• The Journal of the Acoustical Society of Korea
• /
• v.22 no.8
• /
• pp.711-716
• /
• 2003

This paper proposes the method to select the number of optimal mixtures in each state in Continuous Density HMM (Hidden Markov Models), Previously, researchers used the same number of mixture components in each state of HMM regardless spectral characteristic of speaker, To model each speaker as accurately as possible, we propose to use a different number of mixture components for each state, Selection of mixture components considered the probability value of mixture by each state that affects much parameter estimation of continuous density HMM, Also, we use PCA (principal component analysis) to reduce the correlation and obtain the system' stability when it is reduced the number of mixture components, We experiment it when the proposed method used average 10% small mixture components than the conventional HMM, When experiment result is only applied selection of mixture components, the proposed method could get the similar performance, When we used principal component analysis, the feature vector of the 16 order could get the performance decrease of average 0,35% and the 25 order performance improvement of average 0.65%.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.2
• /
• pp.243-250
• /
• 2008

This paper proposes a system which extracts necessary information from call-for-paper (CFP) documents using a hidden Markov model (HMM). Even though a CFP does not follow a strict form, there is, in general, a relatively-fixed sequence of information within most CFPs. Therefore, a hiden Markov model is adopted to analyze CFPs which has an advantage of processing consecutive data. However, when CFPs are intuitively modeled with a hidden Markov model, a problem arises that the boundaries of the information are not recognized accurately. In order to solve this problem, this paper proposes a two-phrase hidden Markov model. In the first step, the P-HMM (Phrase hidden Markov model) which models a document with phrases recognizes CFP documents locally. Then, the D-HMM (Document hidden Markov model) grasps the overall structure and information flow of the document. The experiments over 400 CFP documents grathered on Web result in 0.49 of F-score. This performance implies 0.15 of F-measure improvement over the HMM which is intuitively modeled.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.1
• /
• pp.47-53
• /
• 2004

We propose a new speech enhancement algorithm based on left-right Hidden Markov Model (HMM) with state decision using Log-likelihood Ratio Test (LRT). Since the conventional HMM-based speech enhancement methods try to improve speech quality for all states, they introduce huge computational loads inappropriate to real-time implementation. In the left-right HMM, only the current and the next state are considered for a possible state transition so to reduce the computational complexity. In this paper, we propose a method to decide the current state by using the LRT on the previous state. Experimental results show that the proposed method improves the speed up to 60% with 0.2∼0.4 dB degradation of speech quality compared to the conventional method.

• The Journal of the Korea Contents Association
• /
• v.8 no.5
• /
• pp.44-51
• /
• 2008

Pattern matching and pattern searching in time series data have been active issues in a number of disciplines. This paper suggests a novel pattern matching technology which can be used in the field of stock market analysis as well as in forecasting stock market trend. First, we define conceptual patterns, and extract data forming each pattern from given time series, and then generate learning model using Hidden Markov Model. The results show that the context-based pattern matching makes the matching more accountable and the method would be effectively used in real world applications. This is because the pattern for new data sequence carries not only the matching itself but also a given context in which the data implies.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.4
• /
• pp.31-37
• /
• 2004

Wavelet transform is a useful tool for analysis and process of image. This showed good performance in image compression and noise reduction. Wavelet coefficients can be effectively modeled by hidden Markov tree(HMT) model. However, in application of HMT model to image interpolation, training procedure is needed. Moreover, the parameters obtained from training procedure do not match input image well. In this paper, the structure of HMT is used for image interpolation, and the parameters of HMT are obtained from statistical characteristics across wavelet subbands without training procedure. In the proposed method, wavelet coefficient is modeled as Gaussian mixture model(GMM). In GMM, state transition probabilities are determined from statistical transition characteristic of coefficient across subbands, and the variance of each state is estimated using the property of exponential decay of wavelet coefficient. In simulation, the proposed method shows improvement of performance compared with conventional bicubic method and the method using HMT model with training.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.1139-1142
• /
• 2008

