Efficient $$k$$ k -NN classification based on homogeneous clusters

Ougiaroglou, Stefanos; Evangelidis, Georgios

doi:10.1007/s10462-013-9411-1

Cited by 7 publications

(6 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Classifications for large datasets require high computational costs [16]. It introduced a new method namely Fast Hybrid Classification (FHC), where Two Level Data Structure were made using the k-Means algorithm, which further increased the speed to access TLDS.…”

Section: Methodsmentioning

confidence: 99%

Combination of fast hybrid classification and k value optimization in k-nn for video face recognition

Septiana

Suciati

2020

register. jurnal. ilm. teknologi. sistem. inf.

View full text Add to dashboard Cite

Nowadays, the need for face recognition is no longer include images only but also videos. However, there are some challenges associated with the addition of this new technique such as how to determine the right pre-processing, feature extraction, and classification methods to obtain excellent performance. Although nowadays the k-Nearest Neighbor (k-NN) is widely used, high computational costs due to numerous features of the dataset and large amount of training data makes adequate processing difficult. Several studies have been conducted to improve the performance of k-NN using the FHC (Fast Hybrid Classification) method by optimizing the local k values. One of the disadvantages of the FHC Method is that the k value used is still in the default form. Therefore, this research proposes the use of k-NN value optimization methods in FHC, thereby, increasing its accuracy. The Fast Hybrid Classification which combines the k-means clustering with k-NN, groups the training data into several prototypes called TLDS (Two Level Data Structure). Furthermore, two classification levels are applied to label test data, with the first used to determine the n number of prototypes with the same class in the test data. The second classification using the optimized k value in the k-NN method, is employed to sharpen the accuracy, when the same number of prototypes does not reach n. The evaluation results show that this method provides 86% accuracy and time performance of 3.3 seconds.

show abstract

Section: Methodsmentioning

confidence: 99%

Combination of fast hybrid classification and k value optimization in k-nn for video face recognition

Septiana

Suciati

2020

register. jurnal. ilm. teknologi. sistem. inf.

View full text Add to dashboard Cite

show abstract

“…On a very high level, this is similar to cluster-based NNC where class specific training data (data with same labels) is summarized as (multiple) cluster centers and used as a reduced training set on which NNC is applied -this is also known as prototype-based classification (CC), with the simplest form where there is a single cluster/prototype per class (CC1). A variety of methods exists in literature that adopt this simple idea of data reduction [9,11,24,25,27,35]. These algorithms are designed with the goal of reducing the high computational & storage requirements of NNC.…”

Section: Related Workmentioning

confidence: 99%

Fruit-fly Inspired Neighborhood Encoding for Classification

Sinha

Ram

2021

Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

Inspired by the fruit-fly olfactory circuit, the Fly Bloom Filter [4] is able to efficiently summarize the data with a single pass and has been used for novelty detection. We propose a new classifier that effectively encodes the different local neighborhoods for each class with a per-class Fly Bloom Filter. The inference on test data requires an efficient FlyHash [6] operation followed by a high-dimensional, but very sparse, dot product with the per-class Bloom Filters. On the theoretical side, we establish conditions under which the predictions of our proposed classifier agrees with the predictions of the nearest neighbor classifier. We extensively evaluate our proposed scheme with 71 data sets of varied data dimensionality to demonstrate that the predictive performance of our proposed neuroscience inspired classifier is competitive to the nearest-neighbor classifiers and other single-pass classifiers. CCS CONCEPTS• Computing methodologies → Bio-inspired approaches; Supervised learning by classification; Online learning settings.

show abstract

“…On a very high level, this is similar to cluster-based k-NNC where class specific training data (data with same labels) is summarized as (multiple) cluster centers and used as a reduced training set on which k-NNC is applied. A variety of methods exists in literature that adopt this simple idea of data reduction [Zhou et al, 2010, Parvin et al, 2012, Oigiaroglou and Evangelidis, 2013, 2016, Gallego et al, 2018, Gou et al, 2019. These algorithms are designed with the goal of reducing the high computational & storage requirements of k-NNC.…”

Section: Related Workmentioning

confidence: 99%

Neural Neighborhood Encoding for Classification

Sinha,

Ram

2020

Preprint

View full text Add to dashboard Cite

Inspired by the fruit-fly olfactory circuit, the Fly Bloom Filter [Dasgupta et al., 2018] is able to efficiently summarize the data with a single pass and has been used for novelty detection. We propose a new classifier (for binary and multi-class classification) that effectively encodes the different local neighborhoods for each class with a per-class Fly Bloom Filter. The inference on test data requires an efficient FlyHash [Dasgupta et al., 2017] operation followed by a high-dimensional, but sparse, dot product with the per-class Bloom Filters. The learning is trivially parallelizable. On the theoretical side, we establish conditions under which the prediction of our proposed classifier on any test example agrees with the prediction of the nearest neighbor classifier with high probability. We extensively evaluate our proposed scheme with over 50 data sets of varied data dimensionality to demonstrate that the predictive performance of our proposed neuroscience inspired classifier is competitive the the nearest-neighbor classifiers and other single-pass classifiers.

show abstract

Efficient $$k$$ k -NN classification based on homogeneous clusters

Cited by 7 publications

References 26 publications

Combination of fast hybrid classification and k value optimization in k-nn for video face recognition

Combination of fast hybrid classification and k value optimization in k-nn for video face recognition

Fruit-fly Inspired Neighborhood Encoding for Classification

Neural Neighborhood Encoding for Classification

Contact Info

Product

Resources

About