A Review of various KNN Techniques

Rani, Pooja

doi:10.22214/ijraset.2017.8166

Cited by 11 publications

(7 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In K-NN classification, the class of the query is obtained based on the nearest neighbors to that of the example query. K-NN searches the pattern to the closest data and then assigns to data that is unknown [27]. In this classification, we used Euclidean distance as a criterion along with 10-fold cross validation to predict the class label from the prediction dataset.…”

Section: Resultsmentioning

confidence: 99%

Comparison of Various Data Mining Classification Techniques in the Diagnosis of Diabetic Retinopathy

Vadloori¹,

Huang²,

Wu³

2019

ACTA POLYTECH HUNG

View full text Add to dashboard Cite

Diabetic retinopathy (DR) has been the most frequently occurring complication in the patients suffering from a long-term diabetic condition, that ultimately leads to blindness. Early detection of the disease through biomarkers and effective treatment has been proposed to prevent/delay its occurrence. Several biomarkers have been explored, to help understand the incidence and progression of DR. These included the presence of microaneurysms, exudates, hemorrhages, etc. in the retina of the patients, which contributes to the disease. Investigation of the retinal images from time to time has been proposed as a strategy to prevent blindness. Evaluating the retinal images manually is time-consuming and demands great expertise in the diagnosis of DR. To circumvent such issues computer-aided diagnosis are very promising in the detection of DR. In the present study, we used a DR dataset and applied different classification algorithms in machine learning to predict the occurrence of the DR. The classifiers employed herein, included Knearest neighbor, random forest classifier, support vector machine, regression tree classifier, logistic regression and the Naïve Bayes theorem. Our results showed that the random forest classification model provided the significant detail of attributes in terms of their importance in the diagnosis of the DR. More importantly, our supervised classification models provided the prediction accuracy of the disease and Naïve Bayes classifier demonstrated highest accuracy of 80.15% in the prediction of DR compared to the others. Additionally, receiver operating characteristics (ROC) analysis, with the classifiers and the area under curve (AUC) represented the fitting results of each classifier. The presented approach can prove to be a potential tool for the ophthalmologist in the early diagnosis tool for DR.V. Spandana et al.

show abstract

Section: Resultsmentioning

confidence: 99%

Comparison of Various Data Mining Classification Techniques in the Diagnosis of Diabetic Retinopathy

Vadloori¹,

Huang²,

Wu³

2019

ACTA POLYTECH HUNG

View full text Add to dashboard Cite

show abstract

“…Before the rule was introduced for classification in Cover and Hart [1], the NN rule was mentioned in Nilsson [2] as "minimum distance classifier" and Sebestyen [3] as "proximity algorithm". The kNN method is an instance-based learning algorithm [4] which performs classification on a data point by determining parameter k (number of nearest neighbors), calculating the distance between the test instance and all the training examples and sorting the distance to determine nearest neighbors based on the kth minimum distance, gathering the class of the nearest neighbors and using the majority of the class of nearest neighbors as the prediction value of the test instance. The kNN algorithm are termed instance-based, or lazy learners because they make decisions by comparing the training set with the test set for each classification they perform.…”

Section: Introductionmentioning

confidence: 99%

“…They simply store the instances and they do not do any work on instances until it is given a test tuple. Other methods for classification such as rule-based classification, Decision Tree induction, Classification by Back-propagation, Bayesian classification, and Support Vector Machines (SVM) are examples of non-instance-based learners, otherwise known as eager learners [4,5,6].…”

