• The Journal of the Acoustical Society of Korea
• /
• v.39 no.6
• /
• pp.615-621
• /
• 2020

In this paper, we propose a multi-site based earthquake event classification method using graph convolution networks. In the traditional earthquake event classification methods using deep learning, they used single-site observation to estimate seismic event class. However, to achieve robust and accurate earthquake event classification on the seismic observation network, the method using the information from the multi-site observations is needed, instead of using only single-site data. Firstly, our proposed model employs convolution neural networks to extract informative embedding features from the single-site observation. Secondly, graph convolution networks are used to integrate the features from several stations. To evaluate our model, we explore the model structure and the number of stations for ablation study. Finally, our multi-site based model outperforms up to 10 % accuracy and event recall rate compared to single-site based model.

• Progress in Medical Physics
• /
• v.27 no.3
• /
• pp.131-138
• /
• 2016

We analyzed the terminology and classification related to the risk management of radiation treatment overseas to establish the terminology and classification system for Korea. This study investigated the terminology and classification for radiotherapy risk management through overseas research materials from related organizations and associations, including the IAEA, WHO, British group, EC, and AAPM. Overseas risk management commonly uses the terms "near miss", "incident", and "adverse event", classified according to the degree of severity. However, several organizations have ambiguous terminologies. They use the term "near miss" for events such as a near event, close call, and good catch; the term "incident" for an event; and the term "adverse event" for the likes of an accident and an event. In addition, different organizations use different classifications: a "near miss" is generally classified as "incident" in most cases but not classified as such in BIR et al. Confusion might also be caused by the disunity of the terminology and classification, and by the ambiguity of definitions. Patient safety management of medical institutions in Korea uses the terms "near miss", "adverse event", and "sentinel event", which it classifies into eight levels according to the severity of risk to the patient. Therefore, the terminology and classification for radiotherapy risk management based on the patient safety management of medical institutions in Korea will help in improving the safety and quality of radiotherapy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.6
• /
• pp.2748-2760
• /
• 2018

This paper proposes an audio event classification method based on convolutional neural networks (CNNs). CNN has great advantages of distinguishing complex shapes of image. Proposed system uses the features of audio sound as an input image of CNN. Mel scale filter bank features are extracted from each frame, then the features are concatenated over 40 consecutive frames and as a result, the concatenated frames are regarded as an input image. The output layer of CNN generates probabilities of audio event (e.g. dogs bark, siren, forest). The event probabilities for all images in an audio segment are accumulated, then the audio event having the highest accumulated probability is determined to be the classification result. This proposed method classified thirty audio events with the accuracy of 81.5% for the UrbanSound8K, BBC Sound FX, DCASE2016, and FREESOUND dataset.

• Phonetics and Speech Sciences
• /
• v.7 no.4
• /
• pp.27-33
• /
• 2015

This paper proposes an audio event classification method using Deep Neural Networks (DNN). The proposed method applies Feed Forward Neural Network (FFNN) to generate event probabilities of ten audio events (dog barks, engine idling, and so on) for each frame. For each frame, mel scale filter bank features of its consecutive frames are used as the input vector of the FFNN. These event probabilities are accumulated for the events and the classification result is determined as the event with the highest accumulated probability. For the same dataset, the best accuracy of previous studies was reported as about 70% when the Support Vector Machine (SVM) was applied. The best accuracy of the proposed method achieves as 79.23% for the UrbanSound8K dataset when 80 mel scale filter bank features each from 7 consecutive frames (in total 560) were implemented as the input vector for the FFNN with two hidden layers and 2,000 neurons per hidden layer. In this configuration, the rectified linear unit was suggested as its activation function.

• Journal of the Korean Dietetic Association
• /
• v.12 no.3
• /
• pp.235-242
• /
• 2006

The purpose of this study of was to provide basic data for preparing event menus to increase customer's satisfaction by investigating university students' participation and preference for the event menus in the foodservice operations. The questionnaires were distributed to 300 customers from August 1 to 31, 2005. 88.0% of the questionnaires were analyzed. Statistical analysis of data was performed using SAS package program(version 8.2) for descriptive analysis and $χ^2$-test, t-test, one-way ANOVA, Duncan multiple range test. The results of this study can be summarized as follows : 50.4% of the students have participated in foodservice operation's event and the average degree of the satisfaction was 2.67 out of 5. The type of the events customers have most frequently participated in were the national holiday·subdivisions of the season event(47.3%), the day event(34.1%), environment event(26.9%) and so on. In large classification, preferred were season event(85.2%), international food event(76.9%), and healthy food event(73.1%) and so on. In small classification, orgarnic food event(53.0%), summer fruits festival(41.3%), midsummer event(36.6%) and christmas event(34.4%) and so on. From now on, the event reflecting customers' expectation and requirement should be planned and implemented.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.6
• /
• pp.922-934
• /
• 2015

