Feature optimization

Breast cancer occurs when normal breast cells turn cancerous, grow abnormally and form tumors. The most common cancer impacting women worldwide is breast cancer. Diagnosing breast cancer early and accurately is crucial for giving the correct treatment and ensuring patients receive the best care possible. Due to human error, misdiagnosis is a possibility in the medical field. Over-diagnosis can cause patients to go through unnecessary treatments. Underdiagnosis can allow malignant tumors to become more aggressive and lifethreatening. The aim of our research is to create a dependable model to correctly diagnose breast cancer. We propose to use a rough set ensemble classifier approach to assist doctors in making more accurate diagnosis. The rough set reduct algorithm will be used for feature reductions and the model will be built with logistic regression algorithm, Support Vector Machine (SVM) algorithms and random forest algorithm. The proposed model produced an accuracy of 93% for logistic regression algorithm, 97% for SVM, and 92% for Random Forest when classifying the image data and overall produced a 96% accuracy.

In the current digital era, the advancement of network-based technologies has brought a surge in security vulnerabilities, necessitating complex and dynamic defense mechanisms. This paper explores the integration of machine learning techniques within intrusion detection systems (IDS) to tackle the intricacies of modern network threats. A detailed comparative analysis of various algorithms, including k-nearest neighbors (KNN), logistic regression, and perceptron neural networks, is conducted to evaluate their efficiency in detecting and classifying different types of network intrusions such as denial of service (DoS), probe, user to root (U2R), and remote to local (R2L). Utilizing the national software laboratory knowledge discovery and data mining (NSL-KDD) dataset, a standard in the field, the study examines the algorithms' ability to identify complex patterns and anomalies indicative of security breaches. Principal component analysis is utilized to streamline the dataset into 20 principal components for data processing efficiency. Results indicate that the neural network model is particularly effective, demonstrating exceptional performance metrics across accuracy, precision, and recall in both training and testing phases, affirming its reliability and utility in IDS. The potential for hybrid models combining different machine learning (ML) strategies is also discussed, highlighting a path towards more robust and adaptable IDS solutions.

Existing defense tools against the insider attacks are rare, not in real time fashion and suffer from low detection accuracy as the attacks become more sophisticated. Thus, a detection tool with online learning ability and better accuracy is required urgently. This study proposes an insider attack detection model by leveraging entity behavior analysis technique based on a memory prediction model combined with the recursive feature elimination (RFE) feature selection algorithm. The memory-prediction model provides ability to perform online learning, while the RFE algorithm is deployed to reduce data dimensionality. Dataset for the experiment was created from a real network with 150 active users, and mixed with attacks data from publicly available dataset. The dataset is simulated on a testbed network environment consisting of a server configured to run 4 virtual servers and other two computers as traffic generator and detection tool. The experimental results show 94.01% of detection accuracy, 95.64% of precision, 99.28% of sensitivity, and 96.08% of F1-score. The proposed model is able to perform on-the-fly learning to address evolving nature of the attacks. Combining memory prediction models with the RFE for user behavior analysis is a promising approach, and achieving high accuracy is definitely a positive outcome.

Machine Learning (ML) requires a certain number of features (i.e., attributes) to train the model. One of the main challenges is to determine the right number and the type of such features out of the given dataset's attributes. It is not uncommon for the ML process to use dataset of available features without computing the predictive value of each. Such an approach makes the process vulnerable to overfit, predictive errors, bias, and poor generalization. Each feature in the dataset has either a unique predictive value, redundant, or irrelevant value. However, the key to better accuracy and fitting for ML is to identify the optimum set (i.e., grouping) of the right feature set with the finest matching of the feature's value. This paper proposes a novel approach to enhance the Feature Engineering and Selection (eFES) Optimization process in ML. eFES is built using a unique scheme to regulate error bounds and parallelize the addition and removal of a feature during training. eFES also invents local gain (LG) and global gain (GG) functions using 3D visualizing techniques to assist the feature grouping function (FGF). FGF scores and optimizes the participating feature, so the ML process can evolve into deciding which features to accept or reject for improved generalization of the model. To support the proposed model, this paper presents mathematical models, illustrations, algorithms, and experimental results. Miscellaneous datasets are used to validate the model building process in Python, C#, and R languages. Results show the promising state of eFES as compared to the traditional feature selection process.

