Wrapper-based Feature Selection Method Research Articles

CBDA: Chaos-based binary dragonfly algorithm for evolutionary feature selection

The goal of feature selection in machine learning is to simultaneously maintain more classification accuracy, while reducing lager amount of attributes. In this paper, we firstly design a fitness function that achieves both objectives jointly. Then we come up with a chaos-based binary dragonfly algorithm (CBDA) that incorporates several improvements over the conventional dragonfly algorithm (DA) for developing a wrapper-based feature selection method to solve the fitness function. Specifically, the CBDA innovatively introduces three improved factors, namely the chaotic map, evolutionary population dynamics (EPD) mechanism, and binarization strategy on the basis of conventional DA to balance the exploitation and exploration capabilities of the algorithm and make it more suitable to handle the formulated problem. We conduct experiments on 24 well-known data sets from the UCI repository with three ablated versions of CBDA targeting different components of the algorithm in order to explain their contributions in CBDA and also with five established comparative algorithms in terms of fitness value, classification accuracy, CPU running time, and number of selected features. The results show that the proposed CBDA has remarkable advantages in most of the tested data sets.

A novel multi-objective wrapper-based feature selection method using quantum-inspired and swarm intelligence techniques

A novel multi-objective wrapper-based feature selection method using quantum-inspired and swarm intelligence techniques

Machine learning and deep learning techniques for detecting malicious android applications: An empirical analysis

The open system architecture of android makes it vulnerable to a variety of cyberattacks. Cybercriminals use android applications to intrude into the system and steal confidential data. This situation poses a threat to user privacy and integrity of the system. This paper proposes a static analysis approach to detect malicious and benign Android applications using various machine learning and deep learning algorithms. The proposed work has been validated using a bench marked dataset comprising 11,449 benign and malicious Android applications. The proposed approach applies a wrapper-based feature selection method to filter irrelevant features. The results clearly show that the deep learning algorithms of DBN and MLP outperformed machine learning algorithms in detecting malicious Android applications.

Wrapper-based deep feature optimization for activity recognition in the wearable sensor networks of healthcare systems

The Human Activity Recognition (HAR) problem leverages pattern recognition to classify physical human activities as they are captured by several sensor modalities. Remote monitoring of an individual’s activities has gained importance due to the reduction in travel and physical activities during the pandemic. Research on HAR enables one person to either remotely monitor or recognize another person’s activity via the ubiquitous mobile device or by using sensor-based Internet of Things (IoT). Our proposed work focuses on the accurate classification of daily human activities from both accelerometer and gyroscope sensor data after converting into spectrogram images. The feature extraction process follows by leveraging the pre-trained weights of two popular and efficient transfer learning convolutional neural network models. Finally, a wrapper-based feature selection method has been employed for selecting the optimal feature subset that both reduces the training time and improves the final classification performance. The proposed HAR model has been tested on the three benchmark datasets namely, HARTH, KU-HAR and HuGaDB and has achieved 88.89%, 97.97% and 93.82% respectively on these datasets. It is to be noted that the proposed HAR model achieves an improvement of about 21%, 20% and 6% in the overall classification accuracies while utilizing only 52%, 45% and 60% of the original feature set for HuGaDB, KU-HAR and HARTH datasets respectively. This proves the effectiveness of our proposed wrapper-based feature selection HAR methodology.

A binary Bi-phase mutation-based hybrid Equilibrium Optimizer for feature selection in medical datasets classification

With the rapid expansion in Biological Sciences, biomedical data classification has become challenging. These datasets generally consist of missing values, redundant features and irrelevant information. A novel hybrid wrapper-based feature selection method is proposed to tackle these issues effectively. In order to improve the exploration ability of the particles, the Sine factor is integrated with the Equilibrium Optimizer (EO) technique. A Bi-phase Mutation (BM) scheme is integrated to enhance the exploitation phase of the EO algorithm (BM-based Hybrid EO, BMHEO). The BMHEO method is evaluated by employing four different classifiers – KNN, SVM, Random Forest (RF) and Discriminant Analysis (DA). It is observed that the Random Forest classifier exhibits superior performance compared to the other three classifiers. Eight S-shaped and V-shaped transfer functions are integrated to convert the solutions to binary form. Based on the above enhancements, eight different versions of BMHEO are produced. The performance of these versions is assessed using twenty biomedical datasets. Experiments on these datasets demonstrate that the BMHEO-S2 version outperforms other methods in terms of fitness values and classification accuracies. This approach is also tested on a MED-NODE image dataset for classification, and higher accuracy of 97.02% is achieved.

