Genetic Feature Selection Research Articles

BOD5 prediction using machine learning methods

Abstract Biological oxygen demand (BOD5) is an indicator used to monitor water quality. However, the standard process of measuring BOD5 is time consuming and could delay crucial mitigation works in the event of pollution. To solve this problem, this study employed multiple machine learning (ML) methods such as random forest (RF), support vector regression (SVR) and multilayer perceptron (MLP) to train a best model that can accurately predict the BOD5 values in water samples based on other physical and chemical properties of the water. The training parameters were optimized using genetic algorithm (GA) and feature selection was made using the sequential feature selection (SFS) method. The proposed machine learning framework was first tested on a public dataset (Waterbase). The MLP method produced the best model, with an R2 score of 0.7672791942775417, relative mean squared error (MSE) and relative mean absolute error (MAE) of approximately 15%. Feature importance calculations indicated that chemical oxygen demand (CODCr), ammonium and nitrate are features that highly correlate to BOD5. In the field study with a small private dataset consisting of water samples collected from two different lakes in Jiangsu Province of China, the trained model was found to have a similar range of prediction error (around 15%), a similar relative MAE (around 14%) and achieved about 6% better relative RMSE.

A comparative study of machine learning methods for bio-oil yield prediction – A genetic algorithm-based features selection

A novel genetic algorithm-based feature selection approach is incorporated and based on these features, four different ML methods were investigated. According to the findings, ML models could reliably predict bio-oil yield. The results showed that Random forest (RF) is preferred for bio-oil yield prediction (R2 ~ 0.98) and highly recommended when dealing with the complex correlation between variables and target. Multi-Linear regression model showed relatively poor generalization performance (R2 ~ 0.75). The partial dependence analysis was done for ML models to show the influence of each input variable on the target variable. Lastly, an easy-to-use software package was developed based on the RF model for the prediction of bio-oil yield. The current study offered new insights into the pyrolysis process of biomass and to improve bio-oil yield. It is an attempt to reduce the time-consuming and expensive experimental work for estimating the bio-oil yield of biomass during pyrolysis.

Singer identification for Indian singers using convolutional neural networks

Singer identification is one of the important aspects of music information retrieval (MIR). In this work, traditional feature-based and trending convolutional neural network (CNN) based approaches are considered and compared for identifying singers. Two different datasets, namely artist20 and the Indian popular singers’ database with 20 singers are used in this work to evaluate proposed approaches. Cepstral features such as Mel-frequency cepstral coefficients (MFCCs) and linear prediction cepstral coefficients (LPCCs) are considered to represent timbre information. Shifted delta cepstral (SDC) features are also computed beside the cepstral coefficients to capture temporal information. In addition, chroma features are computed from 12 semitones of a musical octave, overall forming a 46-dimensional feature vector. Experiments are conducted with different feature combinations, and suitable features are selected using the genetic algorithm-based feature selection (GAFS) approach. Two different classification techniques, namely artificial neural networks (ANNs) and random forest (RF), are considered on the features mentioned above. Further, spectrograms and chromagrams of audio clips are directly fed to CNN for classification. The singer identification results obtained using CNNs seem to be better than the traditional isolated and ensemble classifiers. Average accuracy of around 75% is observed with CNN in the case of Indian popular singers database. Whereas, on artist20 dataset, the proposed configuration of feature-based approach and CNN could not give better than 60% accuracy.

An intelligent approach towards very short-term load forecasting

Abstract Very short term load forecasting (VSTLF) plays a pivotal role in helping the utility workers make proper decisions regarding generation scheduling, size of spinning reserve, and maintaining equilibrium between the power generated by the utility to fulfil the load demand. However, the development of an effective VSTLF model is challenging in gathering noisy real-time data and complicates features found in load demand variations from time to time. A hybrid approach for VSTLF using an incomplete fuzzy decision system (IFDS) combined with a genetic algorithm (GA) based feature selection technique for load forecasting in an hour ahead format is proposed in this research work. This proposed work aims to determine the load features and eliminate redundant features to form a less complex forecasting model. The proposed method considers the time of the day, temperature, humidity, and dew point as inputs and generates output as forecasted load. The input data and historical load data are collected from the Northern Regional Load Dispatch Centre (NRLDC) New Delhi for December 2009, January 2010 and February 2010. For validation of proposed method efficacy, it’s performance is further compared with other conventional AI techniques like ANN and ANFIS, which are integrated with genetic algorithm-based feature selection technique to boost their performance. These techniques’ accuracy is tested through their mean absolute percentage error (MAPE) and normalized root mean square error (nRMSE) value. Compared to other conventional AI techniques and other methods provided through previous studies, the proposed method is found to have acceptable accuracy for 1 h ahead of electrical load forecasting.

