Semi-supervised surface defect detection of wind turbine blades with YOLOv4

Timely inspection of defects on the surfaces of wind turbine blades can effectively prevent unpredictable accidents.To this end,this study proposes a semi-supervised object-detection network based on You Only Looking Once version 4(YOLOv4).A semi-supervised structure comprising a generative adversarial network(GAN)was designed to overcome the difficulty in obtaining sufficient samples and sample labeling.In a GAN,the generator is realized by an encoder-decoder network,where the backbone of the encoder is YOLOv4 and the decoder comprises inverse convolutional layers.Partial features from the generator are passed to the defect detection network.Deploying several unlabeled images can significantly improve the generalization and recognition capabilities of defect-detection models.The small-scale object detection capacity of the network can be improved by enhancing essential features in the feature map by adding the concurrent spatial and channel squeeze and excitation(scSE)attention module to the three parts of the YOLOv4 network.A balancing improvement was made to the loss function of YOLOv4 to overcome the imbalance problem of the defective species.The results for both the single-and multi-category defect datasets show that the improved model can make good use of the features of the unlabeled images.The accuracy of wind turbine blade defect detection also has a significant advantage over classical object detection algorithms,including faster R-CNN and DETR.

A method for cleaning wind power anomaly data by combining image processing with community detection algorithms

Current methodologies for cleaning wind power anomaly data exhibit limited capabilities in identifying abnormal data within extensive datasets and struggle to accommodate the considerable variability and intricacy of wind farm data.Consequently,a method for cleaning wind power anomaly data by combining image processing with community detection algorithms(CWPAD-IPCDA)is proposed.To precisely identify and initially clean anomalous data,wind power curve(WPC)images are converted into graph structures,which employ the Louvain community recognition algorithm and graph-theoretic methods for community detection and segmentation.Furthermore,the mathematical morphology operation(MMO)determines the main part of the initially cleaned wind power curve images and maps them back to the normal wind power points to complete the final cleaning.The CWPAD-IPCDA method was applied to clean datasets from 25 wind turbines(WTs)in two wind farms in northwest China to validate its feasibility.A comparison was conducted using density-based spatial clustering of applications with noise(DBSCAN)algorithm,an improved isolation forest algorithm,and an image-based(IB)algorithm.The experimental results demonstrate that the CWPAD-IPCDA method surpasses the other three algorithms,achieving an approximately 7.23%higher average data cleaning rate.The mean value of the sum of the squared errors(SSE)of the dataset after cleaning is approximately 6.887 lower than that of the other algorithms.Moreover,the mean of overall accuracy,as measured by the F1-score,exceeds that of the other methods by approximately 10.49%;this indicates that the CWPAD-IPCDA method is more conducive to improving the accuracy and reliability of wind power curve modeling and wind farm power forecasting.

How to Better Identify Venture Capital Network Communities:Exploration of A Semi-Supervised Community Detection Method

In the field of Venture Capital(VC),researchers have found that VC companies are more likely to jointly invest with other VC companies.This paper attempts to realize a semi-supervised community detection of the VC network based on the data of VC networking and the list of industry leaders.The main research method is to design the initial label of community detection according to the evolution of components of the VC industry leaders.The results show that the community structure of the VC network has obvious distinguishing characteristics,and the aggregation of these communities is affected by the type of institution,the source of capital,the background of personnel,and the field of investment and the geographical position.Meanwhile,by comparing the results of the semi-supervised community detection algorithm with the results of community detection using extremal optimization,it can be shown to some extent that the semi-supervised community detection results in the VC network are more accurate and reasonable.

Detecting While Accessing:A Semi-Supervised Learning-Based Approach for Malicious Traffic Detection in Internet of Things

In the upcoming large-scale Internet of Things(Io T),it is increasingly challenging to defend against malicious traffic,due to the heterogeneity of Io T devices and the diversity of Io T communication protocols.In this paper,we propose a semi-supervised learning-based approach to detect malicious traffic at the access side.It overcomes the resource-bottleneck problem of traditional malicious traffic defenders which are deployed at the victim side,and also is free of labeled traffic data in model training.Specifically,we design a coarse-grained behavior model of Io T devices by self-supervised learning with unlabeled traffic data.Then,we fine-tune this model to improve its accuracy in malicious traffic detection by adopting a transfer learning method using a small amount of labeled data.Experimental results show that our method can achieve the accuracy of 99.52%and the F1-score of 99.52%with only 1%of the labeled training data based on the CICDDoS2019 dataset.Moreover,our method outperforms the stateof-the-art supervised learning-based methods in terms of accuracy,precision,recall and F1-score with 1%of the training data.