Identification of offensive language in Urdu using semantic and embedding models.

Automatic identification of offensive/abusive language is very necessary to get rid of unwanted behavior. However, it is more challenging to generalize the solution due to the different grammatical structures and vocabulary of each language. Most of the prior work targeted western languages, however, one study targeted a low-resource language (Urdu). The prior study used basic linguistic features and a small dataset. This study designed a new dataset (collected from popular Pakistani Facebook pages) containing 7,500 posts for offensive language detection in Urdu. The proposed methodology used four types of feature engineering models: three are frequency-based and the fourth one is the embedding model. Frequency-based are either determined by the term frequency-inverse document frequency (TF-IDF) or bag-of-words or word n-gram feature vectors. The fourth is generated by the word2vec model, trained on the Urdu embeddings using a corpus of 196,226 Facebook posts. The experiments demonstrate that the stacking-based ensemble model with word2vec shows the best performance as a standalone model by achieving 88.27% accuracy. In addition, the wrapper-based feature selection method further improves performance. The hybrid combination of TF-IDF, bag-of-words, and word2vec feature models achieved 90% accuracy and 97% AUC. In addition, it outperformed the baseline with an improvement of 3.55% in accuracy, 3.68% in the recall, 3.60% in f1-measure, 3.67% in precision, and 2.71% in AUC. The findings of this research provide practical implications for commercial applications and future research.

The alterations of brain functional connectivity networks in major depressive disorder detected by machine learning through multisite rs-fMRI data

BackgroundThe current diagnosis of major depressive disorder (MDD) is mainly based on the patient's self-report and clinical symptoms. Machine learning methods are used to identify MDD using resting-state functional magnetic resonance imaging (rs-fMRI) data. However, due to large site differences in multisite rs-fMRI data and the difficulty of sample collection, most of the current machine learning studies use small sample sizes of rs-fMRI datasets to detect the alterations of functional connectivity (FC) or network attribute (NA), which may affect the reliability of the experimental results. MethodsMultisite rs-fMRI data were used to increase the size of the sample, and then we extracted the functional connectivity (FC) and network attribute (NA) features from 1611 rs-fMRI data (832 patients with MDD (MDDs) and 779 healthy controls (HCs)). ComBat algorithm was used to harmonize the data variances caused by the multisite effect, and multivariate linear regression was used to remove age and sex covariates. Two-sample t-test and wrapper-based feature selection methods (support vector machine recursive feature elimination with cross-validation (SVM-RFECV) and LightGBM's "feature_importances_" function) were used to select important features. The Shapley additive explanations (SHAP) method was used to assign the contribution of features to the best classification effect model. ResultsThe best result was obtained from the LinearSVM model trained with the 136 important features selected by SVMRFE-CV. In the nested five-fold cross-validation (consisting of an outer and an inner loop of five-fold cross-validation) of 1611 data, the model achieved the accuracy, sensitivity, and specificity of 68.90 %, 71.75 %, and 65.84 %, respectively. The 136 important features were tested in a small dataset and obtained excellent classification results after balancing the ratio between patients with depression and HCs. ConclusionsThe combined use of FC and NA features is effective for classifying MDDs and HCs. The important FC and NA features extracted from the large sample dataset have some generalization performance and may be used as a reference for the altered brain functional connectivity networks in MDD.

Abstract 6322: Dynamic, stage-course protein interaction network using high power CpG sites in head and neck squamous cell carcinoma

Abstract Head and neck cancer is the most incident and prevalent cancer in men in South Asian countries, especially Pakistan. Prediction of pathological stages of cancer can play a critical role in early diagnosis and personalized medicine. This project ventures into the prediction of different stages of head and neck squamous cell carcinoma (HNSCC) using prioritized DNA methylation patterns. DNA methylation profiles for each HNSCC stage (stage-I-IV) were used to extensively analyze 485,577 methylation CpG sites and prioritize them on the basis of the highest predictive power using a wrapper-based feature selection method, along with different classification models. We identified 68 high-power methylation sites which predicted the pathological stage of HNSCC samples with 90.62% accuracy using a Random Forest classifier. We set out to construct a protein-protein interaction network for the proteins encoded by the 67 genes associated with these sites to study its network topology and also undertook enrichment analysis of nodes in their immediate neighborhood for GO and KEGG Pathway annotations revealing their role in cancer-related pathways, cell differentiation, signal transduction, metabolic and biosynthetic processes. With information on the predictive power of each of the 67 genes in each HNSCC stage, we unveil a dynamic stage-course network for HNSCC that evolves over the course of the four cancer stages. We intend to further study these genes in light of functional datasets from CRISPR, RNAi, drug screens for their putative role in HNSCC initiation and/or progression. Citation Format: Arsalan Riaz, Faisal Khan. Dynamic, stage-course protein interaction network using high power CpG sites in head and neck squamous cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 6322.