Optimization of the C4.5 Algorithm by Using a Genetic Algorithm for the Diagnosis of Life Expectancy for Hepatitis Patients

As technology develops rapidly, the amount of data generated experiencing rapid development, including medical data. Data can help diagnose the life expectancy of people with the disease such as hepatitis using data mining methods in the medical field. In this research, technique data mining uses a classification technique with the C4.5 algorithm and the UCI Machine Learning Repository dataset. This dataset has 19 attributes, 1 class, and 155 samples. C4.5 algorithm is optimized using the Genetic Algorithm feature selection process. This study compares the accuracy of the C4.5 algorithm before and after optimization using a Genetic Algorithm. C4.5 algorithm produces the highest accuracy of 96.23%. Meanwhile, the C4.5 algorithm, after being optimized using Genetic Algorithm, has the highest accuracy of 98.11%. The number of features selected is 15 features. Application of Genetic Algorithms in C4.5 algorithm is proven to improve the accuracy in diagnosing life expectancy of people with hepatitis as much as 1.88%.

A genetic algorithm for prediction of RNA-seq malaria vector gene expression data classification using SVM kernels

Malaria larvae embrace unpredictable variable life periods as they spread across many stratospheres of the mosquito vectors. There are transcriptomes of a thousand distinct species. Ribonucleic acid sequencing (RNA-seq) is a ubiquitous gene expression strategy that contributes to the improvement of genetic survey recognition. RNA-seq measures gene expression transcripts data, including methodological enhancements to machine learning procedures. Scientists have suggested many addressed learning for the study of biological evidence. An enhanced optimized Genetic Algorithm feature selection technique is used in this analysis to obtain relevant information from a high-dimensional Anopheles gambiae dataset and test its classification using SVM-Kernel algorithms. The efficacy of this assay is tested, and the outcome of the experiment obtained an accuracy metric of 93% and 96% respectively.

A hybrid method of genetic algorithm and support vector machine for DNS tunneling detection

With the expansion of the business over the internet, corporations nowadays are investing numerous amounts of money in the web applications. However, there are different threats could make the corporations vulnerable for potential attacks. One of these threats is harnessing the domain name protocol for passing harmful information, this kind of threats is known as DNS tunneling. As a result, confidential information would be exposed and violated. Several studies have investigated the machine learning in order to propose a detection approach. In their approaches, authors have used different and numerous types of features such as domain length, number of bytes, content, volume of DNS traffic, number of hostnames per domain, geographic location and domain history. Apparently, there is a vital demand to accommodate feature selection task in order to identify the best features. This paper proposes a hybrid method of genetic algorithm feature selection approach with the support vector machine classifier for the sake of identifying the best features that have the ability to optimize the detection of DNS tunneling. To evaluate the proposed method, a benchmark dataset of DNS tunneling has been used. Results showed that the proposed method has outperformed the conventional SVM by achieving 0.946 of f-measure

Multiply distorted image quality assessment based on feature level fusion and optimal feature selection

