Employing Feature Selection Research Articles

RadWise: A Rank-Based Hybrid Feature Weighting and Selection Method for Proteomic Categorization of Chemoirradiation in Patients with Glioblastoma.

Glioblastomas (GBM) are rapidly growing, aggressive, nearly uniformly fatal, and the most common primary type of brain cancer. They exhibit significant heterogeneity and resistance to treatment, limiting the ability to analyze dynamic biological behavior that drives response and resistance, which are central to advancing outcomes in glioblastoma. Analysis of the proteome aimed at signal change over time provides a potential opportunity for non-invasive classification and examination of the response to treatment by identifying protein biomarkers associated with interventions. However, data acquired using large proteomic panels must be more intuitively interpretable, requiring computational analysis to identify trends. Machine learning is increasingly employed, however, it requires feature selection which has a critical and considerable effect on machine learning problems when applied to large-scale data to reduce the number of parameters, improve generalization, and find essential predictors. In this study, using 7k proteomic data generated from the analysis of serum obtained from 82 patients with GBM pre- and post-completion of concurrent chemoirradiation (CRT), we aimed to select the most discriminative proteomic features that define proteomic alteration that is the result of administering CRT. Thus, we present a novel rank-based feature weighting method (RadWise) to identify relevant proteomic parameters using two popular feature selection methods, least absolute shrinkage and selection operator (LASSO) and the minimum redundancy maximum relevance (mRMR). The computational results show that the proposed method yields outstanding results with very few selected proteomic features, with higher accuracy rate performance than methods that do not employ a feature selection process. While the computational method identified several proteomic signals identical to the clinical intuitive (heuristic approach), several heuristically identified proteomic signals were not selected while other novel proteomic biomarkers not selected with the heuristic approach that carry biological prognostic relevance in GBM only emerged with the novel method. The computational results show that the proposed method yields promising results, reducing 7k proteomic data to 8 selected proteomic features with a performance value of 96.364%, comparing favorably with techniques that do not employ feature selection.

Enhancing Small Medical Dataset Classification Performance Using GAN

Developing an effective classification model in the medical field is challenging due to limited datasets. To address this issue, this study proposes using a generative adversarial network (GAN) as a data-augmentation technique. The research aims to enhance the classifier’s generalization performance, stability, and precision through the generation of synthetic data that closely resemble real data. We employed feature selection and applied five classification algorithms to thirteen benchmark medical datasets, augmented using the least-square GAN (LS-GAN). Evaluation of the generated samples using different ratios of augmented data showed that the support vector machine model outperforms other methods with larger samples. The proposed data augmentation approach using a GAN presents a promising solution for enhancing the performance of classification models in the healthcare field.

A Survey of Feature Selection Strategies for DNA Microarray Classification

Classification tasks are difficult and challenging in the bioinformatics field, that used to predict or diagnose patients at an early stage of disease by utilizing DNA microarray technology. However, crucial characteristics of DNA microarray technology are a large number of features and small sample sizes, which means the technology confronts a "dimensional curse" in its classification tasks because of the high computational execution needed and the discovery of biomarkers difficult. To reduce the dimensionality of features to find the significant features that can employ feature selection algorithms and not affect the performance of classification tasks. Feature selection helps decrease computational time by removing irrelevant and redundant features from the data. The study aims to briefly survey popular feature selection methods for classifying DNA microarray technology, such as filters, wrappers, embedded, and hybrid approaches. Furthermore, this study describes the steps of the feature selection process used to accomplish classification tasks and their relationships to other components such as datasets, cross-validation, and classifier algorithms. In the case study, we chose four different methods of feature selection on two-DNA microarray datasets to evaluate and discuss their performances, namely classification accuracy, stability, and the subset size of selected features. Keywords: Brief survey; DNA microarray data; feature selection; filter methods; wrapper methods; embedded methods; and hybrid methods. DOI: 10.7176/CEIS/14-2-01 Publication date: March 31 st 2023

A Comparison Between the Performance of Features Selection Techniques: Survey Study

Feature selection is one of the most popular and crucial methods of data processing used in different machine learning and data mining approaches to avoid high dimensionality and increase classification accuracy. Additionally, attribute selection aids in accelerating machine learning algorithms, improving prediction accuracy, data comprehension, decreasing data storage space, and minimizing the computational complexity of learning algorithms. For this reason, several feature selection approaches are used. To determine the essential feature or feature subsets needed to achieve classification objectives, several feature selection techniques have been suggested in the literature. In this research, different widely employed feature selection strategies have been evaluated by using different datasets to see how efficiently these techniques may be applied to achieve high performance of learning algorithms, which improves the classifier's prediction accuracy.