XA-GANomaly: An Explainable Adaptive Semi-Supervised Learning Method for Intrusion Detection Using GANomaly

Intrusion detection involves identifying unauthorized network activity and recognizing whether the data constitute an abnormal network transmission.Recent research has focused on using semi-supervised learning mechanisms to identify abnormal network traffic to deal with labeled and unlabeled data in the industry.However,real-time training and classifying network traffic pose challenges,as they can lead to the degradation of the overall dataset and difficulties preventing attacks.Additionally,existing semi-supervised learning research might need to analyze the experimental results comprehensively.This paper proposes XA-GANomaly,a novel technique for explainable adaptive semi-supervised learning using GANomaly,an image anomalous detection model that dynamically trains small subsets to these issues.First,this research introduces a deep neural network(DNN)-based GANomaly for semi-supervised learning.Second,this paper presents the proposed adaptive algorithm for the DNN-based GANomaly,which is validated with four subsets of the adaptive dataset.Finally,this study demonstrates a monitoring system that incorporates three explainable techniques—Shapley additive explanations,reconstruction error visualization,and t-distributed stochastic neighbor embedding—to respond effectively to attacks on traffic data at each feature engineering stage,semi-supervised learning,and adaptive learning.Compared to other single-class classification techniques,the proposed DNN-based GANomaly achieves higher scores for Network Security Laboratory-Knowledge Discovery in Databases and UNSW-NB15 datasets at 13%and 8%of F1 scores and 4.17%and 11.51%for accuracy,respectively.Furthermore,experiments of the proposed adaptive learning reveal mostly improved results over the initial values.An analysis and monitoring system based on the combination of the three explainable methodologies is also described.Thus,the proposed method has the potential advantages to be applied in practical industry,and future research will explore handling unbalanced real-time datasets in various scenarios.

Semi-supervised community detection on attributed networks using non-negative matrix tri-factorization with node popularity

The World Wide Web generates more and more data with links and node contents,which are always modeled as attributed networks.The identification of network communities plays an important role for people to understand and utilize the semantic functions of the data.A few methods based on non-negative matrix factorization(NMF)have been proposed to detect community structure with semantic information in attributed networks.However,previous methods have not modeled some key factors(which affect the link generating process together),including prior information,the heterogeneity of node degree,as well as the interactions among communities.The three factors have been demonstrated to primarily affect the results.In this paper,we propose a semi-supervised community detection method on attributed networks by simultaneously considering these three factors.First,a semi-supervised non-negative matrix tri-factorization model with node popularity(i.e.,PSSNMTF)is designed to detect communities on the topology of the network.And then node contents are integrated into the PSSNMTF model to find the semantic communities more accurately,namely PSSNMTFC.Parameters of the PSSNMTFC model is estimated by using the gradient descent method.Experiments on some real and artificial networks illustrate that our new method is superior over some related stateof-the-art methods in terms of accuracy.

An Online Fake Review Detection Approach Using Famous Machine Learning Algorithms

Online review platforms are becoming increasingly popular,encouraging dishonest merchants and service providers to deceive customers by creating fake reviews for their goods or services.Using Sybil accounts,bot farms,and real account purchases,immoral actors demonize rivals and advertise their goods.Most academic and industry efforts have been aimed at detecting fake/fraudulent product or service evaluations for years.The primary hurdle to identifying fraudulent reviews is the lack of a reliable means to distinguish fraudulent reviews from real ones.This paper adopts a semi-supervised machine learning method to detect fake reviews on any website,among other things.Online reviews are classified using a semi-supervised approach(PU-learning)since there is a shortage of labeled data,and they are dynamic.Then,classification is performed using the machine learning techniques Support Vector Machine(SVM)and Nave Bayes.The performance of the suggested system has been compared with standard works,and experimental findings are assessed using several assessment metrics.