No reference image quality assessment (NR-IQA) has received considerable importance in the last decade due to a rise in the use of multimedia content in our daily lives. Due to limitations in technology, multiple distortions may be introduced in the images that need to be assessed. Recently feature selection has shown promising results for single distorted NR-IQA and their effectiveness on multiple distorted images still need to be addressed. In this paper, impact of feature level fusion and feature selection on multiple distorted image quality assessment is presented. To this end features are extracted from multiple distorted images using six NR-IQA techniques (BLIINDS-II, BRISQUE, CurveletQA, DIIVINE, GM-LOG, SSEQ) that extract features in different (discrete cosine transform, spatial, curvelet transform, wavelet transform, spatial and gradient, spatial and spectral) domains. The extracted features from different domains are fused to generate a single feature vector. All combinations of feature-level fusion from six different techniques have been evaluated. Three different feature selection algorithms (genetic search, linear forward search, particle swarm optimization) are then applied to select optimum features for NR-IQA. The selected features are then used by the support vector regression model to predict the quality score. The performance of the proposed methodology is evaluated for two multiple distorted IQA databases (LIVE multiple distorted image dataset (LIVEMD), multiply distorted image database (MDID2017)), two singly synthetically distorted IQA databases (Tampere image database (TID2013), Computational and subjective image quality database (CSIQ)), and one screen content IQA database (Screen content image quality database (SIQAD)). Experimental results show that the fusion of features from different domains gives better performance in comparison to existing multiple-distorted NR-IQA techniques with SROCC scores of 0.9555, 0.9587, 0.6892, 0.9452, and 0.7682 on the LIVEMD, MDID, TID2013, CSIQ, and SIQAD databases respectively. Moreover, the performance is further improved when the genetic search feature selection algorithm is applied to fused features to remove the redundant and irrelevant features. The SROCC scores are improved to 0.9691, 0.9723, and 0.6897 for LIVEMD, MDID, and TID2013 databases respectively.

Compact Genetic Algorithm-Based Feature Selection for Sequence-Based Prediction of Dengue-Human Protein Interactions.

Dengue Virus (DENV) infection is one of the rapidly spreading mosquito-borne viral infections in humans. Every year, around 50 million people get affected by DENV infection, resulting in 20,000 deaths. Despite the recent experiments focusing on dengue infection to understand its functionality in the human body, several functionally important DENV-human protein-protein interactions (PPIs) have remained unrecognized. This article presents a model for predicting new DENV-human PPIs by combining different sequence-based features of human and dengue proteins like the amino acid composition, dipeptide composition, conjoint triad, pseudo amino acid composition, and pairwise sequence similarity between dengue and human proteins. A Learning vector quantization (LVQ)-based Compact Genetic Algorithm (CGA) model is proposed for feature subset selection. CGA is a probabilistic technique that simulates the behavior of a Genetic Algorithm (GA) with lesser memory and time requirements. Prediction of DENV-human PPIs is performed by the weighted Random Forest (RF) technique as it is found to perform better than other classifiers. We have predicted 1013 PPIs between 335 human proteins and 10 dengue proteins. All predicted interactions are validated by literature filtering, GO-based assessment, and KEGG Pathway enrichment analysis. This study will encourage the identification of potential targets for more effective anti-dengue drug discovery.

A genetic programming-based feature selection and fusion for facial expression recognition

Emotion recognition has become one of the most active research areas in pattern recognition due to the emergence of human–machine interaction systems. Describing facial expression is a very challenging problem since it relies on the quality of the face representation. A multitude of features have been proposed in the literature to describe facial expression. None of these features is universal for accurately capturing all the emotions since facial expressions vary according to the person, gender and type of emotion (posed or spontaneous). Therefore, some research works have considered combining several features to enhance the recognition rate. But they faced significant problems because of information redundancy and high dimensionality of the resulting features. In this work, we propose a genetic programming framework for feature selection and fusion for facial expression recognition, which we called GP−FER. The main component of this framework is a tree-based genetic program with a three functional layers (feature selection, feature fusion and classification). The proposed genetic program is a binary classifier that performs discriminative feature selection and fusion differently for each pair of expression classes. The final emotion is captured by performing a unique tournament elimination between all the classes using the binary programs. Three different geometric and texture features were fused using the proposed GP−FER. The obtained results, on four posed and spontaneous facial expression datasets (DISFA, DISFA+, CK+ and MUG), show that the proposed facial expression recognition method has outperformed, or achieved a comparable performance to the state-of-the-art methods.