Leveraging Scheme for Cross-Study Microbiome Machine Learning Prediction and Feature Evaluations.

The microbiota has proved to be one of the critical factors for many diseases, and researchers have been using microbiome data for disease prediction. However, models trained on one independent microbiome study may not be easily applicable to other independent studies due to the high level of variability in microbiome data. In this study, we developed a method for improving the generalizability and interpretability of machine learning models for predicting three different diseases (colorectal cancer, Crohn's disease, and immunotherapy response) using nine independent microbiome datasets. Our method involves combining a smaller dataset with a larger dataset, and we found that using at least 25% of the target samples in the source data resulted in improved model performance. We determined random forest as our top model and employed feature selection to identify common and important taxa for disease prediction across the different studies. Our results suggest that this leveraging scheme is a promising approach for improving the accuracy and interpretability of machine learning models for predicting diseases based on microbiome data.

Genome-wide family prediction unveils molecular mechanisms underlying the regulation of agronomic traits in Urochloa ruziziensis.

Tropical forage grasses, particularly those belonging to the Urochloa genus, play a crucial role in cattle production and serve as the main food source for animals in tropical and subtropical regions. The majority of these species are apomictic and tetraploid, highlighting the significance of U. ruziziensis, a sexual diploid species that can be tetraploidized for use in interspecific crosses with apomictic species. As a means to support breeding programs, our study investigates the feasibility of genome-wide family prediction in U. ruziziensis families to predict agronomic traits. Fifty half-sibling families were assessed for green matter yield, dry matter yield, regrowth capacity, leaf dry matter, and stem dry matter across different clippings established in contrasting seasons with varying available water capacity. Genotyping was performed using a genotyping-by-sequencing approach based on DNA samples from family pools. In addition to conventional genomic prediction methods, machine learning and feature selection algorithms were employed to reduce the necessary number of markers for prediction and enhance predictive accuracy across phenotypes. To explore the regulation of agronomic traits, our study evaluated the significance of selected markers for prediction using a tree-based approach, potentially linking these regions to quantitative trait loci (QTLs). In a multiomic approach, genes from the species transcriptome were mapped and correlated to those markers. A gene coexpression network was modeled with gene expression estimates from a diverse set of U. ruziziensis genotypes, enabling a comprehensive investigation of molecular mechanisms associated with these regions. The heritabilities of the evaluated traits ranged from 0.44 to 0.92. A total of 28,106 filtered SNPs were used to predict phenotypic measurements, achieving a mean predictive ability of 0.762. By employing feature selection techniques, we could reduce the dimensionality of SNP datasets, revealing potential genotype-phenotype associations. The functional annotation of genes near these markers revealed associations with auxin transport and biosynthesis of lignin, flavonol, and folic acid. Further exploration with the gene coexpression network uncovered associations with DNA metabolism, stress response, and circadian rhythm. These genes and regions represent important targets for expanding our understanding of the metabolic regulation of agronomic traits and offer valuable insights applicable to species breeding. Our work represents an innovative contribution to molecular breeding techniques for tropical forages, presenting a viable marker-assisted breeding approach and identifying target regions for future molecular studies on these agronomic traits.

Feature selection and association rule learning identify risk factors of malnutrition among Ethiopian schoolchildren.

Previous studies have sought to identify risk factors for malnutrition in populations of schoolchildren, depending on traditional logistic regression methods. However, holistic machine learning (ML) approaches are emerging that may provide a more comprehensive analysis of risk factors. This study employed feature selection and association rule learning ML methods in conjunction with logistic regression on epidemiological survey data from 1,036 Ethiopian school children. Our first analysis used the entire dataset and then we reran this analysis on age, residence, and sex population subsets. Both logistic regression and ML methods identified older childhood age as a significant risk factor, while females and vaccinated individuals showed reduced odds of stunting. Our machine learning analyses provided additional insights into the data, as feature selection identified that age, school latrine cleanliness, large family size, and nail trimming habits were significant risk factors for stunting, underweight, and thinness. Association rule learning revealed an association between co-occurring hygiene and socio-economical variables with malnutrition that was otherwise missed using traditional statistical methods. Our analysis supports the benefit of integrating feature selection methods, association rules learning techniques, and logistic regression to identify comprehensive risk factors associated with malnutrition in young children.

