• The Journal of the Acoustical Society of Korea
• /
• v.14 no.5
• /
• pp.5-10
• /
• 1995

In this paper we present a clustering procedure based on the use of HMM in order to get multiple statistical models which can well absorb the variants of each speaker with different ways of saying words. The HMM-clustered models obtained from the developed technique are applied to the speaker independent isolated word recognition. The HMM clustering method splits off all observation sequences with poor likelihood scores which fall below threshold from the training set and create a new model out of the observation sequences in the new cluster. Clustering is iterated by classifying each observation sequence as belonging to the cluster whose model has the maximum likelihood score. If any clutter has changed from the previous iteration the model in that cluster is reestimated by using the Baum-Welch reestimation procedure. Therefore, this method is more efficient than the conventional template-based clustering technique due to the integration capability of the clustering procedure and the parameter estimation. Experimental data show that the HMM-based clustering procedure leads to $1.43\%$ performance improvements over the conventional template-based clustering method and $2.08\%$ improvements over the single HMM method for the case of recognition of the isolated korean digits.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1987.11a
• /
• pp.70-74
• /
• 1987

본 논문은 화자독립 고립단어 인식에 있어서 LP 모델의 문제점과 그 해결 방안으로서 cepstrum 영역에 있어서 lifter를 이용한 해결에 대해서 고찰하였다. 한편, 각 인식 단어의 기준 패턴을 구하기 위한 방법으로서 집단화의 방법에 대해 논하였다. 집단화의 방법으로서는 UWA 방법과 K-iteration 방법을 변형시킨 KMA 방법을 제시 비교하였다. 인식 실험결과 정현파 lifter와 KMA의 집단화 방법을 사용하였을 때 95%의 최고 인식률을 보였다.

• Proceedings of the Korean Statistical Society Conference
• /
• 2003.10a
• /
• pp.227-232
• /
• 2003

본 논문에서는 집단화된 자료의 평균과 분산을 계산하는 새로운 방범을 제시하였다. 제시된 방법은 각 계급구간 안의 자료값들이 그 구간에 걸쳐 균등한 간격으로 분포하고 있다고 가정하고 평균과 분산을 계산하는 것이다. 개개의 자료값들이 주어진 자료와 모의실험에 의해 생성된 자료를 이용하여 제시된 방법과 기존의 방법을 비교하였다.

• Proceedings of the Korean Statistical Society Conference
• /
• 2005.05a
• /
• pp.147-154
• /
• 2005

본 논문에서는 집단화된 자료의 분위수들을 계산하는 수정된 방법을 제시하였다. 제시된 방법은 각 계급구간 안의 자료들이 그 구간에 걸쳐 균등한 간격으로, 그리고 구간의 중간점에 관하여 대칭으로 분포하고 있다고 가정하고 분위수들을 계산하는 방법이다. 개개의 자료값들이 주어진 자료를 통하여, 제시된 방법과 기존의 방법을 비교하였다.

• Journal of KIISE:Information Networking
• /
• v.27 no.4
• /
• pp.531-543
• /
• 2000

본 논문에서는 ATM PNN망에서 경로 배정을 위해 필요로하는 링크상태 정보를 효율적으로 집단화하는 방법을 제시한다. 이 방법은 집단화할 동료 그룹을 효율적으로 집단화하는 방법을 제시한다. 이방법은 집단화할 동료그룹의 경계노드들을 셔플넷의 노드들로 사상시킴으로써 표현해야 할 링크의 수를 완전 그물망 방법의 $N_2$에서 pN(p는 정수 N는 경계노드수)으로 줄인다 이는 공간 복잡도가 O(N)인 신장트리(spanning tree)방법에서 필요로 하는 링크의 수와 비슷하지만 신장 트리방법과는 달리 비대칭망(asymmetric network)에서 사용할 수 있다는 것이 큰 장접이다. 모의 실험결과 셔플넷 방법은 pNro의 링크만을 표현하면서도 상태 정보의 정확성은 완전 그물망 방법에 근접함을 알 수 있었다.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.9
• /
• pp.2353-2360
• /
• 1998

In the past. as the environment of the established soft ware system changed, most Re-Engineering perforned clustering on the basis of logical operation, In contrast, this paper proposes a method to perfonn clustering efficiently using the infonmltion sharing of each modult, of source programs that constitute the software For the clustering of related modules using the information sharing. We evaluated the result after measuring the degree of clustering using similarity and uniqueness algorithm on the basis of heuristic method of measurement. Thus, we could manipulate and achieve the clustering of related modules and procedures, This paper also prests a method to reconstruct the software system efficiently through the clustering and shows the possibility of its realization through real example.

