A property of learning chunk data using incremental kernel principal component analysis

Tokumoto, Takaomi; Ozawa, Seiichi

doi:10.1109/eais.2012.6232796

Cited by 3 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As mentioned before, Takaomi et al [10] found that more time is needed for the large chunk data. Therefore, the objective of this experiment is to measure the influence of the data division and the data selection when a chunk of data is received.…”

Section: Efficiency Of Ikpca With Different Number Of Chunk Sizesmentioning

confidence: 83%

“…However, large computation and memory costs are needed to obtain an accumulation ratio if a chunk size is large. Takaomi et al [10] investigated the influence of chunk size to the learning time. They found that more time is required for the large chunk data unless the large chunk data are divided into small chunks.…”

Section: A Data Selectionmentioning

confidence: 99%

“…However, when a large chunk of data is given, the computational time of the independent data selection could dominate over those of the eigenvalue decomposition. Hence, the large computation and memory costs still remained an issue for the large chunk data [10].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A fast Incremental Kernel Principal Component Analysis for data streams

Joseph

Ozawa

2014

2014 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

Kernel Principal Component Analysis (KPCA) is widely used feature extraction as it have been proven that KPCA is powerful in many areas in pattern recognition. Considering that the conventional KPCA should decompose a kernel matrix of all training data, this would be an unrealistic assumption for data streams in real-world applications. Therefore, in this paper, we propose an online feature extraction called Chunk Incremental Kernel Principal Component Analysis (CIKPCA) that can handle data streams in an incremental mode. In the proposed method, the training data are assumed to be given in a chunk of multiple data at one time. In CIKPCA, an eigen-feature space is updated by solving the eigenvalue decomposition once whenever a chunk of data is given. However, if a chunk size is large, a kernel matrix to be decomposed is also large, resulting in high computational time. Considering that not all the data are useful for the eigen-feature space learning, the data in a chunk are first selected based on the importance. Several benchmark data sets in the UCI Machine Learning Repository are used to evaluate the performance of the proposed method. The experimental results show that our proposed method can accelerate the learning of the eigenfeature space compared to Takeuchi et al.'s IKPCA without reducing the recognition accuracy.

show abstract

Section: Efficiency Of Ikpca With Different Number Of Chunk Sizesmentioning

confidence: 83%

Section: A Data Selectionmentioning

confidence: 99%

See 1 more Smart Citation

A fast Incremental Kernel Principal Component Analysis for data streams

Joseph

Ozawa

2014

2014 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Up to now, a large number of incremental learning algorithms are investigated extensively [3][4][5][6][7][8]. However, it is interesting to note that most incremental learning algorithms adopt vector model, that is, these algorithms always convert high-dimensional samples into vectors.…”

Section: Introductionmentioning

confidence: 99%

Incremental Tensor Principal Component Analysis for Image Recognition

Zhao

Liu

Yan

2013

AMR

View full text Add to dashboard Cite

Aiming at the disadvantages of the traditional off-line vector-based learning algorithm, this paper proposes a kind of Incremental Tensor Principal Component Analysis (ITPCA) algorithm. It represents an image as a tensor data and processes incremental principal component analysis learning based on update-SVD technique. On the one hand, the proposed algorithm is helpful to preserve the structure information of the image. On the other hand, it solves the training problem for new samples. The experiments on handwritten numeral recognition have demonstrated that the algorithm has achieved better performance than traditional vector-based Incremental Principal Component Analysis (IPCA) and Multi-linear Principal Component Analysis (MPCA) algorithms.

show abstract