In this paper, The classification between precipitation echo(PRE) and non-precipitation echo(N-PRE) (including ground echo and clear echo) is carried out from weather radar data using neuro-fuzzy algorithm. In order to classify between PRE and N-PRE, Input variables are built up through characteristic analysis of radar data. First, the event classifier as the first classification step is designed to classify precipitation event and non-precipitation event using input variables of RBFNNs such as DZ, DZ of Frequency(DZ_FR), SDZ, SDZ of Frequency(SDZ_FR), VGZ, VGZ of Frequency(VGZ_FR). After the event classification, in the precipitation event including non-precipitation echo, the non-precipitation echo is completely removed by the echo classifier of the second classifier step that is built as Type-2 FCM based RBFNNs. Also, parameters of classification system are acquired for effective performance using PSO(Particle Swarm Optimization). The performance results of the proposed echo classifier are compared with CZ. In the sequel, the proposed model architectures which use event classifier as well as the echo classifier of Interval Type-2 FCM based RBFNN show the superiority of output performance when compared with the conventional echo classifier based on RBFNN.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.130-138
• /
• 2021

This paper presents an effective structure by applying various normalization to Convolutional Neural Networks (CNN) for seismic event classification. Normalization techniques can not only improve the learning speed of neural networks, but also show robustness to noise. In this paper, we analyze the effect of input data normalization and hidden layer normalization on the deep learning model for seismic event classification. In addition an effective model is derived through various experiments according to the structure of the applied hidden layer. As a result of various experiments, the model that applied input data normalization and weight normalization to the first hidden layer showed the most stable performance improvement.

• Proceedings of the IEEK Conference
• /
• 2000.06e
• /
• pp.127-130
• /
• 2000

Classification of Electroencephalogram(EEG) makes one of key roles in the field of clinical diagnosis, such as detection for epilepsy. Spectrum analysis using the fourier transform(FT) uses the same window to signals, so classification rate decreases for nonstationary signals such as EEG's. In this paper, wavelet power spectrum method using wavelet transform which is excellent in detection of transient components of time-varying signals is applied to the classification of three types of Event Related Potential(EP) and compared with the result by fourier transform. In the experiments, two types of photic stimulation, which are caused by eye opening/closing and artificial light, are used to collect the data to be classified. After choosing a specific range of scales, scale-averaged wavelet spectrums extracted from the wavelet power spectrum is used to find features by Back-Propagation(13P) algorithm. As a result, wavelet analysis shows superiority to fourier transform for nonstationary EEG signal classification.

• International journal of advanced smart convergence
• /
• v.6 no.4
• /
• pp.41-49
• /
• 2017

In this paper I applies Borer (1993)'s way of classifying English intransitive action verbs such as 'run', walk, among many others, to the corresponding Korean intransitive action verbs such as 'tali-ta' and 'keət-ta', and show how they are different from - or similar with - each other in terms of syntactic structures and verb classification. Unlike the English verb 'run' which can be classified into an unaccusative verb as well as an unergative verb in Borer's theory, the corresponding Korean verbs 'tali-ta' or 't'wi-ta' can behave not only as an unergative and unauucsative verb, but also it can behave as a transitive verb. Though Borer's perspective on classification of verb types may be thought of as somewhat radical mostly due to its heavy dependency on aspectual representation of a whole sentence which a verb is just part of, it is clearly suggesting a new and great insight into the controversial topic of classification of verb types. So it is worth adopting this insightful perspective for the analysis of corresponding Korean verbs and seeing if it also works for the Korean ones.

• Proceedings of the KSR Conference
• /
• 2008.06a
• /
• pp.1518-1523
• /
• 2008

Train collision and derailment are types of accident with low probability of occurrence, but they could lead to disastrous consequences including loss of lives and properties. The development of the risk assessment model has been called upon to predict and assess the risk for a long time. Nevertheless, the risk assessment model is recently introduced to the railway system in Korea. The classification of the hazardous events and causes is the commencement of the risk assessment model. In previous researches related to the classification, the hazardous events and causes were classified by centering the results. That classification was simple, but might not show the root cause of the hazardous events. This study has classified the train collision and derailment based on the relevant hazardous event including faults of the train related the accidents, and investigates the causes related to the hazardous events. For the risk assessment model, FTA (fault tree analysis) and ETA (event tree analysis) methods are introduced to assess the risk.