An Anomaly Detection Algorithm Selection Service for IoT Stream Data Based on Tsfresh Tool and Genetic Algorithm

Anomaly detection algorithms (ADA) have been widely used as services in many maintenance monitoring platforms. However, there are numerous algorithms that could be applied to these fast changing stream data. Furthermore, in IoT stream data due to its dynamic nature, the phenomena of conception drift happened. Therefore, it is a challenging task to choose a suitable anomaly detection service (ADS) in real time. For accurate online anomalous data detection, this paper developed a service selection method to select and configure ADS at run-time. Initially, a time-series feature extractor (Tsfresh) and a genetic algorithm-based feature selection method are applied to swiftly extract dominant features which act as representation for the stream data patterns. Additionally, stream data and various efficient algorithms are collected as our historical data. A fast classification model based on XGBoost is trained to record stream data features to detect appropriate ADS dynamically at run-time. These methods help to choose suitable service and their respective configuration based on the patterns of stream data. The features used to describe and reflect time-series data’s intrinsic characteristics are the main success factor in our framework. Consequently, experiments are conducted to evaluate the effectiveness of features closed by genetic algorithm. Experimentations on both artificial and real datasets demonstrate that the accuracy of our proposed method outperforms various advanced approaches and can choose appropriate service in different scenarios efficiently.

Improving Human Activity Recognition Performance by Data Fusion and Feature Engineering.

Human activity recognition (HAR) is essential in many health-related fields. A variety of technologies based on different sensors have been developed for HAR. Among them, fusion from heterogeneous wearable sensors has been developed as it is portable, non-interventional and accurate for HAR. To be applied in real-time use with limited resources, the activity recognition system must be compact and reliable. This requirement can be achieved by feature selection (FS). By eliminating irrelevant and redundant features, the system burden is reduced with good classification performance (CP). This manuscript proposes a two-stage genetic algorithm-based feature selection algorithm with a fixed activation number (GFSFAN), which is implemented on the datasets with a variety of time, frequency and time-frequency domain features extracted from the collected raw time series of nine activities of daily living (ADL). Six classifiers are used to evaluate the effects of selected feature subsets from different FS algorithms on HAR performance. The results indicate that GFSFAN can achieve good CP with a small size. A sensor-to-segment coordinate calibration algorithm and lower-limb joint angle estimation algorithm are introduced. Experiments on the effect of the calibration and the introduction of joint angle on HAR shows that both of them can improve the CP.

Biomass retrieval based on genetic algorithm feature selection and support vector regression in Alpine grassland using ground-based hyperspectral and Sentinel-1 SAR data

ABSTRACT A general framework for the integration of multi-sensor data for dry and fresh biomass retrieval is proposed and tested in Alpine meadows and pastures. To this purpose, hyperspectral spectroradiometer (as simulation of hyperspectral imagery) and biomass samples were collected in field campaigns and Copernicus Sentinel-1 Interferometric Wide (IW) swath SAR backscattering coefficients were used. First, a genetic algorithm feature selection was performed on hyperspectral data, and afterwards the resulting most sensitive bands where combined with SAR data within a support vector regression (SVR) model. The most sensitive hyperspectral bands were mainly located in different regions of the SWIR range for both fresh and dry biomass, and in the red and near-infrared regions mainly for dry biomass, but with less influence for fresh biomass. The R 2 correlation values between the sampled and the estimated biomass range from 0.24 to 0.71. The relatively low performances are mainly related to the saturation effect in the optical bands, as well as to the paucity of points for high values of biomass. The methodology allows a better understanding of the interaction between grassland systems and the electromagnetic spectrum by offering a model with a reduced number of narrow bands in the context of a multi-sensor integration.

Singer Gender Classification using Feature-based and Spectrograms with Deep Convolutional Neural Network

The task of music information retrieval (MIR) is gaining much importance since the digital cloud is growing sparklingly. An important attribute of MIR is the singer-id, which helps effectively during the recommendation process. It is highly difficult to identify a singer in the case of music as the number of signers available in the digital cloud is high. The process of identifying the gender of a singer may simplify the task of singer identification and also helps with the recommendation. Hence, an effort has been made to detect the gender information of a singer. Two different datasets have been considered. Of which, one is collected from Indian cine industries having 20 different singer details of four regional languages. The other dataset is standard Artist20. Various spectral, temporal, and pitch related features have been used to obtain better accuracy. The features considered for this task are Mel-frequency cepstral coefficients (MFCCs), pitch, velocity, and acceleration of MFCCs. The experimentation has been done on various combinations of the mentioned features with the support of artificial neural networks (ANNs) and random forest (RF). Further, the genetic algorithm-based feature selection (GAFS) has been used to select the suitable features out of the best combination obtained. Moreover, we have also utilized the recent popular convolutional neural networks (CNNs) with the support of spectrograms to obtain better accuracy over the traditional feature vector. Average accuracy of 91.70% is obtained for both the Indian and Western clips, which is an improved accuracy of 3% over hand engineering features.