Distributed denial of service (DDoS) attack is an attempt to make an online service unavailable by overwhelming it with traffic from multiple sources. Therefore, it is necessary to propose an effective method to detect DDoS attack from massive data traffics. However, the existing schemes have some limitations, including that supervised learning methods, need large numbers of labeled data and unsupervised learning algorithms have relatively low detection rate and high false positive rate. In order to tackle these issues, this paper presents a semi-supervised weighted k-means detection method. Specifically, we firstly present a Hadoop-based hybrid feature selection algorithm to find the most effective feature sets and propose an improved density-based initial cluster centers selection algorithm to solve the problem of outliers and local optimal. Then, we provide the Semi-supervised K-means algorithm using hybrid feature selection (SKM-HFS) to detect attacks. Finally, we exploit DARPA DDoS dataset, CAIDA ''DDoS attack 2007'' dataset, CICIDS ''DDoS attack 2017'' dataset and real-world dataset to carry out the verification experiment. The experiment results have demonstrated that the proposed method outperforms the benchmark in the respect of detection performance and technique for order preference by similarity to an ideal solution (TOPSIS) evaluation factor. INDEX TERMS DDoS attack, semi-supervised k-means, Hadoop-based hybrid feature selection, ratio of average sum of squared errors (SSE) to cluster distance (RSD), TOPSIS.

Face recognition in video surveillance helps to identify an individual by comparing facial features of given photograph or sketch with a video for criminal investigations. Generally, face sketch is used by the police when suspect's photo is not available. Manual matching of facial sketch with suspect's image in a long video is tedious and time-consuming task. To overcome these drawbacks, this paper proposes an accurate face recognition technique to recognize a person based on his sketch in an unconstrained video surveillance. In the proposed method, surveillance video and sketch of suspect is taken as an input. Firstly, input video is converted into frames and summarized using the proposed quality indexed three step cross search algorithm. Next, faces are detected by proposed modified Viola-Jones algorithm. Then, necessary features are selected using the proposed salp-cat optimization algorithm. Finally, these features are fused with scale-invariant feature transform (SIFT) features and Euclidean distance is computed between feature vectors of sketch and each face in a video. Face from the video having lowest Euclidean distance with query sketch is considered as suspect's face. The proposed method's performance is analyzed on Chokepoint dataset and the system works efficiently with 89.02% of precision, 91.25% of recall and 90.13% of F-measure.

Network Biology (ISSN 2220-8879; CODEN NBEICS)

Volume 13, Number 2, 1 June 2023 
http://www.iaees.org/publications/journals/nb/articles/2023-13(2)/2023-13(2).asp

Articles

A machine learning approach to predict autism spectrum disorder (ASD) for both children and adults using feature optimization
Network Biology, 2023, 13(2): 37-52
Khandaker Mohammad Mohi Uddin, Hasibur Rahman, Mahadi Hasan, Fatema Akter, Suman Chandra Das
[Abstract] [XML] [EndNote] [RefManager] [BibTex] [DOAJ] [PubMed]
[ Full PDF (1546K)]
[Email Article] [Comment/Review Article]

Information system of acupoint diagnosis and treatment in Traditional Chinese Medicine
Network Biology, 2023, 13(2): 53-73
WenJun Zhang, YanHong Qi
[Abstract] [XML] [EndNote] [RefManager] [BibTex] [DOAJ] [PubMed]
[ Full PDF (1179K)][Supplementary Material (2639K)][Chinese Version (2056K)]
[Email Article] [Comment/Review Article]

Anomaly detection is a significant research area in data science. Anomaly detection is used to find unusual points or uncommon events in data streams. It is gaining popularity not only in the business world but also in different of other fields, such as cyber security, fraud detection for financial systems, and healthcare. Detecting anomalies could be useful to find new knowledge in the data. This study aims to build an effective model to protect the data from these anomalies. We propose a new hyper ensemble machine learning method that combines the predictions from two methodologies the outcomes of isolation forest-k-means and random forest using a voting majority. Several available datasets, including KDD Cup-99, Credit Card, Wisconsin Prognosis Breast Cancer (WPBC), Forest Cover, and Pima, were used to evaluate the proposed method. The experimental results exhibit that our proposed model gives the highest realization in terms of receiver operating characteristic performance, accu...