Understanding risk factors of a new variant outburst through global analysis of Omicron transmissibility

The emergence of a new virus variant is generally recognized by its usually sudden and rapid spread (outburst) in a certain world region. Due to the near-exponential rate of initial expansion, the new strain may not be detected at its true geographical origin but in the area with the most favorable conditions leading to the fastest exponential growth. Therefore, it is crucial to understand better the factors that promote such outbursts, which we address in the example of analyzing global Omicron transmissibility during its global emergence/outburst in November 2021–February 2022. As predictors, we assemble a number of potentially relevant factors: vaccinations (both full and boosters), different measures of population mobility (provided by Google), estimated stringency of measures, the prevalence of chronic diseases, population age, the timing of the outburst, and several other socio-demographic variables. As a proxy for natural immunity (prevalence of prior infections in population), we use cumulative numbers of COVID-19 deaths. As a response variable (transmissibility measure), we use the estimated effective reproduction number (Re) averaged in the vicinity of the outburst maxima. To select significant predictors of Re, we use machine learning regressions that employ feature selection, including methods based on ensembles of decision trees (Random Forest and Gradient Boosting). We identify the young population, earlier infection onset, higher mobility, low natural immunity, and low booster prevalence as likely direct risk factors. Interestingly, we find that all these risk factors were significantly higher for Africa, though curiously somewhat lower in Southern African countries (where the outburst emerged) compared to other African countries. Therefore, while the risk factors related to the virus transmissibility clearly promote the outburst of a new virus variant, specific regions/countries where the outburst actually happens may be related to less evident factors, possibly random in nature.

Minimum relevant features to obtain AI explainable system for predicting breast cancer in WDBC

The potential to explain why a machine learning model produces a certain prediction in incomprehensible terms is becoming increasingly crucial, as it provides accountability and confidence in the algorithm's decision-making process. The interpretation of complex models is difficult. Various approaches to dealing with this issue are being offered. These problems are typically handled in tree ensemble methods by assigning priority levels to input features globally or for a specific prediction. We show that current feature attribution approaches are inconclusive, and develop solutions using SHAP (SHapley Additive Explanation) values, LIME (Local Interpretable Model-Agnostic Explanations), and the Skope Rules package. We employ feature selection methods from SHAP and LIME in this work, which uses the Breast cancer Wisconsin data sets. In the suggested method, features are chosen at the first level of feature selection using Decision tree entropy values. Based on the SHAP and LIME reports, level 2 features are chosen from fewer options. The features are tested on a Decision Tree (DT) model and a DT and Support Vector Machine (SVM) ensemble. Experiments suggest that the ensemble works better as compared to DT. We have also used the Skope Rules package to generate global rules for generalization.

Data‐driven storage operations: Cross‐commodity backtest and structured policies

Storage assets are critical for physical trading of commodities under volatile prices. State‐of‐the‐art methods for managing storage facilities such as the reoptimization heuristic (RH), which are part of commercial software, approximate a Markov Decision Process (MDP) assuming full information regarding the state and the stochastic commodity price process and hence suffer from informational inconsistencies with observed price data and structural inconsistencies with the true optimal policy, which are both components of generalization error. Focusing on spot trades, we find via an extensive backtest that this error can lead to significantly suboptimal RH policies. We develop a forward‐looking data‐driven approach (DDA) to learn policies and reduce generalization error. This approach extends standard (backward‐looking) DDA in two ways: (i) It represents historical and estimated future profits as functions of features in the training objective, which typically includes only past profits; and (ii) it enforces structural properties of the optimal policy. To elaborate, DDA trains parameters of bang‐bang and base‐stock policies, respectively, using linear‐ and mixed‐integer programs, thereby extending known DDAs that parameterize decisions as functions of features without policy structure. We backtest the performance of RH and DDA on six major commodities, employing feature selection across data from Reuters, Bloomberg, and other public data sets. DDA can improve RH on real data, with policy structure needed to realize this improvement. Our research advances the state‐of‐the‐art for storage operations and can be extended beyond spot trading to handle generalization error when also including forward trades.