PM2.5 Prediction Using Genetic Algorithm-Based Feature Selection and Encoder-Decoder Model

The concentration of fine particulate matter (PM2.5), which represents inhalable particles with diameters of 2.5 micrometers and smaller, is a vital air quality index. Such particles can penetrate deep into the human lungs and severely affect human health. This paper studies accurate PM2.5 prediction, which can potentially contribute to reducing or avoiding the negative consequences. Our approach’s novelty is to utilize the genetic algorithm (GA) and an encoder-decoder (E-D) model for PM2.5 prediction. The GA benefits feature selection and remove outliers to enhance the prediction accuracy. The encoder-decoder model with long short-term memory (LSTM), which relaxes the restrictions between the input and output of the model, can be used to effectively predict the PM2.5 concentration. We evaluate the proposed model on air quality datasets from Hanoi and Taiwan. The evaluation results show that our model achieves excellent performance. By merely using the E-D model, we can obtain more accurate (up to 53.7%) predictions than those of previous works. Moreover, the GA in our model has the advantage of obtaining the optimal feature combination for predicting the PM2.5 concentration. By combining the GA-based feature selection algorithm and the E-D model, our proposed approach further improves the accuracy by at least 13.7%.

Stock Prediction Based on Genetic Algorithm Feature Selection and Long Short-Term Memory Neural Network

In the financial market, there are a large number of indicators used to describe the change of stock price, which provides a good data basis for our stock price forecast. Different stocks are affected by different factors due to their different industry types and regions. Therefore, it is very important to find a multi factor combination suitable for a particular stock to predict the price of the stock. This paper proposes to use Genetic Algorithm(GA) for feature selection and develop an optimized Long Short-Term Memory(LSTM) neural network stock prediction model. Firstly, we use the GA to obtain a factors importance ranking. Then, the optimal combination of factors is obtained from this ranking with the method of trial and error. Finally, we use the combination of optimal factors and LSTM model for stock prediction. Thorough empirical studies based upon the China construction bank dataset and the CSI 300 stock dataset demonstrate that the GA-LSTM model can outperform all baseline models for time series prediction.

Sex classification of first molar teeth in cone beam computed tomography images using data mining

ObjectiveThe teeth have been used as a supplementary tool for sex differentiation as they are resistant to post-mortem degradation. The present study aimed to develop a new novel informatics framework for predicting sex from linear tooth dimension measurements achieved from cone beam computed tomography (CBCT) images. Method and materialsA clinical workflow using different machine learning methods was employed to predict the sex in the present study. The CBCT images of 485 subjects (245 men and 240 women) were evaluated for sex differentiation. Nine parameters were measured in both buccolingual and mesiodistal aspects of the teeth. We applied our dataset to Naïve Bayesian (NB), Random Forest (RF), and Support Vector Machine (SVM) as classifiers for prediction. Genetic feature selection was used to discover real features associated with sex classification. ResultsThe 10-fold cross-validation results indicated that NB had higher accuracy than SVM and RF for sex classification. The genetic algorithm (GA) indicated that the model could fit the data without using the enamel thickness and pulp height. The average classification accuracy of our clinical workflow was 92.31 %. ConclusionThe results showed that NB was the best method for sex classification. The application of the first molar teeth in sex prediction indicated an acceptable level of sexual classification. Therefore, these odontometric parameters can be applied as an additional tool for sex determination in forensic anthropology.