인터넷의 이용이 증가함에 따라 네트워크를 통한 다양한 공격 역시 증가 추세에 있다. 따라서 네트워크 이상징후를 사전에 탐지하고 상황에 따라 유연하게 대처할 수 있도록 하기 위한 연구가 절실하다. 본 연구는 은닉마코프모델을 이용해 트래픽에서 이상징후를 탐지하는 기법을 제안한다. 제안하는 기법은 시계열 예측 기법을 이용해 트래픽에서 징후를 추출한다. 징후추출 과정의 결과를 은닉마코프모델을 활용한 징후판단과정을 통해 네트워크 이상징후인지를 판단하고 결정한다. 일련의 과정을 perl로 구현하고, 실제 공격이 포함된 트래픽을 사용하여 검증한다. 하지만 결과가 확연히 증명되지는 않는데, 이는 학습과정의 부족과 실제에 가까운 트래픽의 사용으로 인해 나타나는 현상으로 연구의 본질을 흐리지는 않는다고 판단된다. 오히려 실제 상황을 가정했을 때 접근이나 적용을 판단함에 관리자의 의견을 반영할 수 있으므로 공격의 탐지와 판단에 유연성을 증대시킬 수 있다. 본 연구는 실시간 네트워크의 상황 파악이나 네트워크에서의 신종 공격 탐지 및 분류에 응용가능할 것으로 기대된다.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.10
• /
• pp.723-730
• /
• 2013

This paper presents the design of an arm gesture recognition system using Kinect sensor. A variety of methods have been proposed for gesture recognition, ranging from the use of Dynamic Time Warping(DTW) to Hidden Markov Models(HMM). Our system learns a unique HMM corresponding to each arm gesture from a set of sequential skeleton data. Whenever the same gesture is performed, the trajectory of each joint captured by Kinect sensor may much differ from the previous, depending on the length and/or the orientation of the subject's arm. In order to obtain the robust performance independent of these conditions, the proposed system executes the feature transformation, in which the feature vectors of joint positions are transformed into those of angles between joints. To improve the computational efficiency for learning and using HMMs, our system also performs the k-means clustering to get one-dimensional integer sequences as inputs for discrete HMMs from high-dimensional real-number observation vectors. The dimension reduction and discretization can help our system use HMMs efficiently to recognize gestures in real-time environments. Finally, we demonstrate the recognition performance of our system through some experiments using two different datasets.

• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06a
• /
• pp.477-479
• /
• 2011

본 논문에서는 1차원 은닉 마코프 모델을 2차원으로 확장하기 위하여 노드들의 마코프 특성이 인과적인 관계를 갖는 마코프 메쉬 모델을 이용하여 완전한 2차원 HMM의 구조를 갖는 모델을 제안한다. 마코프메쉬 모델은 이웃시스템을 통하여 이전의 시점을 정의하고, 인과적인 관계를 통하여 전이확률의 계산을 가능하게 한다. 또한 영상의 최적의 분할을 위하여 계층적 디리슐레 과정을 사전분포로 두어 고정된 상태의 수가 아닌 무한의 상태 수를 갖는 2차원 HMM을 제안한다. HDP로 정의된 사전분포와 관측된 표본 자료의 정보를 갖는 우도함수를 결합한 사후분포의 베이스 추정은 깁스샘플링 알고리즘을 이용하여 계산된다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.8A
• /
• pp.786-795
• /
• 2002

In this paper, we present a histogram and moment-based vidoe scencd change detection technique using hierarchical Hidden Markov Models(HMMs). The proposed method extracts histograms from a low-frequency subband and moments of edge components from high-frequency subbands of wavelet transformed images. Then each HMM is trained by using histogram difference and directional moment difference, respectively, extracted from manually labeled video. The video segmentation process consists of two steps. A histogram-based HMM is first used to segment the input video sequence into three categories: shot, cut, gradual scene changes. In the second stage, a moment-based HMM is used to further segment the gradual changes into a fade and a dissolve. The experimental results show that the proposed technique is more effective in partitioning video frames than the previous threshold-based methods.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.5
• /
• pp.19-28
• /
• 2009

Prediction problem of the time-series data has been a research issue for a long time among many researchers and a number of methods have been proposed in the literatures. In this paper, a method is proposed that similarities among time-series data are examined by use of Hidden Markov Model and Likelihood and future direction of the data movement is determined. Query sequence is modeled by Hidden Markov Modeling and then the model is examined over the pre-recorded time-series to find the subsequence which has the greatest similarity between the model and the extracted subsequence. The similarity is evaluated by likelihood. When the best subsequence is chosen, the next portion of the subsequence is used to predict the next phase of the data movement. A number of experiments with different parameters have been conducted to confirm the validity of the method. We used KOSPI to verify suggested method.