Community Detection Using Jaacard Similarity with SIM-Edge Detection Techniques

The structure and dynamic nature of real-world networks can be revealed by communities that help in promotion of recommendation systems.Social Media platforms were initially developed for effective communication,but now it is being used widely for extending and to obtain profit among business community.The numerous data generated through these platforms are utilized by many companies that make a huge profit out of it.A giant network of people in social media is grouped together based on their similar properties to form a community.Commu-nity detection is recent topic among the research community due to the increase usage of online social network.Community is one of a significant property of a net-work that may have many communities which have similarity among them.Community detection technique play a vital role to discover similarities among the nodes and keep them strongly connected.Similar nodes in a network are grouped together in a single community.Communities can be merged together to avoid lot of groups if there exist more edges between them.Machine Learning algorithms use community detection to identify groups with common properties and thus for recommen-dation systems,health care assistance systems and many more.Considering the above,this paper presents alternative method SimEdge-CD(Similarity and Edge between's based Community Detection)for community detection.The two stages of SimEdge-CD initiallyfind the similarity among nodes and group them into one community.During the second stage,it identifies the exact affiliations of boundary nodes using edge betweenness to create well defined communities.Evaluation of proposed method on synthetic and real datasets proved to achieve a better accuracy-efficiency trade-of compared to other existing methods.Our proposed SimEdge-CD achieves ideal value of 1 which is higher than existing sim closure like LPA,Attractor,Leiden and walktrap techniques.

Semi-supervised multi-layered clustering model for intrusion detection

A Machine Learning (ML)-based Intrusion Detection and Prevention System (IDPS)requires a large amount of labeled up-to-date training data to effectively detect intrusions and generalize well to novel attacks.However,the labeling of data is costly and becomes infeasible when dealing with big data,such as those generated by Intemet of Things applications.To this effect,building an ML model that learns from non-labeled or partially labeled data is of critical importance.This paper proposes a Semi-supervised Mniti-Layered Clustering ((SMLC))model for the detection and prevention of network intrusion.SMLC has the capability to learn from partially labeled data while achieving a detection performance comparable to that of supervised ML-based IDPS.The performance of SMLC is compared with that of a well-known semi-supervised model (tri-training)and of supervised ensemble ML models, namely Random.Forest,Bagging,and AdaboostM1on two benchmark network-intrusion datasets,NSL and Kyoto 2006+.Experimental resnits show that SMLC is superior to tri-training,providing a comparable detection accuracy with 20%less labeled instances of training data.Furthermore,our results demonstrate that our scheme has a detection accuracy comparable to that of the supervised ensemble models.

Dynamic evolutionary community detection algorithms based on the modularity matrix

Motivated by the relationship of the dynamic behaviors and network structure, in this paper, we present two efficient dynamic community detection algorithms. The phases of the nodes in the network can evolve according to our proposed differential equations. In each iteration, the phases of the nodes are controlled by several parameters. It is found that the phases of the nodes are ultimately clustered into several communities after a short period of evolution. They can be adopted to detect the communities successfully. The second differential equation can dynamically adjust several parameters, so it can obtain satisfactory detection results. Simulations on some test networks have verified the efficiency of the presented algorithms.

Graph Transformer for Communities Detection in Social Networks

Graphs are used in various disciplines such as telecommunication,biological networks,as well as social networks.In large-scale networks,it is challenging to detect the communities by learning the distinct properties of the graph.As deep learning hasmade contributions in a variety of domains,we try to use deep learning techniques to mine the knowledge from large-scale graph networks.In this paper,we aim to provide a strategy for detecting communities using deep autoencoders and obtain generic neural attention to graphs.The advantages of neural attention are widely seen in the field of NLP and computer vision,which has low computational complexity for large-scale graphs.The contributions of the paper are summarized as follows.Firstly,a transformer is utilized to downsample the first-order proximities of the graph into a latent space,which can result in the structural properties and eventually assist in detecting the communities.Secondly,the fine-tuning task is conducted by tuning variant hyperparameters cautiously,which is applied to multiple social networks(Facebook and Twitch).Furthermore,the objective function(crossentropy)is tuned by L0 regularization.Lastly,the reconstructed model forms communities that present the relationship between the groups.The proposed robust model provides good generalization and is applicable to obtaining not only the community structures in social networks but also the node classification.The proposed graph-transformer shows advanced performance on the social networks with the average NMIs of 0.67±0.04,0.198±0.02,0.228±0.02,and 0.68±0.03 on Wikipedia crocodiles,Github Developers,Twitch England,and Facebook Page-Page networks,respectively.