• Journal of KIISE
• /
• v.42 no.10
• /
• pp.1254-1267
• /
• 2015

Recently, excessive installations of APs have caused WLAN interference, and many techniques have been suggested to solve this problem. The AP aggregation technique serves to reduce active APs by moving station connections to a certain AP. Since this technique forcibly moves station connections, the transmission performance of some stations may deteriorate. The AP transmit power control technique may cause station disconnection or deterioration of transmission performance when power is reduced under a certain level. The combination of these two techniques can reduce interference through AP aggregation and narrow the range of interferences further through detailed power adjustment. However, simply combining these techniques may decrease the probability of power adjustment after aggregation and increase station disconnections upon power control. As a result, improvement in performance may be insignificant. Hence, this study suggests a scheme to combine the AP aggregation and the AP transmit power control techniques in OpenFlow-based WLAN to ameliorate the disadvantages of each technique and to reduce interferences efficiently by performing aggregation for the purpose of increasing the probability of adjusting transmission power. Simulations reveal that the average transmission delay of the suggested scheme is reduced by as much as 12.8% compared to the aggregation scheme and by as much as 18.1% compared to the power control scheme. The packet loss rate due to interference is reduced by as much as 24.9% compared to the aggregation scheme and by as much as 46.7% compared to the power control scheme. In addition, the aggregation scheme and the power control scheme decrease the throughput of several stations as a side effect, but our scheme increases the total data throughput without decreasing the throughput of each station.

• The KIPS Transactions:PartC
• /
• v.18C no.5
• /
• pp.361-368
• /
• 2011

In this study, we present a method for efficiently aggregating networks state information required to determine feasible paths in transport networks that uses the source routing algorithm for hierarchical QoS routing. It is proposed to transform the full mesh topology whose Service Boundary Line serves as its logical link into the star topology. This is an aggregation method that can be used when there are two or more QoS parameters for the link to be aggregated in an asymmetric network, and it improves the information accuracy of the star topology. For this purpose, the Service Boundary Line's 3 attributes, splitting, joining and integrating, are defined in this study, and they are used to present a topology transformation method. The proposed method is similar to space complexity and time complexity of other known techniques. But simulation results showed that aggregated information accuracy and query response accuracy is more highly than that of other known method.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.81-84
• /
• 1999

본 논문에서는 중규모급 단어인식기의 실시간 구현을 위한 무감독 단어집단화 알고리듬을 제안한다. 무감독 단어집단화는 인식대상 어휘 수가 많은 대용량 음성인식 시스템에서 대상 어휘 수를 줄여주는 역할을 하는 전처리기의 성격을 갖는다. 무감독 집단화를 위해 각 단어의 유$\cdot$무성음 고유의 특성을 잘 반영할 수 있는 특징 파라미터 5개를 사용하여 패턴 인식과 회귀분석에서 널리 사용되고 있는 분류$\cdot$회귀트리(Classification And Regression Tree)에 적용시키는 방법으로 접근하였고, 각 단어의 frame 수를 일정하게 n개로 분할(segment)하여 1개의 tree를 생성시키는 방법과 각 segment에 해당하는 tree를 생성시켜 segment들 사이의 교집합 성분으로 단어들을 집단화 하였다 실험결과 탐색 대상단어 22개에서 평균2.21개로 줄어 전체 대상 단어의 $10\%$만을 탐색하여 인식할 수 있는 방법을 제시할 수 있었다.

• The KIPS Transactions:PartC
• /
• v.18C no.6
• /
• pp.441-444
• /
• 2011

In this study, we present a method for efficiently aggregating network state information. It is especially useful for aggregating links that have both delay and bandwidth in an asymmetric network. Proposed method reduces the information distortion of logical link by integration process after similar measure and grouping of logical links in multi-level topology transformation to reduce the space complexity. It is applied to transform the full mesh topology whose Service Boundary Line (SBL) serves as its logical link into a spanning tree topology. Simulation results show that aggregated information accuracy and query response accuracy are higher than that of other known method.