Image fusion technique applied in various field of medical science for diagnose of serious medical disease. The brain stroke detection in head is very difficult task because the tissues of head are important factor. Now the detection process used high intensity multimedia camera for gathering brain stroke image for analysis of stroke. In this paper proposed a feature optimisation technique for brain stroke detection. The proposed technique based on wavelet transforms function and particle of swarm optimisation technique. The proposed algorithm performs better in compression of pervious method such as WT and IWT transform method.

Anomaly detection is a significant research area in data science. Anomaly detection is used to find unusual points or uncommon events in data streams. It is gaining popularity not only in the business world but also in different of other fields, such as cyber security, fraud detection for financial systems, and healthcare. Detecting anomalies could be useful to find new knowledge in the data. This study aims to build an effective model to protect the data from these anomalies. We propose a new hyper ensemble machine learning method that combines the predictions from two methodologies the outcomes of isolation forest-k-means and random forest using a voting majority. Several available datasets, including KDD Cup-99, Credit Card, Wisconsin Prognosis Breast Cancer (WPBC), Forest Cover, and Pima, were used to evaluate the proposed method. The experimental results exhibit that our proposed model gives the highest realization in terms of receiver operating characteristic performance, accuracy, precision, and recall. Our approach is more efficient in detecting anomalies than other approaches. The highest accuracy rate achieved is 99.9%, compared to accuracy without a voting method, which achieves 97%.

Network data is expanding and that too at an alarming rate. Besides, the sophisticated attack tools used by hackers lead to capricious cyber threat landscape. Traditional models proposed in the field of network intrusion detection using machine learning algorithms emphasize more on improving attack detection rate and reducing false alarms but time efficiency is often overlooked. Therefore, in order to address this limitation, a modern solution has been presented using Machine Learning-as-a-Service platform. The proposed work analyses the performance of eight two-class and three multiclass algorithms using UNSW NB-15, a modern intrusion detection dataset. 82,332 testing samples were considered to evaluate the performance of algorithms. The proposed two class decision forest model exhibited 99.2% accuracy and took 6 seconds to learn 1,75,341 network instances. Multiclass classification task was also undertaken wherein attack types like generic, exploits, shellcode and worms were class...

Face recognition in video surveillance helps to identify an individual by comparing facial features of given photograph or sketch with a video for criminal investigations. Generally, face sketch is used by the police when suspect's photo is not available. Manual matching of facial sketch with suspect's image in a long video is tedious and time-consuming task. To overcome these drawbacks, this paper proposes an accurate face recognition technique to recognize a person based on his sketch in an unconstrained video surveillance. In the proposed method, surveillance video and sketch of suspect is taken as an input. Firstly, input video is converted into frames and summarized using the proposed quality indexed three step cross search algorithm. Next, faces are detected by proposed modified Viola-Jones algorithm. Then, necessary features are selected using the proposed salp-cat optimization algorithm. Finally, these features are fused with scale-invariant feature transform (SIFT) features and Euclidean distance is computed between feature vectors of sketch and each face in a video. Face from the video having lowest Euclidean distance with query sketch is considered as suspect's face. The proposed method's performance is analyzed on Chokepoint dataset and the system works efficiently with 89.02% of precision, 91.25% of recall and 90.13% of F-measure.

