An Effective Method To Improve kNN Text Classifier

Hao, Xiaoyan; Tao, Xiaopeng; Zhang, Chenghong; Hu, Yunfa

doi:10.1109/snpd.2007.296

Cited by 13 publications

(7 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nonetheless, the classification problem presented here has many rare classes (see Table 1) and some experiments have shown that Precision and F1 measures may not be an adequate metrics for evaluation this kind of problem [8,13]. Thus we are going to adopt a set of more appropriated metrics for this type of problem [15].…”

Section: Evaluating the Resultsmentioning

confidence: 99%

Using a Probabilistic Neural Network for a Large Multi-label Problem

Oliveira

Ciarelli

Souza³

et al. 2008

2008 10th Brazilian Symposium on Neural Networks

View full text Add to dashboard Cite

show abstract

Section: Evaluating the Resultsmentioning

confidence: 99%

Using a Probabilistic Neural Network for a Large Multi-label Problem

Oliveira

Ciarelli

Souza³

et al. 2008

2008 10th Brazilian Symposium on Neural Networks

View full text Add to dashboard Cite

show abstract

“…When k is 1 the test entity is assignedto the class of its nearest neighbour. [38] Feature vectors andclass of the training samples are stored in the training stage ofthe algorithm while in the classification stage, k is decided bythe user. A test vector is classified by assigning it the labelwhich is most common among the k training data neighbours.…”

Section: K-nearest Neighbour Algorithm Classifiermentioning

confidence: 99%

Face and Fingerprint Fusion System for Identity Authentication Using Fusion Classifiers

M¹,

Y.S²

2018

IJCSES

View full text Add to dashboard Cite

In this work, we propose a feature level fusion and decision level fusion of face and fingerprint for designing a multimodal biometric system. Initially, Gabor and Scale Invariant Feature Transform features are extracted for both offline face and fingerprint of a person and studied the identification accuracy. Later the fusion of the biometric traits is recommended at feature level using all possible combinations of feature vectors. The possible combination of features is fed into fusion classifier of K-Nearest Neighbour(KNN), Support Vector Machine (SVM), Navie Bayes(NB) and Radial Basis Function(RBF). The best combination of feature vectors and fusion classifiers is identified for the proposed multimodal biometric system. Experiments are conducted on Face database and fingerprint database to assess the actualadvantage of the fusion of these biometric traits, in comparison to the unimodal biometric system. Experimental results reveal that fusion combination outperforms individual.

show abstract

“…The improved KNN algorithm has higher accuracy and flexibility. For example, one of the many KNN variations is based on weighted distance to improve accouracy [12]. Because our work presented here mainly deals with the comparison of the KNN algorithm and the EM algorithm in terms of classification accuracy.…”

Section: K-nearest Neighbor Algorithmmentioning

confidence: 99%