Evaluating the impact of feature selection consistency in software prediction

Many empirical software engineering studies have employed feature selection algorithms to exclude the irrelevant and redundant features from the datasets with the aim to improve prediction accuracy achieved with machine learning-based estimation models as well as their generalizability. However, little has been done to investigate how consistently these feature selection algorithms produce features/metrics across different training samples, which is an important point for the interpretation of the trained models. The interpretation of the models largely depends on the features of the analyzed datasets, so it is recommended to evaluate the potential of various feature selection algorithms in terms of how consistently they extract features from the employed datasets. In this study, we consider eight different feature selection algorithms to evaluate how consistently they select features across different folds of k-fold cross-validation as well as when small changes are made in the training data. To provide a stable and generalized conclusion, we investigate data from two different domains, i.e., six datasets from the domain of Software Development Effort Estimation (SDEE) and six datasets from the Software Fault Prediction (SFP) domain. Our results reveal that a feature selection algorithm could produce 20-100% inconsistent features with an SDEE dataset and 18.8-95.3% inconsistent features in the case of an SFP dataset. The analysis also reveals that it is not necessarily true that the most consistent feature selection algorithm results to be the most accurate one (i.e., leads to better prediction accuracy) in the case of SDEE datasets, while with SFP datasets, the analysis highlights that the most consistent feature selection algorithm also results to be the most accurate in predicting faults. • Not all feature selection algorithms are reported as consistent. • Best-First Search (backward) is the most consistent feature selection algorithm. • Harmony Search is the least consistent feature selection algorithm with SDEE.

An optimal weighted machine learning model for detecting financial fraud

ABSTRACT The purpose of this study is to predict enterprises’ financial fraud. After collecting financial data and employing feature selection methods, totally eight features are selected. We select two best-performance machine learning models according to five indicators including accuracy, recall, specificity, AUC and the misclassification cost. Besides, an optimal weighted machine learning model, based on the two best-performance models, is proposed and the results confirm its good performance in forecasting enterprises’ financial fraud.

B-PO05-177 USING MACHINE LEARNING TO PREDICT CARDIOMYOPATHY IN PATIENTS WITH CORONARY ARTERY DISEASE

Low left ventricular ejection fraction (LVEF) related to CAD has a risk of sudden cardiac death and requires evaluation for ICD placement. To develop a neural network model and identify features that would help predict the LVEF. Patients (n=2116) (table 1) hospitalized with a diagnosis of CAD (from 2010- 2019) with ECG, cardiac catheterization, baseline LVEF, and follow-up LVEF after 60 days from admission were studied. We employed feature selection techniques of Pearson correlation and mutual information for identifying the most related features to LVEF. The final dataset was then applied to a neural network model for predicting the follow-up LVEF (flow chart). Of the patients studied, (mean age 64.9 + 13.03 yrs), 302 had LVEF ≤ 35% on presentation and 370 had LVEF ≤ 35% on 60 days follow-up. Then, we trained neural network and linear regression models with the above features to predict LVEF within a scale of [0,1] on an 80-20% train test dataset. Of the 220 variables assessed using statistical-based feature selection techniques to determine a smaller dimension frame dataset of 21 high strength features (table 2). The model’s performance was validated using the mean squared error (0.01 error) and area under receiver operator characteristic curves and obtained a score of around 0.68. We trained a neural network model for predicting LVEF in patients with CAD. We plan on applying more data to develop a robust model that improves our ability to predict sudden cardiac death in CAD patients.

On a clustering-based mining approach with labeled semantics for significant place discovery

With the rapid increase in GPS data collection through pervasive use of mobile devices, it has become an important problem to discover significant places of moving objects from complex spatial and temporal trajectories. This problem is challenging mainly because such trajectory data suffer from several issues including incompleteness, low quality, high redundancy, and oftentimes trajectory points do not follow Gaussian distribution. We propose a clustering-based method with temporal and spatial semantics, referred to as Stops and Moves of Trajectories using Attribute Selection (SMoTAS), whose technical advantages are multifold. Firstly, it improves data availability by using a self-adaptive algorithm to correct the deviation in traditional speed-based methods. Secondly, it improves place mining accuracy by filtering multi-label clustering results when there is a lack of detailed geographic data. Thirdly, it employs feature selection to exploit the core attributes of clustering and simplify the clustering results with Grubbs criterion. Experimental results on real-life datasets show that SMoTAS not only achieves substantial improvement of accuracy over existing methods in discovering significant places, but also exhibits superior adaptability to different trajectories and application scenarios.

