• Journal of the Institute of Convergence Signal Processing
• /
• v.4 no.4
• /
• pp.30-39
• /
• 2003

In this paper, we carried out the performance evaluation of HM-Net(Hidden Markov Network) speech recognition system for Korean speech databases. We adopted to construct acoustic models using the HM-Nets modified by HMMs(Hidden Markov Models), which are widely used as the statistical modeling methods. HM-Nets are carried out the state splitting for contextual and temporal domain by PDT-SSS(Phonetic Decision Tree-based Successive State Splitting) algorithm, which is modified the original SSS algorithm. Especially it adopted the phonetic decision tree to effectively express the context information not appear in training speech data on contextual domain state splitting. In case of temporal domain state splitting, to effectively represent information of each phoneme maintenance in the state splitting is carried out, and then the optimal model network of triphone types are constructed by in the parameter. Speech recognition was performed using the one-pass Viterbi beam search algorithm with phone-pair/word-pair grammar for phoneme/word recognition, respectively and using the multi-pass search algorithm with n-gram language models for sentence recognition. The tree-structured lexicon was used in order to decrease the number of nodes by sharing the same prefixes among words. In this paper, the performance evaluation of HM-Net speech recognition system is carried out for various recognition conditions. Through the experiments, we verified that it has very superior recognition performance compared with the previous introduced recognition system.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.3 no.1
• /
• pp.5-25
• /
• 2009

In recent years the popularity of multi-hop wireless networks has been growing. Its flexible topology and abundant routing path enables many types of applications. However, the lack of a centralized controller often makes it difficult to design a reliable service in multi-hop wireless networks. While packet routing has been the center of attention for decades, recent research focuses on data discovery such as file sharing in multi-hop wireless networks. Although there are many peer-to-peer lookup (P2P-lookup) schemes for wired networks, they have inherent limitations for multi-hop wireless networks. First, a wired P2P-lookup builds a search structure on the overlay network and disregards the underlying topology. Second, the performance guarantee often relies on specific topology models such as random graphs, which do not apply to multi-hop wireless networks. Past studies on wireless P2P-lookup either combined existing solutions with known routing algorithms or proposed tree-based routing, which is prone to traffic congestion. In this paper, we present two wireless P2P-lookup schemes that strictly build a topology-dependent structure. We first propose the Ring Interval Graph Search (RIGS) that constructs a DHT only through direct connections between the nodes. We then propose the ValleyWalk, a loosely-structured scheme that requires simple local hints for query routing. Packet-level simulations showed that RIGS can find the target with near-shortest search length and ValleyWalk can find the target with near-shortest search length when there is at least 5% object replication. We also provide an analytic bound on the search length of ValleyWalk.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10a
• /
• pp.283-286
• /
• 1996

This study suggests a new methodology for the fault diagnosis based on the signed digraph in developing the fault diagnosis system of a boiler plant. The suggested methodology uses the new model, fault-effect tree. The SDG has the advantage, which is simple and graphical to represent the causal relationship between process variables, and therefore is easy to understand. However, it cannot handle the broken path cases arisen from data uncertainty as it assumes consistent path. The FET is based on the SDG to utilize the advantages of the SDG, and also covers the above problem. The proposed FET model is constructed by clustering of measured variables, decomposing knowledge base and searching the fault propagation path from the possible faults. The search is performed automatically. The fault diagnosis system for a boiler plant, ENDS was constructed using the expert system shell G2 and the advantages of the presented method were confirmed through case studies.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.35 no.2
• /
• pp.71-88
• /
• 2010

A density-based outlier detection such as an LOF (Local Outlier Factor) tries to find an outlying observation by using density of its surrounding space. In spite of several advantages of a density-based outlier detection method, the computational complexity of outlier detection has been one of major barriers in its application. In this paper, we present an LOF algorithm that can reduce computation time of a density based outlier detection algorithm. A kd-tree indexing and approximated k-nearest neighbor search algorithm (ANN) are adopted in the proposed method. A set of experiments was conducted to examine performance of the proposed algorithm. The results show that the proposed method can effectively detect local outliers in reduced computation time.

