Feature Selection on K-Nearest Neighbor Algorithm Using Similarity Measure

Puspadini, Ratih; Mawengkang, Herman; Efendi, Syahril

doi:10.1109/mecnit48290.2020.9166612

Cited by 7 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Sequential Random k-nearest neighbors (SRk-NN) algorithm is used to select features based on the majority vote of nearest neighbor classifiers [24], and a distanceand attribute-weighted k-NN-based algorithm has also been utilized for feature selection [25]. The performance of k -NN-based feature selections has already been examined [26], and a method for tuning the number of neighbors (k) has also been developed, along with goal-oriented similarity measures [27]. There has been precedent for the use of genetic algorithm-based feature selection to improve the performance of k-NN in a classification problem [28].…”

Section: G Related Work Of Neighborhood-based Methodsmentioning

confidence: 99%

Neighborhood Ranking-Based Feature Selection

Ipkovich,

Abonyi

2024

IEEE Access

View full text Add to dashboard Cite

This article aims to integrate k-NN regression, false-nearest neighborhood (FNN), and trustworthiness and continuity (T&C) neighborhood-based measures into an efficient and robust feature selection method to support the identification of nonlinear regression models. The proposed neighborhood ranking-based feature selection technique (NRFS) is validated in three problems, in a linear regression task, in the nonlinear Friedman database, and in the problem of determining the order of nonlinear dynamical models. A neural network is also identified to validate the resulting feature sets. The analysis of the distance correlation also confirms that the method is capable of exploring the nonlinear correlation structure of complex systems. The results illustrate that the proposed NRFS method can select relevant variables for nonlinear regression models.

show abstract

Section: G Related Work Of Neighborhood-based Methodsmentioning

confidence: 99%

Neighborhood Ranking-Based Feature Selection

Ipkovich,

Abonyi

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The curse of dimensionality means that Euclidean distance is unhelpful in very high dimensions because all points in the training set are almost equidistant to the search point. Hence, it is advisable to preliminarily reduce the number of variables by applying, for example, a technique for feature selection or feature extraction [ 35 ].…”

Section: Methodsmentioning

confidence: 99%

“…First, a 50 Hz notch filter was applied to EEG data in order to remove line noise due to signal acquisition and digital conversion. Afterwards, a band-pass filter was applied on the noise-free EEG data in order to extract the signal content in six different frequency bands corresponding to well-known oscillations of brain activity, namely, (8-13) Hz (α), (13-21) Hz (β1), (21)(22)(23)(24)(25)(26)(27)(28)(29)(30) Hz (β2), (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) Hz (low 𝛾), (40-70) Hz (medium 𝛾), and (70-120) Hz (high 𝛾). Finally, a sliding window paradigm was applied to the noise-free, band-passed EEG data by setting two temporal parameters: 𝐿, which represents the time length of the analysis window, and 𝑆, which is the time shift of the window which slides on the signal.…”

Section: Eeg Preprocessingmentioning

confidence: 99%

Machine Learning Characterization of Ictal and Interictal States in EEG Aimed at Automated Seizure Detection

Zazzaro

Pavone²

2022

Biomedicines

View full text Add to dashboard Cite

Background: The development of automated seizure detection methods using EEG signals could be of great importance for the diagnosis and the monitoring of patients with epilepsy. These methods are often patient-specific and require high accuracy in detecting seizures but also very low false-positive rates. The aim of this study is to evaluate the performance of a seizure detection method using EEG signals by investigating its performance in correctly identifying seizures and in minimizing false alarms and to determine if it is generalizable to different patients. Methods: We tested the method on about two hours of preictal/ictal and about ten hours of interictal EEG recordings of one patient from the Freiburg Seizure Prediction EEG database using machine learning techniques for data mining. Then, we tested the obtained model on six other patients of the same database. Results: The method achieved very high performance in detecting seizures (close to 100% of correctly classified positive elements) with a very low false-positive rate when tested on one patient. Furthermore, the model portability or transfer analysis revealed that the method achieved good performance in one out of six patients from the same dataset. Conclusions: This result suggests a strategy to discover clusters of similar patients, for which it would be possible to train a general-purpose model for seizure detection.

show abstract

“…Banyak fungsi metrik jarak yang digunakan dalam KNN dan algoritma machine learning lainnya. Pada umumnya KNN menggunakan metode perhitungan jarak dengan Euclidean distance yang berfungsi untuk menguji ukuran sebagai bentuk jarak kedekatan antara dua objek [20]. Tetangga terdekat pada KNN didefinisikan dalam rentang Euclidean antara dua objek 𝑋 = (𝑥 1 , 𝑥 2 , .…”

Section: Gambar 1 Diagram Alir Penelitianunclassified

Comparison of k-Nearest Neighbor and Support Vector Machine using Binary Dragonfly Algorithm Optimization

Nugroho,

Khomeini,

Heraldi

2024

SISTEMASI

View full text Add to dashboard Cite

BDA merupakan adaptasi dari Dragonfly Algorithm (DA) yang bertujuan untuk mengoptimalkan komputasi permasalahan single-objecivet, discrete, dan multi-objective. Penggabungan algoritma optimasi BDA dengan algoritma klasifikasi KNN dan SVM bertujuan untuk meningkatkan kinerja model prediksi. Penelitian ini bertujuan membandingkan dan menguji keakuratan akurasi dari algoritma KNN dan SVM terhadap dataset diabetes yang digunakan pada penelitian untuk mengetahui algoritma terbaik dalam memprediksi diabetes. Penelitian ini menggunakan algoritma optimasi BDA untuk menyeleksi fitur-fitur terbaik pada dataset, kemudian algoritma klasifikasi KNN dan SVM, dalam mengklasifikasikan data, memprediksi, dan membandingkan keakuratan akurasi dari kedua algoritma terhadap dataset diabetes. Data rekam medis dari pengidap diabetes diolah menggunakan algoritma KNN dan SVM, yang kemudian akan menghasilkan tingkat akurasi yang dapat digunakan dalam memprediksi diabetes. Penelitian terdahulu, telah melakukan perbandingan antara algoritma-algoritma klasifikasi dalam memprediksi penyakit diabetes. Pada penelitian terdahulu diatas belum ada yang menggabungkan BDA dengan algoritma klasifikasi, karena BDA sendiri merupakan suatu metode yang relatif baru dan belum banyak diteliti, sehingga peneliti menggunakan algoritma optimasi ini. Hasil dari penelitian yang dilakukan mendapatkan hasil akurasi tertinggi pada algoritma BDA + KNN dengan

show abstract

Feature Selection on K-Nearest Neighbor Algorithm Using Similarity Measure

Cited by 7 publications

References 12 publications

Neighborhood Ranking-Based Feature Selection

Neighborhood Ranking-Based Feature Selection

Machine Learning Characterization of Ictal and Interictal States in EEG Aimed at Automated Seizure Detection

Comparison of k-Nearest Neighbor and Support Vector Machine using Binary Dragonfly Algorithm Optimization

Contact Info

Product

Resources

About