An improved group teaching optimization algorithm based on local search and chaotic map for feature selection in high-dimensional data

The current study proposes a novel binary group teaching optimization algorithm with local search and chaos mapping (BGTOALC) as a wrapper-based feature selection method to solve high-dimensional feature selection problems. The local search and chaos mapping enhance the performance of the proposed algorithm. Also, two novel binary operators called Binary Teacher Phase Good Group (BTPGG) and Binary Teacher Phase Bad Group (BTPBG) are applied to the teacher’s phase for increasing the exploration and exploitation of the algorithm. Moreover, a new Binary Student Opposition-Based Learning (BSOBL) operator is introduced for the student phase, using an opposition-based strategy to achieve better exploitation. Finally, the teacher allocation phase is designed in a binary manner using the new Mean Binary Select (MBS) operator to increase the algorithm’s convergence rate. Subsequently, two other binary group teaching optimization algorithms, named BGTOAV and BGTOAS, are developed utilizing the S-shaped and V-shaped transfer functions to compare their performance with the BGTOALC algorithm. The proposed approaches are compared to other state-of-the-art binary algorithms on 30 datasets with different dimensions. Different experiments prove that the BGTOALC method outperforms the previous methods in terms of reducing the number of selected features and increasing the accuracy of the machine learning algorithm. Eventually, statistical analyses indicate the superiority of the BGTOALC method in terms of efficiency and convergence rate against other binary metaheuristic algorithms.

Binary Golden Eagle Optimizer with Time-Varying Flight Length for feature selection

The concept of any method to resolve feature selection issues is to identify a subset of the original features. However, determining a minimal feature subset is considered an NP-hard problem. Many existing feature selection methods, namely particle swarm optimization (PSO), grey wolf optimization (GWO) etc., are complex and consume more time. This article proposes a new wrapper-based method that employs Binary Golden Eagle Optimizer with Time Varying Flight Length (BGEO-TVFL) to reduce the feature selection issues. The proposed method provides optimal feature selection based on Golden Eagle Optimizer (GEO). GEO is a recently presented swarm-based meta-heuristics algorithm that imitates the smartness of Golden Eagles (GEs). GEO algorithm performs on continuous space; however, the feature selection is in discrete space. Hence, the continuous space has transformed into a discrete space by employing transfer functions. Here, eight different transfer functions have been applied to determine the best transfer function and test the BGEO-TVFL. Moreover, a time-varying flight length (TVFL) has been employed to balance the exploration and exploitation in GEO. Finally, the performance of the proposed BGEO-TVFL has been determined by focusing on various metrics under different UC Irvine (UCI) datasets. The best variant has been selected and analyzed the performance with existing standard filter feature selection approaches and wrapper based feature selection methods. The existing methods include Bat Algorithm (BAT), Ant Colony Optimization (ACO) algorithm, PSO, GWO, Genetic Algorithm (GA), Cuckoo Search (CS), Golden Eagle Optimizer (GEO), ReliefF, Information Gain (IG), Correlation-based feature selection (CFS), and Gain Ratio (GR) respectively. In addition, the performance of BGEO-TVFL is also compared and analyzed with state-of-art methods to determine its superiority. Furthermore, the proposed BGEO-TVFL is analyzed with existing algorithms by employing 30 functions from CEC’2017, CEC’2018 and CEC’2020 benchmarks. As a result, the proposed BGEO-TVFL has achieved a better result than existing methods and is suitable for dealing with dimensionality reduction issues.