• Journal of the Korean Society for Precision Engineering
• /
• v.17 no.2
• /
• pp.176-184
• /
• 2000

Possible representation methods for the product structure have been compared and analyzed to determine optimal disassembly path of a product. Petri net is selected as the most optimal method to represent disassembly path of the product. In this method, a reachability tree for the product is generated and disassembly time for each path is calculated. A path with the smallest disassembly time is selected as the optimal path. A software far DPN(Disassembly Petri Net) has been developed and applied to search the optimal disassembly path for a ballpoint pen disassembly process as an example.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.06c
• /
• pp.75-78
• /
• 1998

이 논문에서는 대용량 연속음성 인식에서 인식 속도를 향상시키기 위한 방법들에 대해서 연구하였다. 음성인식에 있어서 많은 양의 계산을 요하는 부분은 관측 확률의 계산과 탐색에 필요한 계산이다. 탐색에 필요한 계산을 줄이기 위하여 빔 탐색법과 phoneme look-ahead기법을 통해 탐색 공간을 줄였으며, 관측 확률을 계산하는데 소요되는 시간을 줄이기 위하여 입력 특징 벡터와 이웃 관계에 있는 가우시안 성분들만 정확한 계산을 하는 VQ에 의한 계산량 감축 방법과 tree-structured pdf 방법을 구현하였다. 3천개의 어휘와 2천여개의 트라이폰 모델로 구성된 연속 음성인식 시스템에서 보통의 Viterbi 빔 탐색법을 적용한 경우에 실시간의 2.73배의 인식 속도로 93.39%의 단어 인식률을 얻을 수 있는데 phoneme look-ahead 기법과 tree-structured pdf 방법을 추가 적용함으로써 비슷한 인식 성능에서 1.55배의 인식 속도를 얻을 수 있었다.

• 한국데이터정보과학회:학술대회논문집
• /
• 2004.04a
• /
• pp.39-50
• /
• 2004

The decision tree approach is most useful in classification problems and to divide the search space into rectangular regions. Decision tree algorithms are used extensively for data mining in many domains such as retail target marketing, fraud dection, data reduction and variable screening, category merging, etc. CHAID(Chi-square Automatic Interaction Detector) uses the chi-squired statistic to determine splitting and is an exploratory method used to study the relationship between a dependent variable and a series of predictor variables. In this paper we propose CHAID algorithm by cube-based proportional sampling and explore CHAID algorithm in view of accuracy and speed by the number of variables.

• Journal of the Korean Data and Information Science Society
• /
• v.15 no.4
• /
• pp.803-816
• /
• 2004

The decision tree approach is most useful in classification problems and to divide the search space into rectangular regions. Decision tree algorithms are used extensively for data mining in many domains such as retail target marketing, fraud dection, data reduction and variable screening, category merging, etc. CHAID uses the chi-squired statistic to determine splitting and is an exploratory method used to study the relationship between a dependent variable and a series of predictor variables. In this paper we propose CHAID algorithm by cube-based proportional sampling and explore CHAID algorithm in view of accuracy and speed by the number of variables.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2002.05a
• /
• pp.120-127
• /
• 2002

The multicasting is defined as the distribution of the same information stream from one to many nodes concurrently. There has been an intensive research effort to design protocols and construct multicast routing graphs for a single multicast group. However. there have been few researches about the relation between multiple and concurrent multicast groups. In this paper, the multiple multicast tree allocation algorithm to avoid congestion is proposed. The multicast group with different bandwidth requirement is also considered. A two-phase algorithm is proposed. The first phase is for basic search and the second phase for further improvement. The performance of the proposed algorithm is experimented with computational results. Computational results show that the proposed algorithm outperforms an existing algorithm.

• Journal of Microbiology
• /
• v.38 no.2
• /
• pp.62-65
• /
• 2000

Partial sequence comparisons of mitochondrial small subunit rDNA (mt SSU rDNA) were used to examine taxonomic and evolutionary relationships among seven Penicillium species : two monoverticillate species, two biverticillate species, and three terverticillate species. Amplified fragments of mt SSU rDNA highly varied among seven species in size, suggesting the existence of multiple insertions or deletions in the region. A phylogengtic tree was constructed by exhaustive search of parsimony analysis. The phylogenetic tree distinguished two statistically supported monophyletic groups, one for two monoverticillate species and the other for three terverticillate species and ont biverticillate species, P. vulpinum. The phylogenetic relationship of P. waksmanii, the biverticillate species, was not clear.