In this article, authors focus on designing an algorithm using Hybrid SVMKNN as component classifier for AdaBoost. To obtain a set of effective SVM-KNN component classifiers, algorithm adaptively adjusts the kernel parameters of KNN using optimized parameters of SVM to form Hybrid SVMKNN as component classifier. In the proposed method, first SVM is applied on training samples with σ = 1000. Initially equal weights are assigned to each tuple and updated weights are obtained. The updated weights from SVM are used as initial weights for KNN. Weighted average of predictions from both the models is obtained. Mammograms are represented by Grey Level Co-occurrence Matrix (GLCM) texture features. High dimensionality of feature space reduces the classification accuracy. Extraction of relevant features to distinguish between benign and malignant class is important to improve the classification performance. Authors use K-means clustering algorithm for features optimization to remove the redund...

Authors presents an effective breast mammogram classification technique by the application of computer assisted tools for augmenting human functions, namely radiologists. Authors present AdaBoost with RBFSVM and Hybrid KNN-RBFSVM as base estimator by adaptively adjusting kernel parameters in RFBSVM (c and C). KNN-RBFSVM classifier is developed by adaptively adjusting the kernel parameters of SVM using optimized parameters of KNN. In the proposed Hybrid KNN-RBFSVM algorithm, first weighted KNN is applied on training mammograms. Initially equal weights are assigned to each mammogram and updated weights are found. The updated weights from KNN are used as initial weights for RBFSVM. This Hybrid KNN-RBFSVM algorithm is used as base estimator in AdaBoost with number of estimators = 200, for prediction of test mammogram as benign or malignant. GLCM features are extracted from mammograms. Inconsistent and irrelevent features of mammogram affect the classification accuracy. Authors use proposed PreARM algorithm for features optimization. The classification results of the AdaBoost with DT, KNN, RBFSVM and Hybrid KNN-RBFSVM as base estimator are compared in terms of accuracy and area under ROC curve value. DDSM mammogram image dataset is used for experiment. The result shows, AdaBoost with Hybrid KNN-RBFSVM as base estimator achieves improved accuracy and AUC value compared to above ensembles.

In this article, authors focus on designing an algorithm using Hybrid SVMKNN as component classifier for AdaBoost. To obtain a set of effective SVM-KNN component classifiers, algorithm adaptively adjusts the kernel parameters of KNN using optimized parameters of SVM to form Hybrid SVMKNN as component classifier. In the proposed method, first SVM is applied on training samples with σ = 1000. Initially equal weights are assigned to each tuple and updated weights are obtained. The updated weights from SVM are used as initial weights for KNN. Weighted average of predictions from both the models is obtained. Mammograms are represented by Grey Level Co-occurrence Matrix (GLCM) texture features. High dimensionality of feature space reduces the classification accuracy. Extraction of relevant features to distinguish between benign and malignant class is important to improve the classification performance. Authors use K-means clustering algorithm for features optimization to remove the redund...

Authors presents concept of image mining, an extension of data mining, for discovering image data relationship from a large collection of mammograms images. Association rule mining is the process of discovering useful and interesting rules from large datasets based on user specified minimum support and confidence values. The set of all possible item sets grows exponentially with the number of items in the database. These constraints lead to exponential search space and dataset dependent minimum support and confidence values. It generates a huge number of unnecessary rules from frequent item sets and results in weak mining performance. The authors propose two association rule optimization techniques for overcoming these problems. The first graph theory approach (OARGT) is based on objective function such that graph generated by the optimized rules is a simple graph with simple walk. The second approach is based on Multi-Objective Genetic algorithm (MOGA) with adaptive crossover and m...

Authors presents an effective breast mammogram classification technique by the application of computer assisted tools for augmenting human functions, namely radiologists. Authors present AdaBoost with RBFSVM and Hybrid KNN-RBFSVM as base estimator by adaptively adjusting kernel parameters in RFBSVM (c and C). KNN-RBFSVM classifier is developed by adaptively adjusting the kernel parameters of SVM using optimized parameters of KNN. In the proposed Hybrid KNN-RBFSVM algorithm, first weighted KNN is applied on training mammograms. Initially equal weights are assigned to each mammogram and updated weights are found. The updated weights from KNN are used as initial weights for RBFSVM. This Hybrid KNN-RBFSVM algorithm is used as base estimator in AdaBoost with number of estimators = 200, for prediction of test mammogram as benign or malignant. GLCM features are extracted from mammograms. Inconsistent and irrelevent features of mammogram affect the classification accuracy. Authors use proposed PreARM algorithm for features optimization. The classification results of the AdaBoost with DT, KNN, RBFSVM and Hybrid KNN-RBFSVM as base estimator are compared in terms of accuracy and area under ROC curve value. DDSM mammogram image dataset is used for experiment. The result shows, AdaBoost with Hybrid KNN-RBFSVM as base estimator achieves improved accuracy and AUC value compared to above ensembles.

