• The KIPS Transactions:PartB
• /
• v.8B no.6
• /
• pp.699-704
• /
• 2001

ABSTRACT In this paper, a new hybrid modular recurrent TDNN (time-delay neural network)-HMM (hidden Markov model) architecture for speech recognition has been studied. In TDNN, the recognition rate could be increased if the signal window is extended. To obtain this effect in the neural network, a high-level memory generated through a feedback within the first hidden layer of the neural network unit has been used. To increase the ability to deal with the temporal structure of phonemic features, the input layer of the network has been divided into multiple states in time sequence and has feature detector for each states. To expand the network from small recognition task to the full speech recognition system, modular construction method has been also used. Furthermore, the neural network and HMM are integrated by feeding output vectors from the neural network to HMM, and a new parameter smoothing method which can be applied to this hybrid system has been suggested.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.8
• /
• pp.565-570
• /
• 2013

Previous researches on mining sequential patterns mainly focused on discovering patterns from the point-based event. Interval events with a time interval occur in the real world that have the start and end point. Existing interval pattern mining methods that discover relationships among interval events based on the Allen operators have some problems. These are that interval patterns having three or more interval events can be interpreted as several meanings. In this paper, we propose the I_TPrefixSpan algorithm, which is an efficient sequence pattern mining technique for removing ambiguity in the Interval Patterns Mining. The proposed algorithm generates event sequences that have no ambiguity. Therefore, the size of generated candidate set can be minimized by searching sequential pattern mining entries that exist only in the event sequence. The performance evaluation shows that the proposed method is more efficient than existing methods.

• Annual Conference on Human and Language Technology
• /
• 2000.10d
• /
• pp.405-410
• /
• 2000

본 연구에서는 기능적 자기공명영상을 이용하여 한글 문장의 의미와 통사 처리에 관한 뇌의 활성화 양상을 비교함으로써 한글문장 이해의 과정에 대한 신경해부학적 증거를 찾고자 하였다. 6명의 자원자를 대상으로 문장진위판단과제를 이용하여 활성화를 유도하였다. 1.5T 초전도 자기공명영상 장치에서 EPI로 BOLD 기법을 이용하여 기능적 영상을 얻었으며 영상 후 처리는 SPM99 분석 프로그램을 이용하였다. 의미관련 통사관련 문장 모두에서 좌 우 전두회(frontal gyrus) 영역에서 활성화되었다. 의미와 통사처리 영역을 구분하기 위하여 감산법을 적용한 결과, 의미처리는 좌반구의 중측두회(middle temporal gyrus) 영역에서, 통사처리는 우반구의 하전두회(BA44) 부위에서 더 많이 활성화되었다. 의미처리에서 더 우세한 성향을 띠는 부위로 밝혀진 중측두회 영역은 의미처리시에 활성화되는 영역으로 보고하는 기존의 연구와 일치하는 결과이다. 의미와 통사 문장처리시의 뇌 활성화 양상은 뇌의 여러 영역에서 중첩되어 있기는 하지만, 특정영역에서의 차이를 보이고 있으므로, 의미와 통사처리는 다른 기전(mechanism)에 의해서 일어남을 시사해 준다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.10C
• /
• pp.951-962
• /
• 2002

A DMT transceiver recovers the sampling time from reserved sub-carriers, the pilots. Since the pilots are available after the FFT, the symbol synchronization must be done before sample synchronization. In DMT VDSL system, symbol synchronization is handled separately from sample synchronization, although the two processes are intimately related. The DMT symbol itself contains sufficient information, the cyclic extension, for symbol synchronization. Using only the sign bit of received signal, the Maximum Likelihood Estimation solution is derived. The Tx windowing in the transmitter of DMT VDSL system results in the blurring of MLE peaks. We propose the weighted summing MLE method using the sign bit which produces the clearly sharp top of MLE peaks. The stability of symbol synchronization is improved significantly by averaging over a few symbols. This paper presents the study of the original MLE and the weighted summing MLE using sign bit. A clock difference between transmitter and receiver destroys the oahogonality of the carriers. Therefore, a receiver using asynchronous sampling must perform timing correction in the discrete-time domain. We introduce an efficient digital sample synchronization method which is based on temporal and frequency domain digital signal processing.