Feature selection with multi-objective genetic algorithm based on a hybrid filter and the symmetrical complementary coefficient

With the expansion of data size and data dimension, feature selection attracts more and more attention. In this paper, we propose a novel feature selection algorithm, namely, Hybrid filter and Symmetrical Complementary Coefficient based Multi-Objective Genetic Algorithm feature selection (HSMOGA). HSMOGA contains a new hybrid filter, Symmetrical Complementary Coefficient which is a well-performed metric of feature interactions proposed recently, and a novel way to limit feature subset’s size. A new Pareto-based ranking function is proposed when solving multi-objective problems. Besides, HSMOGA starts with a novel step called knowledge reserve, which precalculate the knowledge required for fitness function calculation and initial population generation. In this way, HSMOGA is classifier-independent in each generation, and its initial population generation makes full use of the knowledge of data set which makes solutions converge faster. Compared with other GA-based feature selection methods, HSMOGA has a much lower time complexity. According to experimental results, HSMOGA outperforms other nine state-of-art feature selection algorithms including five classic and four more recent algorithms in terms of kappa coefficient, accuracy, and G-mean for the data sets tested.

Machine learning guided prediction of liquid chromatography–mass spectrometry ionization efficiency for genotoxic impurities in pharmaceutical products

The limitation and control of genotoxic impurities (GTIs) has continued to receive attention from pharmaceutical companies and authorities for several decades. Because GTIs have the ability to damage deoxyribonucleic acid (DNA) and the potential to cause cancer, low-level quantitation is required to protect patients. A quick and easy method of determining the liquid chromatography-mass spectrometry (LC/MS) conditions for high-sensitivity analysis of GTIs may prospectively accelerate pharmaceutical development. In this study, a quantitative structure-property relationship (QSPR) model was developed for predicting the ionization efficiency of compounds using liquid-chromatography-mass spectrometry (LC/MS) parameters and molecular descriptors. Before implementing the QSPR prediction model, linear regression analysis was performed to model the relationship between the ionization efficiency and the LC/MS parameters for each compound. Comparison of the predicted peak areas with the experimentally observed peak areas showed good agreement based on the coefficient of determination (R2 > 0.96). The machine learning-based QSPR approach begins with computation of the molecular descriptors expressing the physicochemical properties of a compound, followed by a genetic algorithm-based feature selection. Linear and nonlinear regression were performed, and support vector machine (SVM) was selected as the best machine learning algorithm for the prediction. The SVM algorithm was developed and optimized using 1031 experimental data points for nine compounds, including well-known GTIs. Validation of the model by comparison of the predicted and observed relative ionization efficiencies (RIE) showed a high coefficient of determination (R2 = 0.96) and low root mean squared error value (RMSE = 0.118). Finally, this established prediction model was applied to hydrophilic interaction liquid chromatography coupled with MS for a new compound in new mobile phase compositions and new MS parameter settings. The RMSE of the predicted versus observed RIE was 0.203. This prediction accuracy was sufficient to determine the starting point of the LC/MS method development. The methodology demonstrated in this study can be used to determine the LC/MS conditions for high sensitivity analysis of GTIs.

Genetic Algorithm-based Feature Selection Approach for Enhancing the Effectiveness of Similarity Searching in Ligand-based Virtual Screening

Background: In the last years, similarity searching has gained wide popularity as a method for performing Ligand-Based Virtual Screening (LBVS). This screening technique functions by making a comparison of the target compound’s features with that of each compound in the database of compounds. It is well known that none of the individual similarity measures could provide the best performances each time pertaining to an active compound structure, representing all types of activity classes. In the literature, we find several techniques and strategies that have been proposed to improve the overall effectiveness of ligand-based virtual screening approaches. Objective: In this work, our main objective is to propose a features selection approach based on genetic algorithm (FSGASS) to improve similarity searching pertaining to ligand-based virtual screening. Methods: Our contribution allows us to identify the most important and relevant characteristics of chemical compounds and to minimize their number in their representations. This will allow the reduction of features space, the elimination of redundancy, the reduction of training execution time, and the increase of the performance of the screening process. Results: The obtained results demonstrate superiority in the performance compared with these obtained with Tanimoto coefficient, which is considered as the most widely coefficient to quantify the similarity in the domain of LBVS. Conclusion: Our results show that significant improvements can be obtained by using molecular similarity research methods at the basis of features selection.