Mutually Enhancing Community Detection and Sentiment Analysis on Twitter Networks

The burgeoning use of Web 2.0-powered social media in recent years has inspired numerous studies on the content and composition of online social networks (OSNs). Many methods of harvesting useful information from social networks’ immense amounts of user-generated data have been successfully applied to such real-world topics as politics and marketing, to name just a few. This study presents a novel twist on two popular techniques for studying OSNs: community detection and sentiment analysis. Using sentiment classification to enhance community detection and community partitions to permit more in-depth analysis of sentiment data, these two techniques are brought together to analyze four networks from the Twitter OSN. The Twitter networks used for this study are extracted from four accounts related to Microsoft Corporation, and together encompass more than 60,000 users and 2 million tweets collected over a period of 32 days. By combining community detection and sentiment analysis, modularity values were increased for the community partitions detected in three of the four networks studied. Furthermore, data collected during the community detection process enabled more granular, community-level sentiment analysis on a specific topic referenced by users in the dataset.

Enhancing Sentiment Analysis on Twitter Using Community Detection

The increasing popularity of social media in recent years has created new opportunities to study the interactions of different groups of people. Never before have so many data about such a large number of individuals been readily available for analysis. Two popular topics in the study of social networks are community detection and sentiment analysis. Community detection seeks to find groups of associated individuals within networks, and sentiment analysis attempts to determine how individuals are feeling. While these are generally treated as separate issues, this study takes an integrative approach and uses community detection output to enable community-level sentiment analysis. Community detection is performed using the Walktrap algorithm on a network of Twitter users associated with Microsoft Corporation’s @technet account. This Twitter account is one of several used by Microsoft Corporation primarily for communicating with information technology professionals. Once community detection is finished, sentiment in the tweets produced by each of the communities detected in this network is analyzed based on word sentiment scores from the well-known SentiWordNet lexicon. The combination of sentiment analysis with community detection permits multilevel exploration of sentiment information within the @technet network, and demonstrates the power of combining these two techniques.

Overlapping Community Detection in Dynamic Networks

Due to the increasingly large size and changing nature of social networks, algorithms for dynamic networks have become an important part of modern day community detection. In this paper, we use a well-known static community detection algorithm and modify it to discover communities in dynamic networks. We have developed a dynamic community detection algorithm based on Speaker-Listener Label Propagation Algorithm (SLPA) called SLPA Dynamic (SLPAD). This algorithm, tested on two real dynamic networks, cuts down on the time that it would take SLPA to run, as well as produces similar, and in some cases better, communities. We compared SLPAD to SLPA, LabelRankT, and another algorithm we developed, Dynamic Structural Clustering Algorithm for Networks Overlapping (DSCAN-O), to further test its validity and ability to detect overlapping communities when compared to other community detection algorithms. SLPAD proves to be faster than all of these algorithms, as well as produces communities with just as high modularity for each network.

Density-based rough set model for hesitant node clustering in overlapping community detection

Overlapping community detection in a network is a challenging issue which attracts lots of attention in recent years.A notion of hesitant node（HN） is proposed. An HN contacts with multiple communities while the communications are not strong or even accidental, thus the HN holds an implicit community structure.However, HNs are not rare in the real world network. It is important to identify them because they can be efficient hubs which form the overlapping portions of communities or simple attached nodes to some communities. Current approaches have difficulties in identifying and clustering HNs. A density-based rough set model（DBRSM） is proposed by combining the virtue of densitybased algorithms and rough set models. It incorporates the macro perspective of the community structure of the whole network and the micro perspective of the local information held by HNs, which would facilitate the further ＂growth＂ of HNs in community. We offer a theoretical support for this model from the point of strength of the trust path. The experiments on the real-world and synthetic datasets show the practical significance of analyzing and clustering the HNs based on DBRSM. Besides, the clustering based on DBRSM promotes the modularity optimization.

The Effects of Centrality Ordering in Label Propagation for Community Detection

In many cases randomness in community detection algorithms has been avoided due to issues with stability. Indeed replacing random ordering with centrality rankings has improved the performance of some techniques such as Label Propagation Algorithms. This study evaluates the effects of such orderings on the Speaker-listener Label Propagation Algorithm or SLPA, a modification of LPA which has already been stabilized through alternate means. This study demonstrates that in cases where stability has been achieved without eliminating randomness, the result of removing random ordering is over fitting and bias. The results of testing seven various measures of centrality in conjunction with SLPA across five social network graphs indicate that while certain measures outperform random orderings on certain graphs, random orderings have the highest overall accuracy. This is particularly true when strict orderings are used in each run. These results indicate that the more evenly distributed solution space which results from complete random ordering is more valuable than the more targeted search that results from centrality orderings.