Among various types of cancer, breast cancer is considered as second largest hazardous diseases that cause death. Initially, small lump like structure grow from breast cells which are considered as malignant lumps. To find the available of malignancy in breast area, several checkups such as self-test and periodic has to be done to reduce the death rate due to breast cancer. But, the classification of breast cancer by medical physicians using available techniques is not sufficient, so it is important to improve the classification technique using neural network. The four important phases namely preprocessing, segmentation, feature extraction, and classification can be done in constructed Deep Belief Network (DBN) whereas preprocessing makes to remove the noise and artifacts of mammogram image and then the glands are enhanced. The preprocessed output is given to fuzzy c-means segmentation process with the help of masking. Again the feature extractors such as Scale Invariant Feature Tra...

Breast cancer is one of the major health issues across the world. In this study, a new computer-aided detection (CAD) system is introduced. First, the mammogram images were enhanced to increase the contrast. Second, the pectoral muscle was eliminated and the breast was suppressed from the mammogram. Afterward, some statistical features were extracted. Next, k-nearest neighbor (k-NN) and decision trees classifiers were used to classify the normal and abnormal lesions. Moreover, multiple classifier systems (MCS) was constructed as it usually improves the classification results. The MCS has two structures, cascaded and parallel structures. Finally, two wrapper feature selection (FS) approaches were applied to identify those features, which influence classification accuracy. The two data sets (1) the mammographic image analysis society digital mammogram database (MIAS) and (2) the digital mammography dream challenge were combined together to test the CAD system proposed. The highest acc...

In domains such as finance, explainable machine learning approaches are important if they are to back decision-making systems especially in detecting potential bankruptcy where there are a myriad of attributes such as net profit, liquidity ratio that varies with each institution. With the business expansion and increase in company data, to foster explainable machine learning processes and potentially improve performance, there is a need to identify the most important indicators from highly dimensional data that can enhance bankruptcy detection and enable company owners to investigate financial statements with less need for external audit. Hence, this research investigates an optimal selection of features to detect bankruptcy using an ensemble approach combining six feature selection techniques namely Pearson’s correlation, information gain, exhaustive feature selection, gradient boosting trees feature importance, random shuffling, and recursive feature elimination, through different...

Breast cancer is one of the major health issues across the world. In this study, a new computer-aided detection (CAD) system is introduced. First, the mammogram images were enhanced to increase the contrast. Second, the pectoral muscle was eliminated and the breast was suppressed from the mammogram. Afterward, some statistical features were extracted. Next, k-nearest neighbor (k-NN) and decision trees classifiers were used to classify the normal and abnormal lesions. Moreover, multiple classifier systems (MCS) was constructed as it usually improves the classification results. The MCS has two structures, cascaded and parallel structures. Finally, two wrapper feature selection (FS) approaches were applied to identify those features, which influence classification accuracy. The two data sets (1) the mammographic image analysis society digital mammogram database (MIAS) and (2) the digital mammography dream challenge were combined together to test the CAD system proposed. The highest acc...