Section: Introductionmentioning

confidence: 99%

An Enhanced Adaptive k-Nearest Neighbor Classifier Using Simulated Annealing

Onyezewe¹,

Kana²,

Abdullahi³

et al. 2021

IJISA

View full text Add to dashboard Cite

The k-Nearest Neighbor classifier is a non-complex and widely applied data classification algorithm which does well in real-world applications. The overall classification accuracy of the k-Nearest Neighbor algorithm largely depends on the choice of the number of nearest neighbors(k). The use of a constant k value does not always yield the best solutions especially for real-world datasets with an irregular class and density distribution of data points as it totally ignores the class and density distribution of a test point’s k-environment or neighborhood. A resolution to this problem is to dynamically choose k for each test instance to be classified. However, given a large dataset, it becomes very tasking to maximize the k-Nearest Neighbor performance by tuning k. This work proposes the use of Simulated Annealing, a metaheuristic search algorithm, to select optimal k, thus eliminating the prospect of an exhaustive search for optimal k. The results obtained in four different classification tasks demonstrate a significant improvement in the computational efficiency against the k-Nearest Neighbor methods that perform exhaustive search for k, as accurate nearest neighbors are returned faster for k-Nearest Neighbor classification, thus reducing the computation time.

show abstract

“…Secara skema yang diterapkan k-NN mampu mendeskripsikan atribut atau fitur dari sebuah data [8]. Sebagai algoritma yang digunakan pada kasus data mining, algoritma k-NN dapat diklasifikasi menjadi kategori k-NN terstruktur atau Structure base k-NN dan k-NN tidak terstruktur Non-structure base k-NN [9], [10]. Algoritma terstruktur tidak banyak digunakan karena dianggap kurang efisien dan membutuhkan waktu yang relatif lebih lama dalam pencarian variabel k pada saat awal proses, khususnya terhadap data dengan atribut banyak, sehingga Teknik Non-structure base k-NN dianggap lebih sederhana dan efisien namun berpengaruh pada akurasi terhadap metode penentuan nilai k dan penentuan pengukuran jarak terdekat.…”

unclassified

“…Fungsi pengukuran jarak berbasis Euclidean atau dikenal dengan Euclidean distance menjadi fungsi penentu jarak terdekat yang paling umum digunakan pada k-NN [11]. Selain pengukuran jarak berbasis gometri Euclid, fungsi jarak yang digunakan pada algoritma k-NN adalah Minkowski distance dan Manhattan distance [10]. Kinerja algoritma k-NN sangat dipengaruhi dari beberapa faktor, diantaranya adalah penggunaan fungsi jarak, penentuan nilai k dan nilai atribut yang tidak relevan [6] [3].…”

unclassified

Penerapan Fungsi Exponential Pada Pembobotan Fungsi Jarak Euclidean Algoritma K-Nearest Neighbor

Vikri

Rohmah²

2022

generation

View full text Add to dashboard Cite

– k-Nearest Neighbor (k-NN) is one of the popular classification algorithms and is widely used to solve classification cases. This is because the k-NN algorithm has advantages such as being simple, easy to explain, and easy to implement. However, the k-NN algorithm has a lack of classification results that are strongly influenced by the scale of input data and Euclidean which treats attribute data evenly, not according to the relevance of each data attribute. This causes a decrease in the classification results. One way to improve the classification accuracy performance of the k-NN algorithm is the method of weighting its features when measuring the Euclidean distance. The exponential function of the optimized Euclidean distance measurement is applied to the k-NN algorithm as a distance measurement method. Improving the performance of the k-NN method with the Exponential function for weighting features on k-NN will be carried out by experimentation using the Data Mining method. Then the results of the performance of the objective method will be compared with the original k-NN method and the previous k-NN weighting research method. As a result of the closest distance decision, taking the closest distance to k-NN will be determined with a value of k=5. After the experiment, the goal algorithm was compared with the k-NN, Wk-NN, and DWk-NN algorithms. Overall the comparison results obtained an average value of k-NN 85.87%, Wk-NN 86.98%, DWk-NN 88.19% and the k-NN algorithm given the weighting of the Exponential function obtained a value of 90.17%.

show abstract

A Review of various KNN Techniques

Cited by 11 publications

References 34 publications

Comparison of Various Data Mining Classification Techniques in the Diagnosis of Diabetic Retinopathy

Comparison of Various Data Mining Classification Techniques in the Diagnosis of Diabetic Retinopathy

An Enhanced Adaptive k-Nearest Neighbor Classifier Using Simulated Annealing

Penerapan Fungsi Exponential Pada Pembobotan Fungsi Jarak Euclidean Algoritma K-Nearest Neighbor

Contact Info

Product

Resources

About