Investigation of a composite two-phase hedging rule policy for a multi reservoir system using streamflow forecast

Long-term changes in reservoir inflow due to climate changes and human interference violate the assumptions of hydrologic stationarity especially in the reservoir design. Utilization of uncertain prediction into a reservoir operating rule curves somehow reflects the challenges that imposed by nonstationary conditions. This study proposes a hedging based policy incorporated forecast term to manage release decisions in two separate phases. Hedging is applied firstly regarding to reservoir water level similar to conventional hedging rules and secondary according to an extra simulation in the near future. To determine the time interval of future effects, an exterior optimization model is introduced to handle the trade-off between forecast uncertainty and future information which imposed by forecast horizon. Future inflows are forecasted introducing a model including a wrapper-based feature selection method and AdaBoost.RT as a learning algorithm. The results of applying the model to a real six reservoir system in IRAN showed that incorporating future inflows into the real time decisions significantly improves the total squared relative deficit about 20% and 10% compared to conventional hedging rule curve (CHRC) and standard operation policy as objective function. Also having a glance at the near future reduces the vulnerability of the system about 5% and 27% respectively against CHRC and SOP. The results also showed that, although the SOP reaches to a best reliability of satisfying water demands in total system as 31% and 27% better than CHRC and the proposed two-phase policy, but the number of intensified failures was higher than two others which somehow influences on volume-based indices like vulnerability.

Diabetes Prediction Algorithm Using Recursive Ridge Regression L2

At present, the prevalence of diabetes is increasing because the human body cannot metabolize the glucose level. Accurate prediction of diabetes patients is an important research area. Many researchers have proposed techniques to predict this disease through data mining and machine learning methods. In prediction, feature selection is a key concept in preprocessing. Thus, the features that are relevant to the disease are used for prediction. This condition improves the prediction accuracy. Selecting the right features in the whole feature set is a complicated process, and many researchers are concentrating on it to produce a predictive model with high accuracy. In this work, a wrapper-based feature selection method called recursive feature elimination is combined with ridge regression (L2) to form a hybrid L2 regulated feature selection algorithm for overcoming the overfitting problem of data set. Overfitting is a major problem in feature selection, where the new data are unfit to the model because the training data are small. Ridge regression is mainly used to overcome the overfitting problem. The features are selected by using the proposed feature selection method, and random forest classifier is used to classify the data on the basis of the selected features. This work uses the Pima Indians Diabetes data set, and the evaluated results are compared with the existing algorithms to prove the accuracy of the proposed algorithm. The accuracy of the proposed algorithm in predicting diabetes is 100%, and its area under the curve is 97%. The proposed algorithm outperforms existing algorithms.

Software Defect Prediction Using Wrapper Feature Selection Based on Dynamic Re-Ranking Strategy

Finding defects early in a software system is a crucial task, as it creates adequate time for fixing such defects using available resources. Strategies such as symmetric testing have proven useful; however, its inability in differentiating incorrect implementations from correct ones is a drawback. Software defect prediction (SDP) is another feasible method that can be used for detecting defects early. Additionally, high dimensionality, a data quality problem, has a detrimental effect on the predictive capability of SDP models. Feature selection (FS) has been used as a feasible solution for solving the high dimensionality issue in SDP. According to current literature, the two basic forms of FS approaches are filter-based feature selection (FFS) and wrapper-based feature selection (WFS). Between the two, WFS approaches have been deemed to be superior. However, WFS methods have a high computational cost due to the unknown number of executions available for feature subset search, evaluation, and selection. This characteristic of WFS often leads to overfitting of classifier models due to its easy trapping in local maxima. The trapping of the WFS subset evaluator in local maxima can be overcome by using an effective search method in the evaluator process. Hence, this study proposes an enhanced WFS method that dynamically and iteratively selects features. The proposed enhanced WFS (EWFS) method is based on incrementally selecting features while considering previously selected features in its search space. The novelty of EWFS is based on the enhancement of the subset evaluation process of WFS methods by deploying a dynamic re-ranking strategy that iteratively selects germane features with a low subset evaluation cycle while not compromising the prediction performance of the ensuing model. For evaluation, EWFS was deployed with Decision Tree (DT) and Naïve Bayes classifiers on software defect datasets with varying granularities. The experimental findings revealed that EWFS outperformed existing metaheuristics and sequential search-based WFS approaches established in this work. Additionally, EWFS selected fewer features with less computational time as compared with existing metaheuristics and sequential search-based WFS methods.

Machine Learning Prediction of Treatment Outcome in Late-Life Depression.

