In the proposed method, an adaptive ordered weighted averaging (OWA) operator is used for aggregating the votes of cluster members in the cluster head. Since the network topology, traffic, and environmental conditions of a MANET as well as the number of nodes in each cluster dynamically change, the mere use of a fixed quantifier-based weight generation approach for OWA operator is not efficient. We propose a condition-based weight generation method for OWA operator in which the number of cluster members that participate in decision making may be varying in time and OWA weights are calculated periodically and dynamically based on the conditions of the network. Simulation results demonstrate the effectiveness of the proposed method in detecting rushing, RouteError fabrication, and wormhole attacks.