Feature Selection Method Based on Grey Wolf Optimization and Simulated Annealing

Background: Feature selection sometimes also known as attribute subset selection is a process in which optimal subset of features are elected with respect to target data by reducing dimensionality and removing irrelevant features. There will be 2n possible solutions for a dataset having n number of features that is difficult to solve by conventional attribute selection method. In such cases metaheuristic-based methods generally outruns the conventional methods. Objective: The main aim of this paper is to enhance the classification accuracy and minimize the number of selected features and error rate. Methods: To achieve the objective, a binary metaheuristic feature selection method bGWOSA based on grey wolf optimization and simulated annealing has been introduced. The proposed feature selection method uses simulated annealing for equalizing the trade-off between exploration and exploitation. The performance of the proposed binary feature selection method has been examined on the ten feature selection benchmark datasets taken from UCI repository and compared with binary cuckoo search, binary particle swarm optimization, binary grey wolf optimization, binary bat algorithm and binary hybrid whale optimization method. Results: The proposed feature selection method achieves the highest accuracy for the most of datasets compared to state-of-the-art. Further, from the experimental and statistical results, efficacy of the proposed feature selection method has been validated. Conclusion: Classification accuracy can be enhanced by employing feature selection methods. Moreover, performance can also be enhanced by tuning the control parameters of metaheuristic methods.

Performance Analysis of Machine Learning Algorithms for Thyroid Disease

Thyroid disease arises from an anomalous growth of thyroid tissue at the verge of the thyroid gland. Thyroid disorderliness normally ensues when this gland releases abnormal amounts of hormones where hypothyroidism (inactive thyroid gland) and hyperthyroidism (hyperactive thyroid gland) are the two main types of thyroid disorder. This study proposes the use of efficient classifiers by using machine learning algorithms in terms of accuracy and other performance evaluation metrics to detect and diagnose thyroid disease. This research presents an extensive analysis of different classifiers which are K-nearest neighbor (KNN), Naive Bayes, support vector machine, decision tree and logistic regression implemented with or without feature selection techniques. Thyroid data were taken from DHQ Teaching Hospital, Dera Ghazi Khan, Pakistan. Thyroid dataset was unique and different from other existing studies because it included three additional features which were pulse rate, body mass index and blood pressure. Experiment was based on three iterations; the first iteration of the experiment did not employ feature selection while the second and third were with L1-, L2-based feature selection technique. Evaluation and analysis of the experiment have been done which consisted of many factors such as accuracy, precision and receiver operating curve with area under curve. The result indicated that classifiers which involved L1-based feature selection achieved an overall higher accuracy (Naive Bayes 100%, logistic regression 100% and KNN 97.84%) compared to without feature selection and L2-based feature selection technique.

LangAware: A robust ML‐based platform for the quick and accurate assessment of neurodegenerative diseases

AbstractBackgroundAlthough Alzheimer's disease (AD) is associated with changes in spoken language, these have seldom been subjected to systematic analysis on a large scale. We evaluated the effectiveness of LangAware to detect the language indicators that are coupled with early AD, thus assisting with diagnosis. We evaluated LangAware using recordings of speech samples obtained from AD patients and matched healthy controls (NC) derived from various elicitation tasks in two languages, English and Greek*.MethodEnglish and Greek datasets were analyzed employing feature selection techniques to choose the most prominent multi‐level linguistic analysis features differentiating the AD from the NC group in both languages. The platform’s diagnostic performance was evaluated on its ability to classify "unseen" audio recordings employing these salient features.ResultEvaluation results indicated that LangAware achieved equally high classification scores for both English and Greek. Most significantly, these scores were achieved by employing a custom set of LangAware‐developed cross‐linguistic markers.ConclusionThe current evaluation verified the robustness of the platform’s predictive models using audio datasets in two languages. Based on the findings, we conclude that LangAware could provide a time and cost‐effective platform for cognitive screening across languages and across tasks pertaining to neurodegenerative diseases in a range of clinical settings. Such findings also advocate in favor of the robustness of LangAware platform in pursuing cognitive assessment on spontaneous speech across languages. *Peter Garrard and Antonis A. Mougias contributed to this work by providing data for the evaluation of the LangAware platform.