Background: Recent evidence suggests that integration of multi-modal data improves performance in machine learning prediction of depression treatment outcomes. Here, we compared the predictive performance of three machine learning classifiers using differing combinations of sociodemographic characteristics, baseline clinical self-reports, cognitive tests, and structural magnetic resonance imaging (MRI) features to predict treatment outcomes in late-life depression (LLD).Methods: Data were combined from two clinical trials conducted with depressed adults aged 60 and older, including response to escitalopram (N = 32, NCT01902004) and Tai Chi (N = 35, NCT02460666). Remission was defined as a score of 6 or less on the 24-item Hamilton Rating Scale for Depression (HAMD) at the end of 24 weeks of treatment. Features subsets were constructed from baseline sociodemographic and clinical features, gray matter volumes (GMVs), or both. Three classification algorithms were compared: (1) Support Vector Machine-Radial Bias Function (SVMRBF), (2) Random Forest (RF), and (3) Logistic Regression (LR). A repeated 5-fold cross-validation approach with a wrapper-based feature selection method was used for model fitting. Model performance metrics included Area under the ROC Curve (AUC) and Matthews correlation coefficient (MCC). Cross-validated performance significance was tested by permutation analysis. Classifiers were compared by Cochran's Q and post-hoc pairwise comparisons using McNemar's Chi-Square test with Bonferroni correction.Results: For the RF and SVMRBF algorithms, the combined feature set outperformed the clinical and GMV feature sets with a final cross-validated AUC of 0.83 ± 0.11 and 0.80 ± 0.11, respectively. Both classifiers passed permutation analysis. The LR algorithm performed best using GMV features alone (AUC 0.79 ± 0.14) but failed to pass permutation analysis using any feature set. Performance of the three classifiers differed significantly for all three features sets. Important predictive features of treatment response included anterior and posterior cingulate volumes, depression characteristics, and self-reported health-related quality scores.Conclusion: This preliminary exploration into the use of ML and multi-modal data to identify predictors of general treatment response in LLD indicates that integration of clinical and structural MRI features significantly increases predictive capability. Identified features are among those previously implicated in geriatric depression, encouraging future work in this arena.

BREAST CANCER DIAGNOSIS USING WRAPPER-BASED FEATURE SELECTION AND ARTIFICIAL NEURAL NETWORK

Breast cancer is commonest type of cancers among women. Early diagnosis plays a significant role in reducing the fatality rate. The main objective of this study is to propose an efficient approach to classify breast cancer tumor into either benign or malignant based on digitized image of a fine needle aspirate (FNA) of a breast mass represented by the Wisconsin Breast Cancer Dataset. Two wrapper-based feature selection methods, namely, sequential forward selection(SFS) and sequential backward selection (SBS) are used to identify the most discriminant features which can contribute to improve the classification performance. The feed forward neural network (FFNN) is used as a classification algorithm. The learning algorithm hyper-parameters are optimized using the grid search process. After selecting the optimal classification model, the data is divided into training set and testing set and the performance was evaluated. The feature space is reduced from nine feature to seven and six features using SFS and SBS respectively. The highest classification accuracy recorded was 99.03% with FFNN using the seven SFS selected features. While accuracy recorded with the six SBS selected features was 98.54%. The obtained results indicate that the proposed approach is effective in terms of feature space reduction leading to better accuracy and efficient classification model.

Student Performance Prediction Using A Cascaded Bi-level Feature Selection Approach

Features in educational data are ambiguous which leads to noisy features and curse of dimensionality problems. These problems are solved via feature selection. There are existing models for features selection. These models were created using either a single-level embedded, wrapperbased or filter-based methods. However single-level filter-based methods ignore feature dependencies and ignore the interaction with the classifier. The embedded and wrapper based feature selection methods interact with the classifier, but they can only select the optimal subset for a particular classifier. So their selected features may be worse for other classifiers. Hence this research proposes a robust Cascade Bi-Level (CBL) feature selection technique for student performance prediction that will minimize the limitations of using a single-level technique. The proposed CBL feature selection technique consists of the Relief technique at first-level and the Particle Swarm Optimization (PSO) at the second-level. The proposed technique was evaluated using the UCI student performance dataset. In comparison with the performance of the single-level feature selection technique the proposed technique achieved an accuracy of 94.94% which was better than the values achieved by the single-level PSO with an accuracy of 93.67% for the binary classification task. These results show that CBL can effectively predict student performance.

