• Journal of KIISE:Software and Applications
• /
• v.32 no.8
• /
• pp.789-794
• /
• 2005

LDA (Linear Discriminant Analysis) provides the projection that discriminates the data well, and shows a very good performance for face recognition. However, since LDA provides only one transformation matrix over whole data, it is not sufficient to discriminate the complex data consisting of many classes like honan faces. To overcome this weakness, we propose a new face recognition method, called LDA mixture model, that the set of alf classes are partitioned into several clusters and we get a transformation matrix for each cluster. This detailed representation will improve the classification performance greatly. In the simulation of face recognition, LDA mixture model outperforms PCA, LDA, and PCA mixture model in terms of classification performance.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.16 no.6
• /
• pp.759-765
• /
• 2006

In this paper, it is proposed an effective face expression recognition LDA mixed mode using a triangularity membership fuzzy function and wavelet basis. The proposal algorithm gets performs the optimal image, fuzzy wavelet algorithm and Expression recognition is consisted of face characteristic detection step and face Expression recognition step. This paper could applied to the PCA and LDA in using some simple strategies and also compares and analyzes the performance of the LDA mixed model which is combined and the facial expression recognition based on PCA and LDA. The LDA mixed model is represented by the PCA and the LDA approaches. And then we calculate the distance of vectors dPCA, dLDA from all fates in the database. Last, the two vectors are combined according to a given combination rule and the final decision is made by NNPC. In a result, we could showed the superior the LDA mixed model can be than the conventional algorithm.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.7
• /
• pp.17-22
• /
• 1999

In this paper we propose a new mixed method of LDA and tri-tone model to predict Korean prosodic break indices(PBI) for a given utterance. PBI can be used as an important cue of syntactic discontinuity in continuous speech recognition(CSR). The model consists of three steps. At the first step, PBI was predicted with the information of syllable and pause duration through the linear discriminant analysis (LDA) method. At the second step, syllable tone information was used to estimate PBI. In this step we used vector quantization (VQ) for coding the syllable tones and PBI is estimated by tri-tone model. In the last step, two PBI predictors were integrated by a weight factor. The proposed method was tested on 200 literal style spoken sentences. The experimental results showed 72% accuracy.

• Journal of KIISE:Software and Applications
• /
• v.32 no.8
• /
• pp.781-788
• /
• 2005

LDA (Linear Discriminant Analysis) is a data discrimination technique that seeks transformation to maximize the ratio of the between-class scatter and the within-class scatter While it has been successfully applied to several applications, it has two limitations, both concerning the underfitting problem. First, it fails to discriminate data with complex distributions since all data in each class are assumed to be distributed in the Gaussian manner; and second, it can lose class-wise information, since it produces only one transformation over the entire range of classes. We propose three extensions of LDA to overcome the above problems. The first extension overcomes the first problem by modeling the within-class scatter using a PCA mixture model that can represent more complex distribution. The second extension overcomes the second problem by taking different transformation for each class in order to provide class-wise features. The third extension combines these two modifications by representing each class in terms of the PCA mixture model and taking different transformation for each mixture component. It is shown that all our proposed extensions of LDA outperform LDA concerning classification errors for handwritten digit recognition and alphabet recognition.

• ETRI Journal
• /
• v.38 no.3
• /
• pp.487-493
• /
• 2016

Two new methods are proposed for an unsupervised adaptation of a language model (LM) with a single sentence for automatic transcription tasks. At the training phase, training documents are clustered by a method known as Latent Dirichlet allocation (LDA), and then a domain-specific LM is trained for each cluster. At the test phase, an adapted LM is presented as a linear mixture of the now trained domain-specific LMs. Unlike previous adaptation methods, the proposed methods fully utilize a trained LDA model for the estimation of weight values, which are then to be assigned to the now trained domain-specific LMs; therefore, the clustering and weight-estimation algorithms of the trained LDA model are reliable. For the continuous speech recognition benchmark tests, the proposed methods outperform other unsupervised LM adaptation methods based on latent semantic analysis, non-negative matrix factorization, and LDA with n-gram counting.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.2
• /
• pp.131-136
• /
• 2020