Jointly optimization for activity recognition in secure IoT-enabled elderly care applications

Elderly care is a significant livelihood project in the increasingly serious aging society. Nowadays, the wider application of the Internet of Things (IoT) technology on assistant means has making an importance contribution to elderly care in institutions and at home. On the broader data foundation collected by IoT devices, Human Activity Recognition (HAR) with its high demand in various elderly care applications also has grabbed considerable research attentions. However, the elderly’s sensitive data transmitted over the wireless communication channel need to address many security concerns, and the accuracy of activity recognition is susceptible to many influencing factors, especially, the employed feature selection method and classifier. In this paper, to ensure the confidentiality of the elderly’s sensitive data, a secure and efficient group-based key establishment and authentication framework is first proposed. Subsequently, activity recognition is investigated within the security data sensitive framework, where a feature reorganization based feature selection method is proposed with the demonstrated relationship between the recognition accuracy and strongly correlative features, and three classifiers are investigated based on classical ones. By contrast, an improved convolutional neural network (CNN) enabled classifier is singled out to conduct the activity recognition with the feature reorganization method. Finally, security proofs show that the proposed security framework can ensure data confidentiality and be resilient to possible attacks, and experimental analyses show investigations in activity recognition can achieve more cost-effective results.

Estimating Postural Stability Using Improved Permutation Entropy via TUG Accelerometer Data for Community-Dwelling Elderly People.

To develop an effective fall prevention program, clinicians must first identify the elderly people at risk of falling and then take the most appropriate interventions to reduce or eliminate preventable falls. Employing feature selection to establish effective decision making can thus assist in the identification of a patient’s fall risk from limited data. This work therefore aims to supplement professional timed up and go assessment methods using sensor technology, entropy analysis, and statistical analysis. The results showed the different approach of applying logistic regression analysis to the inertial data on a fall-risk scale to allow medical practitioners to predict for high-risk patients. Logistic regression was also used to automatically select feature values and clinical judgment methods to explore the differences in decision making. We also calculate the area under the receiver-operating characteristic curve (AUC). Results indicated that permutation entropy and statistical features provided the best AUC values (all above 0.9), and false positives were avoided. Additionally, the weighted-permutation entropy/statistical features test has a relatively good agreement rate with the short-form Berg balance scale when classifying patients as being at risk. Therefore, the proposed methodology can provide decision-makers with a more accurate way to classify fall risk in elderly people.

Subject-Specific feature selection for near infrared spectroscopy based brain-computer interfaces

Background and ObjectiveBrain-computer interfaces (BCIs) enable people to control an external device by analyzing the brain's neural activity. Functional near-infrared spectroscopy (fNIRS), which is an emerging optical imaging technique, is frequently used in non-invasive BCIs. Determining the subject-specific features is an important concern in enhancing the classification accuracy as well as reducing the complexity of fNIRS based BCI systems. In this study, the effectiveness of subject-specific feature selection on classification accuracy of fNIRS signals is examined. MethodsIn order to determine the subject-specific optimal feature subsets, stepwise regression analysis based on sequential feature selection (SWR-SFS) and ReliefF methods were employed. Feature selection is applied on time-domain features of fNIRS signals such as mean, slope, peak, skewness and kurtosis values of signals. Linear discriminant analysis, k nearest neighborhood and support vector machines are employed to evaluate the performance of the selected feature subsets. The proposed techniques are validated on benchmark motor imagery (MI) and mental arithmetic (MA) based fNIRS datasets collected from 29 healthy subjects. ResultsBoth SWR-SFS and reliefF feature selection methods have significantly improved the classification accuracy. However, the best results (88.67% (HbR) and 86.43% (HbO) for MA dataset and 77.01% (HbR) and 71.32% (HbO) for MI dataset) were achieved using SWR-SFS while feature selection provided extremely high feature reduction rates (89.50% (HbR) and 93.99% (HbO) for MA dataset and 94.04% (HbR) and 97.73% (HbO) for MI dataset). ConclusionsThe results of the study indicate that employing feature selection improves both MA and MI-based fNIRS signals classification performance significantly.