Image processing-based automatic detection of asphalt pavement rutting using a novel metaheuristic optimized machine learning approach

Pavement rutting refers to surface depression in the wheel-path along an asphalt road which causes loss of steering control and consequently leads to serious traffic accidents. Hence, it is necessary to develop powerful methods to accurately recognize pavement rutting during road condition survey. This study presents a novel computer vision-based model to automatically identify rutting on asphalt pavement road. The model is established based on a hybridization of image processing techniques (ITPs), least squares support vector classification (LSSVC), dynamic feature selection (FS) method, and forensic-based investigation (FBI). The ITPs, including Gabor filter and discrete cosine transform were employed to implement texture computation for image data. These techniques are used to generate an initial set of extracted features describing rutting and non-rutting states. The extracted features were then refined by a wrapper-based feature selection (FS) method to determine set of highly relevant features. LSSVC models were used to learn the categorization of rutting and non-rutting based on the refined features and hyper-parameters optimized by the FBI metaheuristic. The final LSSVC prediction model with the most desired prediction accuracy can be obtained once the process of the FBI’s optimization terminates. A dataset of 2000 image samples has been collected during field trip of pavement survey in Da Nang city (Vietnam) to construct and evaluate the newly developed model. The statistical results obtained from a k-fold cross-validation have demonstrated that the hybrid FBI-LSSVC-FS model can achieve the most desired rutting recognition performance with accuracy rate, precision, recall, and F1 score of 98.9%, 0.994, 0.984 and 0.989, respectively. Therefore, this paper contributes to the body of knowledge by proposing a novel AI-based prediction model to assist transportation agencies in the task of periodic asphalt pavement survey.

Comparison of Filter and Wrapper Based Feature Selection Methods on Spam Comment Classification

The continuous growth of the internet has led to the use of social media for various purposes increase. For instance, some irresponsible parties take advantage of the comment feature on social media platforms to harm others by providing spam comments on the shared object. Furthermore, variation of comments creates many features to be processed, thereby negatively impacting the performance of a classification algorithm. Therefore, this study aims to solve the problem associated with spam comments by comparing filter and wrapper based feature selection using text classification techniques. Data collected from training and test data of 4944 and 100 comments showed that the best accuracy, precision, recall, and f-measure of MNB are 96%, 100%, 92%, and 95.8%. The best accuracy is achieved using feature selection by combining Chi-Square and Sequential Forward Selection methods with a subset of 500 features. Furthermore, the accuracy increase in the MNB and SVM classifications are 8% and 4%. This research concludes that the combination of feature selection improves the classification performance of Indonesian language spam comments.

A Statistical Comparison of Feature Selection Techniques for Solar Energy Forecasting Based on Geographical Data

In recent years, solar energy forecasting has been increasingly embraced as a sustainable low-energy solution to environmental awareness. It is a subject of interest to the scientific community, and machine learning techniques have proven to be a powerful means to construct an automatic learning model for an accurate prediction. Along with the various machine learning and data mining utilities applied to solar energy prediction, the process of feature selection is becoming an ultimate requirement for improving model building efficiency. In this paper, we consider the feature selection (FS) approach potential. We provide a detailed taxonomy of various feature selection techniques and examine their usability and ability to deal with a solar energy forecasting problem, given meteorological and geographical data. We focus on filter-based, wrapper-based, and embedded-based feature selection methods. We use the reduced number of selected features, stability, and regression accuracy and compare feature selection techniques. Moreover, the experimental results demonstrate how the feature selection methods studied can considerably improve the prediction process and how the selected features vary by method, depending on the given data constraints.

Enhancing sentiment classification performance using hybrid Query Expansion Ranking and Binary Particle Swarm Optimization with Adaptive Inertia Weights

Machine learning-based sentiment classification is the best-performing method to understand public sentiment. However, the method has some problems, such as noisy features and high-dimensional feature space which affect the sentiment classification performance. To address the problems, this paper proposes a new feature selection using hybrid Query Expansion Ranking and Binary Particle Swarm Optimization with Adaptive Inertia Weights. The proposed method was validated using five tweet datasets on different topics both in Indonesian and English, and compared with state-of-the-art of filter and wrapper-based feature selection methods. Experimental results show the proposed method significantly improves sentiment classification performance and decrease computational time.