• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.35 no.8
• /
• pp.921-931
• /
• 2011

This paper presents a data-fitting technique in which a B-spline hypervolume is used to approximate a given data set of scattered data samples. We describe the implementation of the data structure of a B-spline hypervolume, and we measure its memory size to show that the representation is compact. The proposed technique includes two algorithms. One is for the determination of the knot vectors of a B-spline hypervolume. The other is for the control points, which are determined by solving a linear least-squares minimization problem where the solution is independent of the data-set complexity. The proposed approach is demonstrated with various data-set configurations to reveal its performance in terms of approximation accuracy, memory use, and running time. In addition, we compare our approach with existing methods and present unconstrained optimization examples to show the potential for various applications.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.8 no.3
• /
• pp.230-238
• /
• 2003

In this paper, Inverter drives adopting 2 phase space vector SRP-PWM (Separately Randomized Pulse Position PWM) with fixed switching frequency is proposed. The proposed 2 phase space vector SRP-PWM scheme is based on the 3 phase SRP-PWM. In the proposed SRP-PWM scheme, each of two phase pulses is located randomly in each switching interval. The experimental results show that the voltage and acoustic noise harmonics are spread to a wide band area. Also, the performance of the proposed 2 phase SRP-PWM and the conventional center aligned SVM are compared to each other. In result, the speed response is nearly similar to each other from the viewpoint of the v/f constant control.

• 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.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.1
• /
• pp.96-99
• /
• 1997

This paper proposes a very fast preprocessor for isolated word recognition. The proposed preprocessor has a small computational cost for extracting candidate words. In the preprocessor, we used a feature sorting algorithm instead of vector quantization to reduce the computational cost. In order to show the effectiveness of our preprocessor, we compared it to a speech recognition system based on semi-continuous hidden Markov Model and a VQ-based preprocessor by computing their recognition performances of a speaker independent isolated word recognition. For the experiments, we used the speech database consisting of 244 words which were uttered by 40 male speakers. The set of speech data uttered by 20 male speakers was used for training, and the other set for testing. As the results, the accuracy of the proposed preprocessor was 99.9% with 90% reduction rate for the speech database.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.4
• /
• pp.35-41
• /
• 2001

This paper describes Linear Discriminant Analysis and common vector extraction for speech recognition. Voice signal contains psychological and physiological properties of the speaker as well as dialect differences, acoustical environment effects, and phase differences. For these reasons, the same word spelled out by different speakers can be very different heard. This property of speech signal make it very difficult to extract common properties in the same speech class (word or phoneme). Linear algebra method like BT (Karhunen-Loeve Transformation) is generally used for common properties extraction In the speech signals, but common vector extraction which is suggested by M. Bilginer et at. is used in this paper. The method of M. Bilginer et al. extracts the optimized common vector from the speech signals used for training. And it has 100％ recognition accuracy in the trained data which is used for common vector extraction. In spite of these characteristics, the method has some drawback-we cannot use numbers of speech signal for training and the discriminant information among common vectors is not defined. This paper suggests advanced method which can reduce error rate by maximizing the discriminant information among common vectors. And novel method to normalize the size of common vector also added. The result shows improved performance of algorithm and better recognition accuracy of 2％ than conventional method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.6C
• /
• pp.634-641
• /
• 2009

The goal of OSIC-series detection methods is to approach the ML performance with feasible complexity. However, since they sometimes suffer from the empty vector problem, they can not achieve the soft-output ML performance or many candidate vectors are required to achieve the soft-output ML performance. In this paper, we propose the novel detection method, which can generate the reliable soft-outputs without suffering from empty vector problem. The proposed detector can approach the near soft-output ML performance as well as hard-output. Further, the complexity study shows that the proposed detection method has the lowest complexity compared to the other detectors having the near ML performance.

• Proceedings of The KACE
• /
• 2017.08a
• /
• pp.157-160
• /
• 2017

최근 자연 언어 처리 분야에서는 단어를 실수벡터로 임베딩하는 워드 임베딩(Word embedding) 기술이 많은 각광을 받고 있다. 최근에는 서로 다른 두 언어를 이용한 이중 언어 위드 임베딩(Bilingual word embedding) 방법을 사용하는 연구가 많이 이루어지고 있는데, 이중 언어 워드 임베딩에서 임베딩 절과의 질은 학습하는 코퍼스의 정렬방식에 따라 많은 영향을 받는다. 본 논문은 자막 병렬 코퍼스를 이용하여 밑바탕 어휘집(Seed lexicon)을 구축하여 번역 연결 강도를 향상시키고, 이중 언어 워드 임베딩의 사천(Vocabulary) 확장을 위한 언어별 연결 함수(Language-specific mapping function)을 학습하는 새로운 방식의 모델을 제안한다. 제안한 모델은 기존 모델과의 성능비교에서 비교할만한 수준의 결과를 얻었다.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.5
• /
• pp.1043-1048
• /
• 2011

Test vector reordering is a very effective way to reduce power consumption during test application. But, it is time-consuming and complicated processes, and it does not consider internal circuit structure, which may limit the effectiveness. In this paper, we order test vectors using fanout count of primary inputs that consider the internal circuit structure, which may reduce the switching activity. Then, we reorder test test vectors again by using Hamming distance between test vectors. We proposed FOVO algorithm to perform these two ideas. FOVO is an effective way to reduce power consumption during test application. The algorithm is applied to benchmark circuits and we get an average of 3.5% or more reduction of the power consumption.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.2
• /
• pp.276-283
• /
• 1994

Using a neural network for vector quantization, we can expect to have better codebook design algorithm for its adaptive process. Also, the designed codebook puts the codewords in order by its self-organizing characteristics, which makes it possible to partially search the codebook for real time process. To exploit these features of the neural network, in this paper, we propose a new codebook design algorithm that modified the KSFM(Kohonen`s Self-organizing Feature Map) and then combines the K-means algorithm. Experimental results show the performance improvment and the ability of the partical seach of the codebook for the real time process.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.945-946
• /
• 2008

Feature matching is often used for image alignment. It, however, isconsidered as motion estimation problem in case of video. In that case we need only a motion vector in an image. Then we can compute the distance between two images although the images are far away each other. So we propose affine transformation from camera motion for spatial positioning of frames and aligning those frames. The data from this method can be useful for calculating the distance, stabilizing video, photographing panorama and so on.