• Journal of Korean Elementary Science Education
• /
• v.24 no.1
• /
• pp.86-101
• /
• 2005

In this study, a brain-basrd model for science teaching and learning was developed based on the natural processes which human acquire knowledge about a natural object or on event, the major domains of science educational objectives of the national curriculum, and the human brain's organizational patterns and functions. In the model, each educational objective domain is related to the brain regions as follows: The affective domain is related to the limbic system, especially amygdala of human brain which is involved in emotions, the psychomotor domain is related to the occipital lobes of human brain which perform visual processing, temporal lobes which perform functions of language generating and understandng, and parietal lobes which receive and process sensory information and execute motor activities of body, and the cognitive domain is related to the frontal and prefrontal lobes which are involved in think-ing, planning, judging, and problem solving. The model is a kind of procedural model which proceed fiom affective domain to psychomotor domain, and to cognitive domain of science educational objective system, and emphasize the order of each step and authentic assessment at each step. The model has both properties of circularity and network of activities. At classrooms, the model can be used as various forms according to subjects and student characteristics. STS themes can be appropriately covered by the model.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38B no.9
• /
• pp.765-774
• /
• 2013

In this paper, we present a cooperative compressed spectrum sensing scheme for correlated signals in decentralized wideband cognitive radio networks. Compressed sensing is a signal processing technique that can recover signals which are sampled below the Nyquist rate with high probability, and can solve the necessity of high-speed analog-to-digital converter problem for wideband spectrum sensing. In compressed sensing, one of the main issues is to design recovery algorithms which accurately recover original signals from compressed signals. In this paper, in order to achieve high recovery performance, we consider the multiple measurement vector model which has a sequence of compressed signals, and propose a cooperative sparse Bayesian recovery algorithm which models the temporal correlation of the input signals.

• The KIPS Transactions:PartC
• /
• v.16C no.2
• /
• pp.237-244
• /
• 2009

The congestion in wireless sensor network increases the ratio of data loss and causes the delay of data. The existing congestion protocols for wireless sensor network reduces the amount of transmission by control the sampling frequency of the sensor nodes related to the congestion when the congestion has occurred and was detected. However, the control method of sampling frequency is not applicable on the situation which is sensitive to the temporal data loss. In the paper, we propose a new congestion control, ACT - Adaptive Congestion conTrol. The ACT monitors the network traffic with the queue usage and detects the congestion based on the multi level threshold of queue usage. Given network congestion, the ACT increases the efficiency of network by adaptive flow control method which adjusts the frequency of packet transmission and guarantees the fairness of packet transmission between nodes. Furthermore, ACT increases the quality of data by using the variable compression method. Through experiment, we show that ACT increases the network efficiency and guarantees the fairness to sensor nodes compared with existing method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.11C
• /
• pp.1573-1580
• /
• 2004

In this paper, an adaptation mode controller for adaptive microphone array system is proposed for high-quality speech acquisition in real environments. To ensure proper adaptation of the adaptive array algorithm, the proposed adaptation mode controller uses not only temporal information, but also spatial information. The proposed adaptation mode controller is constructed with two processing stages: an initialization stage and a running stage. In the initialization stage, a sound source localization technique is adopted, and a signal correlation characteristic is used in the running stage. For the adaptive may algorithm, a generalized sidelobe canceller with an adaptive blocking matrix is used. The proposed adaptation mode controller can be used even when the adaptive blocking matrix is not adapted, and is much stable than the power ratio method. The proposed algorithm is evaluated in real environment, and simulation results show 13dB SINR improvement with the speaker sitting 2m distance from the may.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.10
• /
• pp.1327-1333
• /
• 2018

Numerical analysis has the advantage that no actual flow pathways need to be formulated, making this technique especially useful for simulation analysis such as pathway design. However, it does require that the complete physical parameters of the fluid and the complete boundary conditions be known. If any of them are unknown, either the calculation will become impossible, or even if the calculation does converge, the reliability of the result will be low. Therefore, a means of more accurate acquisition of flow information is required. In this paper, we present techniques for estimating flow field from a constraint equation for image information and velocity field, based on the image intensity changes accompanying the motion of dye in waterway. In the equation, we entered a stabilizing term to suppress estimation error. We show the effectiveness of our method through experiments with generated and real images of a Karman vortex.

• The KIPS Transactions:PartC
• /
• v.19C no.1
• /
• pp.19-28
• /
• 2012

Wireless sensor network provides services anytime and anywhere they are requested. Especially, medical sensor network based on biosensors is applied a lot to biotechnology and medical engineering. In medical sensor network, people can make their health checked at home free from temporal and spatial constraints. In ubiquitous healthcare environment, people can get instant help even in the emergency, and in hospital, patients can be taken care of efficiently. In this environment, health and life related data are delivered, and the privacy and security of personal data are very important. In this paper, we propose user authentication and data communication mechanism in two modes, normal and urgent situation using cellular phone. Through our proposal, data can be transferred in quick and secure manner.