• MALSORI
• /
• no.50
• /
• pp.51-69
• /
• 2004

In this paper, we present a one-pass semi-dynamic network decoding framework that inherits both advantages of fast decoding speed from static network decoders and memory efficiency from dynamic network decoders. Our method is based on the novel language model network representation that is essentially of finite state machine (FSM). The static network derived from the language model network [1][2] is partitioned into smaller subnetworks which are static by nature or self-structured. The whole network is dynamically managed so that those subnetworks required for decoding are cached in memory. The network is near-minimized by applying the tail-sharing algorithm. Our decoder is evaluated on the 25k-word Korean broadcast news transcription task. In case of the search network itself, the network is reduced by 73.4% from the tail-sharing algorithm. Compared with the equivalent static network decoder, the semi-dynamic network decoder has increased at most 6% in decoding time while it can be flexibly adapted to the various memory configurations, giving the minimal usage of 37.6% of the complete network size.

• Structural Engineering and Mechanics
• /
• v.75 no.3
• /
• pp.295-309
• /
• 2020

The construction of Reinforced Concrete (R/C) buildings with unreinforced masonry infills is part of the traditional building practice in many countries with regions of high seismicity throughout the world. When these buildings are subjected to seismic motions the presence of masonry infills and especially their configuration can highly influence the seismic damage state. The capability to avoid configurations of masonry infills prone to seismic damage at the stage of initial architectural concept would be significantly definitive in the context of Performance-Based Earthquake Engineering. Along these lines, the present paper investigates the potential of instant prediction of the damage response of R/C buildings with various configurations of masonry infills utilizing Artificial Neural Networks (ANNs). To this end, Multilayer Feedforward Perceptron networks are utilized and the problem is formulated as pattern recognition problem. The ANNs' training data-set is created by means of Nonlinear Time History Analyses of 5 R/C buildings with a large number of different masonry infills' distributions, which are subjected to 65 earthquakes. The structural damage is expressed in terms of the Maximum Interstorey Drift Ratio. The most significant conclusion which is extracted is that the ANNs can reliably estimate the influence of masonry infills' configurations on the seismic damage level of R/C buildings incorporating their optimum design.

• Human Ecology Research
• /
• v.55 no.1
• /
• pp.45-55
• /
• 2017

This study analyzes diverse factors in time poverty and quality of life in dual-earner families with preschool children that pertain to the individual, family, and occupation. Data were taken from the 17th edition of the Korean Labor and Income Panel Study developed by the Korea Labor Institute in 2014. The sample consists of 826 households who are dual-earner families with preschool children. The major findings are as follows. First, this study identified inadequacies in personal care time for dual-earner families with preschool children. Second, the results show that gender, recognition of gender role, and overall satisfaction of occupation are related to the time poverty of dual-earner families. Men are more likely to experience time poverty than women, and equal recognition of gender role and satisfaction of occupation indicate a negative relation on the time poverty of dual-earner families with preschool children. Last, quality of life in non-time-poor groups is higher than for groups who experience time poverty. Also, health state, earned income, work-family life conflict, and overall satisfaction of occupation are commonly related to quality of life in both groups. The results suggest implications for comprehensive policies to address family time issues.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.2 no.2
• /
• pp.122-126
• /
• 2002

In this study, we find features of 3 emotions (Happiness, Angry, Surprise) as the fundamental research of emotion recognition. Speech signal with emotion has several elements. That is, voice quality, pitch, formant, speech speed, etc. Until now, most researchers have used the change of pitch or Short-time average power envelope or Mel based speech power coefficients. Of course, pitch is very efficient and informative feature. Thus we used it in this study. As pitch is very sensitive to a delicate emotion, it changes easily whenever a man is at different emotional state. Therefore, we can find the pitch is changed steeply or changed with gentle slope or not changed. And, this paper extracts formant features from speech signal with emotion. Each vowels show that each formant has similar position without big difference. Based on this fact, in the pleasure case, we extract features of laughter. And, with that, we separate laughing for easy work. Also, we find those far the angry and surprise.

• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2001.05a
• /
• pp.225-230
• /
• 2001

Intensity of background noise is a factor significantly affecting both subjective evaluation of experienced stress level and associated electroencephalographic (EEG) responses during mental load in noisy environments. In the study on 27 subjects we analyzed the influence of the background white noise (WN) intensity on psychophysiological responses during a word recognition test. Electrocortical activity were recorded during baseline resting state and 40 s long performance on 3 similar Korean word recognition tests with different intensities of background WN (55, 70 and 85 dB).. An important finding in terms of physiological reactivity was similarity of all physiological response profiles between 55 and 70dB WN, i.e., none of physiological variables differentiated the two conditions, while 85dB WN resulted in a significantly different profile of reactions (higher fast beta power in EEG spectra). This condition was characterized by highest subjective rating of experienced stress, had more fast beta activity and had tendency of right hemisphere dominance, emphasizing the role of brain lateralization in negative affect control.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.5 no.7
• /
• pp.1274-1292
• /
• 2011

Due to the variety of household electric devices and different power consumption habits of consumers at present, general home energy management (HEM) systems suffer from the lack of dynamic identification of various household appliances and a unidirectional information display. This study presented a set of intelligent interconnection network systems for electric appliances, which can measure the power consumption of household appliances through a current sensing device based on OSGi platform. The system establishes the characteristics and categories of related electric appliances, and searches the corresponding cluster data and eliminates noise for recognition functionality and error detection mechanism of electric appliances by applying the clustering algorithm. The system also integrates household appliance control network services so as to control them according to users' power consumption plans or through mobile devices, thus realizing a bidirectional monitoring service. When the system detects an abnormal operating state, it can automatically shut off electric appliances to avoid accidents. In practical tests, the system reached a recognition rate of 95%, and could successfully control general household appliances through the ZigBee network.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.21-26
• /
• 2009

A sensor system capable of automatically recognize activities would allow many potential Ubiquitous applications. This paper presents a new system for recognizing the activities of daily living(ADL) like walking, running, standing, sitting, lying etc. The system based on the state-dependent motion analysis using Tri-Accelerometer and Zigbee tag. Two accelerometers are used for the classification of body and hand activities. Classification of the environment and instrumental activities is performed based on the hand interaction with an object ID using.

• Journal of Society of Preventive Korean Medicine
• /
• v.17 no.2
• /
• pp.89-103
• /
• 2013

Objectives: This study was aimed to contribute to the establishment of base for the development of new health technology in Korean Medicine. Methods: Survey was performed with 200 samples obtained through stratified sampling from the list of members of Association of Korean Medicine. Results: For the question about the recognition of new health technology, 54.0% answered 'yes' and 45.0% answered 'no', For the question about whether using the therapy not listed in the medical care of national health insurance, 43.5% answered 'use', Conclusion: Doctors of Korean Medicine seem to want the enlargement of new health technology in the Korean Medicine.

• Journal of the Korea Computer Industry Society
• /
• v.4 no.4
• /
• pp.647-660
• /
• 2003

The present of state is situation that is realized by necessity of maintenance of public security about great many information is real condition been increasing continually in knowledge info-age been situating in wide field of national defense, public peace, banking, politics, education etc. Also, loss or forgetfulness, and peculation by ID for individual information and number increase of password in Internet called that is sea of information is resulting various social problem. By alternative about these problem, including Biometrics, several authentication systems through sign(Signature), Smart Card, Watermarking technology are developed. Therefore, This paper shows that extract factor that efficiency can get into peculiar feature in physical features for good fingerprint recognition algorithm implementation with old study finding that take advantage of special quality of these fingerprint.

• Proceedings of the KIEE Conference
• /
• 1999.07b
• /
• pp.546-549
• /
• 1999

For the safty driving of an automobile which is become individual requisites, a new Neural Network algorithm which recognized the load vehicles in real time is proposed. The proposed neural network technique is the real time computation method through the inter-node diffusion. In the network, a node corresponds to a state in the quantized input space. Each node is composed of a processing unit and fixed weights from its neighbor nodes as well as its input terminal. The most reliable algorithm derived for real time recognition of vehicles, is a dynamic programming based algorithm based on sequence matching techniques that would process the data as it arrives and could therefore provide continuously updated neighbor information estimates. Through several simulation experiments, real time reconstruction of the nonlinear image information is processed 1-D LIPN hardware has been composed and various experiments with static and dynamic signals have been implmented.