A Review of Feature Selection and Feature Extraction Methods Applied on Microarray Data

Hira, Zena M.; Gillies, Donald

doi:10.1155/2015/198363

Cited by 770 publications

(521 citation statements)

References 85 publications

Supporting

Mentioning

498

Contrasting

Unclassified

Order By: Relevance

“…FE involves the construction of a new (smaller) feature set derived from the full dataset (Hira & Gillies, 2015). In this study, this dataset was constructed using the principle component analysis (PCA) tool within ArcMap v10.3.…”

Section: Methodsmentioning

confidence: 99%

Impact of ecological redundancy on the performance of machine learning classifiers in vegetation mapping

MacIntyre

Niekerk

Dobrowolski

et al. 2018

Ecology and Evolution

View full text Add to dashboard Cite

Vegetation maps are models of the real vegetation patterns and are considered important tools in conservation and management planning. Maps created through traditional methods can be expensive and time‐consuming, thus, new more efficient approaches are needed. The prediction of vegetation patterns using machine learning shows promise, but many factors may impact on its performance. One important factor is the nature of the vegetation–environment relationship assessed and ecological redundancy. We used two datasets with known ecological redundancy levels (strength of the vegetation–environment relationship) to evaluate the performance of four machine learning (ML) classifiers (classification trees, random forests, support vector machines, and nearest neighbor). These models used climatic and soil variables as environmental predictors with pretreatment of the datasets (principal component analysis and feature selection) and involved three spatial scales. We show that the ML classifiers produced more reliable results in regions where the vegetation–environment relationship is stronger as opposed to regions characterized by redundant vegetation patterns. The pretreatment of datasets and reduction in prediction scale had a substantial influence on the predictive performance of the classifiers. The use of ML classifiers to create potential vegetation maps shows promise as a more efficient way of vegetation modeling. The difference in performance between areas with poorly versus well‐structured vegetation–environment relationships shows that some level of understanding of the ecology of the target region is required prior to their application. Even in areas with poorly structured vegetation–environment relationships, it is possible to improve classifier performance by either pretreating the dataset or reducing the spatial scale of the predictions.

show abstract

Section: Methodsmentioning

confidence: 99%

Impact of ecological redundancy on the performance of machine learning classifiers in vegetation mapping

MacIntyre

Niekerk

Dobrowolski

et al. 2018

Ecology and Evolution

View full text Add to dashboard Cite

show abstract

“…Moreover, using extracted features to predict the class of new cases sometimes leads to poor performance, which means that the method per forms well in training data but fails to classify the testing data well. Hira and Gillies (2015) comprehensively discuss the weaknesses and advantages of feature selection and extraction. This situation leads to the suggestion that the full features should be used in some cases.…”

Section: Introductionmentioning

confidence: 99%

Random Forest Classification and Support Vector Machine for Detecting Epilepsyusing Electroencephalograph Records

Kuswanto¹,

Salamah²,

Fachruddin³

2017

American Journal of Applied Sciences

View full text Add to dashboard Cite

Complexity in data structure has led to the rapid development of computational statistics methods. Machine learning approaches have been introduced and applied to solve complex problems in many fields. This paper applies two common machine learning approaches, Random Forest (RF) and Support Vector Machine (SVM), in the detection of epilepsy. The diagnosis of epilepsy can usually only be made when a seizure is happening, which leads to some difficulties in the diagnostic process. The most recent way of diagnosing epilepsy is by using an Electroencephalograph (EEG) record. However, detecting epilepsy cases through EEG records takes a long time and may lead to misleading diagnostic results. The use of machine learning approaches is intended to generate fast and accurate classification results. As the EEG only generates a signal, direct analysis using RF or SVM cannot be carried out and the EEG record needs to be pre-processed. This paper uses Discrete Wavelet Transform and Line Length Features in the data preprocessing stage to decompose the signal by frequency and time. The classification results show that both RF and SVM perform very well and are able to classify cases of epilepsy accurately. The RF outperforms the SVM in the training dataset, while the SVM has a better performance in testing, with almost nom is classified cases. Several open problems relating to interpretation as well as parameter settings are described.

show abstract

“…Feature selection methods can be structured into three clusters: filter-based, wrapper-based and embedded [193]. All have some major drawbacks and we therefore propose a hybrid feature selection (HFS) algorithm.…”

Section: Feature Selectionmentioning

confidence: 99%

“…Deterministic wrappers such as sequential selection and its variants, sequential forward selection (SFS), add or remove features sequentially. They are computationally intensive ('greedy search') and have a risk of overfitting [193]. On the other hand, randomised methods such as simulated annealing, genetic algorithm and differential evaluation incorporate randomness into their search procedure to escape local minima but still tend to suffer from overfitting [193].…”

Section: Background Problemsmentioning

confidence: 99%

“…They are computationally intensive ('greedy search') and have a risk of overfitting [193]. On the other hand, randomised methods such as simulated annealing, genetic algorithm and differential evaluation incorporate randomness into their search procedure to escape local minima but still tend to suffer from overfitting [193]. The embedded approach is less computationally intensive than wrappers but provides a classifier-dependent selection.…”

Section: Background Problemsmentioning

confidence: 99%

See 1 more Smart Citation

Design and optimisation of time-frequency analysis for multichannel neonatal EEG background features in term neonates with hypoxic ischaemic encephalopathy: characterisation, classification and neurodevelopmental outcome prediction

Awal¹

View full text Add to dashboard Cite

Neonatal Hypoxic Ischaemic Encephalopathy (HIE) is a major cause of morbidity and mortality in newborns throughout the world with an incidence of 1-8/1000 newborns in developed countries and up to 26/1000 newborns in developing countries. It causes one million neonatal deaths globally per year. Electroencephalography (EEG) is a useful method for assessment of brain activity in newborns with HIE. EEG performs well in the early diagnosis and classification of HIE severity and in predicting neurodevelopmental outcome. A normal EEG is highly predictive of a normal outcome and various abnormal EEG features are associated with neurological abnormalities or death.A systematic review and meta-analysis of the literature was undertaken to determine which specific background features of the EEG best predict outcome (Aim 1). Automatic detection and classification

show abstract

A Review of Feature Selection and Feature Extraction Methods Applied on Microarray Data

Cited by 770 publications

References 85 publications

Impact of ecological redundancy on the performance of machine learning classifiers in vegetation mapping

Impact of ecological redundancy on the performance of machine learning classifiers in vegetation mapping

Random Forest Classification and Support Vector Machine for Detecting Epilepsyusing Electroencephalograph Records

Design and optimisation of time-frequency analysis for multichannel neonatal EEG background features in term neonates with hypoxic ischaemic encephalopathy: characterisation, classification and neurodevelopmental outcome prediction

Contact Info

Product

Resources

About