Over the last few years, web phishing attacks have been constantly evolving causing customers to lose trust in e-commerce and online services. Various tools and systems based on a blacklist of phishing websites are applied to detect the phishing websites. Unfortunately, the fast evolution of technology has led to the born of more sophisticated methods when building websites to attract users. Thus, the latest and newly deployed phishing websites; for example, zero-day phishing websites, cannot be detected by using these blacklist-based approaches. Several recent research studies have been adopting machine learning techniques to identify phishing websites and utilizing them as an early alarm method to identify such threats. However, the important website features have been selected based on human experience or frequency analysis of website features in most of these approaches. In this paper, intelligent phishing website detection using particle swarm optimization-based feature weighting is proposed to enhance the detection of phishing websites. The proposed approach suggests utilizing particle swarm optimization (PSO) to weight various website features effectively to achieve higher accuracy when detecting phishing websites. In particular, the proposed PSO-based website feature weighting is used to differentiate between the various features in websites, based on how important they contribute towards recognizing the phishing from legitimate websites. The experimental results indicated that the proposed PSO-based feature weighting achieved outstanding improvements in terms of classification accuracy, true positive and negative rates, and false positive and negative rates of the machine learning models using only fewer websites features utilized in the detection of phishing websites. INDEX TERMS Feature weighting, machine learning, particle swarm optimization, phishing website.

In this paper, to optimize the process of detecting cyber-attacks, we choose to propose 2 main optimization solutions: Optimizing the detection method and optimizing features. Both of these two optimization solutions are to ensure the aim is to increase accuracy and reduce the time for analysis and detection. Accordingly, for the detection method, we recommend using the Random Forest supervised classification algorithm. The experimental results in section 4.1 have proven that our proposal that use the Random Forest algorithm for abnormal behavior detection is completely correct because the results of this algorithm are much better than some other detection algorithms on all measures. For the feature optimization solution, we propose to use some data dimensional reduction techniques such as information gain, principal component analysis, and correlation coefficient method. The results of the research proposed in our paper have proven that to optimize the cyberattack detection process, it is not necessary to use advanced algorithms with complex and cumbersome computational requirements, it must depend on the monitoring data for selecting the reasonable feature extraction and optimization algorithm as well as the appropriate attack classification and detection algorithms.

All over the world, breast cancer is the second leading cause of death in women above 40 years of age. To design an efficient classification system for breast cancer diagnosis, one has to use efficient algorithms for feature selection to reduce the feature space of mammogram classification. The current work investigates the use of hybrid genetic ensemble method for feature selection and classification of masses. Genetic algorithm (GA) is used to select a subset of features and to evaluate the fitness of the selected features, Adaptive boosting (AdaBoost) and Random Forest (RF) ensembles with 10-fold cross-validation are employed. The selected features are used to classify masses into benign or malignant using AdaBoost, RF, and single Decision Tree (DT) classifiers. The performance evaluation of classifiers indicates that AdaBoost outperforms both RF and single DT classifiers. AdaBoost achieves an accuracy of 96.15%, with 97.32% sensitivity, 95.90% speci-ficity, and area under curve of A Z = 0.982 ± 0.004. The results obtained with the proposed method are better when compared with extant research work.

In this paper presents the recent development and application of image analysis and computer technology system in an automatic fruit recognition system. It is very much essential to through light on basic concepts and technologies associated with computer technology system, a tool used in image analysis of fruit characterization. In India the ever-increasing population, losses in handling and processing and the increased expectation of food products of high quality and safety standards, there is a need for the growth of accurate, fast and objective quality determination of fruits. A number of challenges had to be overcome to enable the system to perform automatic recognition of the kind of fruit or fruit variety using the images from the camera. Many different types of fruits are subject to significant variation in color and texture, depending on how ripe they are. There are many processes in agriculture where decisions are made based on the appearance of the product. Applications for grading the fruit by its quality, size or ripeness are based on its appearance, as well as a decision on whether it is healthy or diseased. The objective of this paper is to provide in depth introduction of machine vision system, its components and recent work reported on an automatic fruit characterization system.

In the world of Automation there is agriculture which is come into play to increase productivity, quality as well as economic growth of the country. Fruit classification is an important process for separating different fruits. For this purpose support vector machine (SVM) and Genetic algorithm (GA) is using to give best result. Here GA and SVM perform separate operation where GA is use for optimization of feature and then SVM is use for classification. Three type of fruit images i.e. apples, banana and grapes are using for classification. 178 fruit images is using for series of experiment. Feature extraction process performed after preprocessing process. In feature extraction process, shape, texture and color of image characteristics extracted. Than above optimization process is use for improving classification process and get better accuracy.