• Journal of Korea Multimedia Society
• /
• v.21 no.8
• /
• pp.952-959
• /
• 2018

Clustering is one of the most fundamental algorithms in machine learning. The performance of clustering is affected by the distribution of data, and when there are more data or more dimensions, the performance is degraded. For this reason, we use a stacked auto encoder, one of the deep learning algorithms, to reduce the dimension of data which generate a feature vector that best represents the input data. We use k-means, which is a famous algorithm, as a clustering. Sine the feature vector which reduced dimensions are also multi dimensional, we use the Euclidean distance as well as the cosine similarity to increase the performance which calculating the similarity between the center of the cluster and the data as a vector. A deep clustering networks combining a stacked auto encoder and k-means re-trains the networks when the k-means result changes. When re-training the networks, the loss function of the stacked auto encoder and the loss function of the k-means are combined to improve the performance and the stability of the network. Experiments of benchmark image ad document dataset empirically validated the power of the proposed algorithm.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2006.05a
• /
• pp.203-207
• /
• 2006

This paper deals with a problem of detecting BCC using confocal raman spectrum. Specially, we propose Fisher measure based filtering for rejection of frequency components being noisy or non-discriminative. we use PCA (principal component analysis) for reduction of feature space dimension. Also, we apply MAP detector for classification of BCC raman spectrum. The experimental results shows that our proposed method can reduce the feature dimension and also raise the detection ratio.

• Journal of Internet Computing and Services
• /
• v.5 no.3
• /
• pp.57-62
• /
• 2004

In this paper, an efficient speaker identification based on robust vector quantizationprincipal component analysis (VQ-PCA) is proposed to solve the problems from outliers and high dimensionality of training feature vectors in speaker identification, Firstly, the proposed method partitions the data space into several disjoint regions by roust VQ based on M-estimation. Secondly, the robust PCA is obtained from the covariance matrix in each region. Finally, our method obtains the Gaussian Mixture model (GMM) for speaker from the transformed feature vectors with reduced dimension by the robust PCA in each region, Compared to the conventional GMM with diagonal covariance matrix, under the same performance, the proposed method gives faster results with less storage and, moreover, shows robust performance to outliers.

• Journal of Korea Multimedia Society
• /
• v.16 no.7
• /
• pp.820-828
• /
• 2013

Extracting unique features from an image is a fundamental issue when making panorama images, acquiring stereo images, recognizing objects and analyzing images. Generally, the task to compare features to other images requires much computing time because some features are formed as a vector which has many elements. In this paper, we present a method that compares features after reducing the feature dimension extracted from an image using PCA(principal component analysis) and sorting the features in a linked list. SURF(speeded up robust features) is used to describe image features. When the dimension reduction method is applied, we can reduce the computing time without decreasing the matching accuracy. The proposed method is proved to be fast and robust in experiments.

• Proceedings of the KIEE Conference
• /
• 2004.07c
• /
• pp.1784-1786
• /
• 2004

PD(Partial Discharge) signal emitted from PD sources has their intrinsic features in the region of time and frequency. STFT(Short Time Fourier Transform) shows time-frequency distribution at the same time. 2-Dimensional matrices(33${\times}$77) from STFT for PD pulse signals are a good feature vectors and can be decreased in dimension by wavelet 2D data compression technique. Decreased feature vectors(13${\times}$24) were used as inputs of Back-propagation ANN(Artificial Neural Network) for discrimination of Multi-PD sources(air discharge sources(3), surface discharge(1)). They are a good feature vectors for discriminating Multi-PD sources.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.31 no.3
• /
• pp.41-51
• /
• 2006

Clustering is to group similar objects into clusters. Until now there are a lot of approaches using Self-Organizing feature Maps (SOFMS) But they have problems with a small output-layer nodes and initial weight. For example, one of them is a one-dimension map of c output-layer nodes, if they want to make c clusters. This approach has problems to classify elaboratively. This Paper suggests one-dimensional output-layer nodes in SOFMs. The number of output-layer nodes is more than those of clusters intended to find and the order of output-layer nodes is ascending in the sum of the output-layer node's weight. We un find input data in SOFMs output node and classify input data in output nodes using Euclidean distance. The proposed algorithm was tested on well-known IRIS data and TSPLIB. The results of this computational study demonstrate the superiority of the proposed algorithm.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1994.04a
• /
• pp.443-448
• /
• 1994

As a basic step of interfacing CAD and CAM, it is required to convert thedrawing data into manufacturing information automatically. In this study the developed system carries out feature recognition from the orthographic views of press mold containing various pockets. Using rhe recognized output data, 3 dimension- al model is built using ADS and AME in order to check the result of recognition. The system consists of 4 main parts, suchas, Preprocessing, Coordinate handling, Feature recognition and 3D-modeling. The system shows a good application example which can interface the design and manufacturing stage in CAD/CAM system on PC level

• Journal of Korea Multimedia Society
• /
• v.22 no.12
• /
• pp.1347-1355
• /
• 2019

Deep learning approach based on convolution neural network (CNN) has extensively studied in the field of computer vision. However, periocular feature extraction using CNN was not well studied because it is practically impossible to collect large volume of biometric data. This study uses the ResNet model which was trained with the ImageNet dataset. To overcome the problem of insufficient training data, we focused on the training of multi-layer perception (MLP) having simple structure rather than training the CNN having complex structure. It first extracts features using the pretrained ResNet model and reduces the feature dimension by principle component analysis (PCA), then trains a MLP classifier. Experimental results with the public periocular dataset UBIPr show that the proposed method is effective in person authentication using periocular region. Especially it has the advantage which can be directly applied for other biometric traits.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.54 no.7
• /
• pp.332-338
• /
• 2005

PD(Partial Discharge) signal emitted from PD sources has their intrinsic features in the region of time and frequency STFT(Short Time Fourier Transform) shows time-frequency distribution at the same time. 2-Dimensional matrices(33$\times$77) from STFT for PD pulse signals are a good feature vectors and can be decreased in dimension by wavelet 2D data compression technique. Decreased feature vectors(13$\times$24) were used as inputs of Back-propagation ANN(Artificial Neural Network) for discrimination of Multi-PD sources(air discharge sources(3), surface discharge(1)). They are a good feature vectors for discriminating Multi-PD sources in the air.

• Proceedings of the KIEE Conference
• /
• 2005.10b
• /
• pp.9-11
• /
• 2005

We propose a new content-based image retrieval using a block color co-occurrence matrix (BCCM) and pattern correlogram. In the proposed method, the color feature vectors are extracted by using BCCM that represents the probability of the co-occurrence of two mean colors within blocks. Also the pattern feature vectors are extracted by using pattern correlogram which is combined with spatial correlation of pattern. In the proposed pattern correlogram method. after block-divided image is classified into 48 patterns with respect to the change of the RGB color of the image, joint probability between the same pattern from the surrounding blocks existing at the fixed distance and the center pattern is calculated. Experimental results show that the proposed method can outperform the conventional methods as regards the precision and the size of the feature vector dimension.