Wake Up Word (WUW) is a short utterance used to convert speech recognizer to recognition mode. The WUW defined by the user who actually use the speech recognizer is called user-defined WUW. In this paper, to recognize user-defined WUW, we construct traditional Gaussian Mixture Model-Hidden Markov Model (GMM-HMM), Linear Discriminant Analysis (LDA)-GMM-HMM and LDA-Deep Neural Network (DNN)-HMM based system and compare their performances. Also, to improve recognition accuracy of the WUW system, a threshold method is applied to each model, which significantly reduces the error rate of the WUW recognition and the rejection failure rate of non-WUW simultaneously. For LDA-DNN-HMM system, when the WUW error rate is 9.84 %, the rejection failure rate of non-WUW is 0.0058 %, which is about 4.82 times lower than the LDA-GMM-HMM system. These results demonstrate that LDA-DNN-HMM model developed in this paper proves to be highly effective for constructing user-defined WUW recognition system.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2006.11a
• /
• pp.167-170
• /
• 2006

본 논문은 선형분류기인 LDA 융합모델과 최소거리패턴분류법을 이용한 얼굴표정인식 알고리즘 연구에 관한 것이다. 제안된 알고리즘은 얼굴 표정을 인식하기 위해 두 단계의 특징 추출과정과 인식단계를 거치게 된다. 먼저 특징추출 단계에서는 얼굴 표정이 담긴 영상을 PCA를 이용해 고차원에서 저차원의 공간으로 변환한 후, LDA 이용해 특징벡터를 클래스 별로 나누어 분류한다. 다음 단계로 LDA융합모델을 통해 계산된 특징벡터에 최소거리패턴분류법을 적용함으로서 얼굴 표정을 인식한다. 제안된 알고리즘은 6가지 기본 감정(기쁨, 화남, 놀람, 공포, 슬픔, 혐오)으로 구성된 데이터베이스를 이용해 실험한 결과, 기존알고리즘에 비해 향상된 인식률과 특정 표정에 관계없이 고른 인식률을 보임을 확인하였다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2020.07a
• /
• pp.385-386
• /
• 2020

In this paper, we propose a collaborative filtering recommendation algorithm based on the LDA2Vec topic model. By extracting and analyzing the article's content, calculate their semantic similarity then combine the traditional collaborative filtering algorithm to recommend. This approach may promote the system's recommend accuracy.

• Journal of Korean Library and Information Science Society
• /
• v.52 no.1
• /
• pp.79-108
• /
• 2021

Information retrieval in the health field has several challenges. Health information terminology is difficult for consumers (laypeople) to understand. Formulating a query with professional terms is not easy for consumers because health-related terms are more familiar to health professionals. If health terms related to a query are automatically added, it would help consumers to find relevant information. The proposed query expansion (QE) models show how to expand a query using MeSH terms. The documents were represented by MeSH terms (i.e. Bag-of-MeSH), found in the full-text articles. And then the MeSH terms were used to generate LDA (Latent Dirichlet Analysis) topic models. A query and the top k retrieved documents were used to find MeSH terms as topic words related to the query. LDA topic words were filtered by threshold values of topic probability (TP) and word probability (WP). Threshold values were effective in an LDA model with a specific number of topics to increase IR performance in terms of infAP (inferred Average Precision) and infNDCG (inferred Normalized Discounted Cumulative Gain), which are common IR metrics for large data collections with incomplete judgments. The top k words were chosen by the word score based on (TP *WP) and retrieved document ranking in an LDA model with specific thresholds. The QE model with specific thresholds for TP and WP showed improved mean infAP and infNDCG scores in an LDA model, comparing with the baseline result.

• Journal of the Korean Society for Library and Information Science
• /
• v.58 no.1
• /
• pp.5-30
• /
• 2024

The purpose of this study is to extract topics from experimental data using the topic modeling methods(LDA, Top2Vec, and BERTopic) and compare the characteristics and differences between these models. The experimental data consist of 55,442 papers published in 85 academic journals in the field of library and information science, which are indexed in the Web of Science(WoS). The experimental process was as follows: The first topic modeling results were obtained using the default parameters for each model, and the second topic modeling results were obtained by setting the same optimal number of topics for each model. In the first stage of topic modeling, LDA, Top2Vec, and BERTopic models generated significantly different numbers of topics(100, 350, and 550, respectively). Top2Vec and BERTopic models seemed to divide the topics approximately three to five times more finely than the LDA model. There were substantial differences among the models in terms of the average and standard deviation of documents per topic. The LDA model assigned many documents to a relatively small number of topics, while the BERTopic model showed the opposite trend. In the second stage of topic modeling, generating the same 25 topics for all models, the Top2Vec model tended to assign more documents on average per topic and showed small deviations between topics, resulting in even distribution of the 25 topics. When comparing the creation of similar topics between models, LDA and Top2Vec models generated 18 similar topics(72%) out of 25. This high percentage suggests that the Top2Vec model is more similar to the LDA model. For a more comprehensive comparison analysis, expert evaluation is necessary to determine whether the documents assigned to each topic in the topic modeling results are thematically accurate.