Community Detection in Dynamic Social Networks

There are many community detection algorithms for discovering communities in networks, but very few deal with networks that change structure. The SCAN (Structural Clustering Algorithm for Networks) algorithm is one of these algorithms that detect communities in static networks. To make SCAN more effective for the dynamic social networks that are continually changing their structure, we propose the algorithm DSCAN (Dynamic SCAN) which improves SCAN to allow it to update a local structure in less time than it would to run SCAN on the entire network. We also improve SCAN by removing the need for parameter tuning. DSCAN, tested on real world dynamic networks, performs faster and comparably to SCAN from one timestamp to another, relative to the size of the change. We also devised an approach to genetic algorithms for detecting communities in dynamic social networks, which performs well in speed and modularity.

Alleviating the Cold Start Problem in Recommender Systems Based on Modularity Maximization Community Detection Algorithm

Recommender system (RS) has become a very important factor in many eCommerce sites. In our daily life, we rely on the recommendation from other persons either by word of mouth, recommendation letters, movie, item and book reviews printed in newspapers, etc. The typical Recommender Systems are software tools and techniques that provide support to people by identifying interesting products and services in online store. It also provides a recommendation for certain users who search for the recommendations. The most important open challenge in Collaborative filtering recommender system is the cold start problem. If the adequate or sufficient information is not available for a new item or users, the recommender system runs into the cold start problem. To increase the usefulness of collaborative recommender systems, it could be desirable to eliminate the challenge such as cold start problem. Revealing the community structures is crucial to understand and more important with the increasing popularity of online social networks. The community detection is a key issue in social network analysis in which nodes of the communities are tightly connected each other and loosely connected between other communities. Many algorithms like Givan-Newman algorithm, modularity maximization, leading eigenvector, walk trap, etc., are used to detect the communities in the networks. To test the community division is meaningful we define a quality function called modularity. Modularity is that the links within a community are higher than the expected links in those communities. In this paper, we try to give a solution to the cold-start problem based on community detection algorithm that extracts the community from the social networks and identifies the similar users on that network. Hence, within the proposed work several intrinsic details are taken as a rule of thumb to boost the results higher. Moreover, the simulation experiment was taken to solve the cold start problem.

Research community detection from multi-relation researcher network based on structure/attribute similarities

Purpose: This paper aims to provide a method to detect research communities based on research interest in researcher network, which combines the topological structure and vertex attributes in a unified manner.Design/methodology/approach: A heterogeneous researcher network has been constructed by combining multiple relations of academic researchers. Vertex attributes and their similarities were considered and calculated. An approach has been proposed and tested to detect research community in research organizations based on this multi-relation researcher network.Findings: Detection of topologically well-connected, semantically coherent and meaningful research community was achieved.Research limitations: The sample size of evaluation experiments was relatively small. In the present study, a limited number of 72 researchers were analyzed for constructing researcher network and detecting research community. Therefore, a large sample size is required to give more information and reliable results.Practical implications: The proposed multi-relation researcher network and approaches for discovering research communities of similar research interests will contribute to collective innovation behavior such as brainstorming and to promote interdisciplinary cooperation.Originality/value: Recent researches on community detection devote most efforts to singlerelation researcher networks and put the main focus on the topological structure of networks.In reality, there exist multi-relation social networks. Vertex attribute also plays an important role in community detection. The present study combined multiple single-relational researcher networks into a multi-relational network and proposed a structure-attribute clustering method for detecting research community in research organizations.

Community detection in signed networks based on discrete-time model

Community detection in signed networks has been studied widely in recent years. In this paper, a discrete difference equation is proposed to imitate the consistently changing phases of the nodes. During the interaction, each node will update its phase based on the difference equation. Each node has many different nodes connected with it, and these neighbors have different influences on it. The similarity between two nodes is applied to describe the influences between them. Nodes with high positive similarities will get together and nodes with negative similarities will be far away from each other.Communities are detected ultimately when the phases of the nodes are stable. Experiments on real world and synthetic signed networks show the efficiency of detection performance. Moreover, the presented method gains better detection performance than two existing good algorithms.