• Journal of Information Technology Applications and Management
• /
• v.23 no.3
• /
• pp.87-97
• /
• 2016

This paper proposes a fast sound source localization method using a TDOA sign-based algorithm. We present an L-shaped microphone set-up which creates four major regions in the range of $0^{\circ}{\sim}360^{\circ}$ by the intersection of the positive and negative regions of the individual microphone pairs. Then, we make an initial source region prediction based on the signs of two TDOA estimates before computing the azimuth value. Also, we apply a threshold and angle comparison to tackle the existing front-back confusion problem. Our experimental results show that the proposed method is comparable in accuracy to previous three microphone array methods; however, it takes a shorter computation time because we compute only two TDOA values.

• Journal of Biomedical Engineering Research
• /
• v.44 no.2
• /
• pp.147-156
• /
• 2023

Efforts to employ smart home sensors to monitor the indoor activities of elderly single residents have been made to assess the feasibility of a safe and healthy lifestyle. However, the bathroom remains an area of blind spot. In this study, we have developed and evaluated a new edge computer device that can automatically detect water usage activities in the bathroom and record the activity log on a cloud server. Three kinds of sound as flushing, showering, and washing using wash basin generated during water usage were recorded and cut into 1-second scenes. These sound clips were then converted into a 2-dimensional image using MEL-spectrogram. Sound data augmentation techniques were adopted to obtain better learning effect from smaller number of data sets. These techniques, some of which are applied in time domain and others in frequency domain, increased the number of training data set by 30 times. A deep learning model, called CRNN, combining Convolutional Neural Network and Recurrent Neural Network was employed. The edge device was implemented using Raspberry Pi 4 and was equipped with a condenser microphone and amplifier to run the pre-trained model in real-time. The detected activities were recorded as text-based activity logs on a Firebase server. Performance was evaluated in two bathrooms for the three water usage activities, resulting in an accuracy of 96.1% and 88.2%, and F1 Score of 96.1% and 87.8%, respectively. Most of the classification errors were observed in the water sound from washing. In conclusion, this system demonstrates the potential for use in recording the activities as a lifelog of elderly single residents to a cloud server over the long-term.

• Asian Pacific Journal of Cancer Prevention
• /
• v.17 no.2
• /
• pp.835-838
• /
• 2016

Background: Lung cancer remains one of the most common cancers in the world, both in terms of new cases (about 13% of total per year) and deaths (nearly one cancer death in five), because of the high case fatality. Errors in lung cancer type or malignant growth determination lead to degraded treatment efficacy, because anticancer strategy depends on tumor morphology. Materials and Methods: We have made an attempt to evaluate effectiveness of machine learning algorithms in the task of lung cancer classification based on gene expression levels. We processed four publicly available data sets. The Dana-Farber Cancer Institute data set contains 203 samples and the task was to classify four cancer types and sound tissue samples. With the University of Michigan data set of 96 samples, the task was to execute a binary classification of adenocarcinoma and non-neoplastic tissues. The University of Toronto data set contains 39 samples and the task was to detect recurrence, while with the Brigham and Women's Hospital data set of 181 samples it was to make a binary classification of malignant pleural mesothelioma and adenocarcinoma. We used the k-nearest neighbor algorithm (k=1, k=5, k=10), naive Bayes classifier with assumption of both a normal distribution of attributes and a distribution through histograms, support vector machine and C4.5 decision tree. Effectiveness of machine learning algorithms was evaluated with the Matthews correlation coefficient. Results: The support vector machine method showed best results among data sets from the Dana-Farber Cancer Institute and Brigham and Women's Hospital. All algorithms with the exception of the C4.5 decision tree showed maximum potential effectiveness in the University of Michigan data set. However, the C4.5 decision tree showed best results for the University of Toronto data set. Conclusions: Machine learning algorithms can be used for lung cancer morphology classification and similar tasks based on gene expression level evaluation.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.5
• /
• pp.409-415
• /
• 2009

The head-related transfer function (HRTF), which has an important role in virtual sound localization has different characteristics across the subjects. Measuring HRTF is very time-consuming and requires a set of specific apparatus. Accordingly, HRTF customization is often employed. In this paper, we propose a method to search an adequate HRTF from a set of the HRTFs. To achieve rapid and reliable customization of HRTF, all HRTFs in the database are partitioned, where a binary search tree was employed. The distortion measurement adopted in HRTF partitioning was determined in a heuristic way, which predicts the differences in perceived sound location well. The DC-Davis CIPIC HRTF database set was used to evaluate the effectiveness of the proposed method. In the listening test, where 10 subjects were participated, the stimuli filtered by the HRTF obtained by the proposed method were closer to those by the personalized HRTF in terms of sound localization. Moreover, performance of the proposed method was shown to be superior to the previous customization method, where the HRFT is selected by using anthropometric data.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.23 no.2
• /
• pp.184-189
• /
• 2013

In this paper, we propose the injurious moving picture filtering system using certain sounds contained in the injurious moving picture to filter injurious moving picture which is distributed without limitation in internet and internet storage space. For this purpose, the Gaussian Mixture Model which can well represent the characteristics of the sound, is used and frame level likelihood estimation is used to calculate the likelihood between filtering target data and the sound models. Also, the pruning method which can real-time proceed by reducing the comparing number of data, is applied for real-time processing, and MWMR method which showed good performance from existing speaker identification, is applied for the distinguish performance of high precision. In the identification experiment result, in case of the frame rate which is the proportion of total frame to high likelihood frame, is set to 50%, identification error rate is 6.06%, and in case of frame rate is set to 60%, error rate is 3.03%. As the result, the proposed system can distinguish between general and injurious moving picture effectively.

• Proceedings of the KOSOMBE Conference
• /
• v.1997 no.11
• /
• pp.555-558
• /
• 1997

A new approach to silent MR imaging using a rotating DC gradient has been explored and experimentally studied. As is known, acoustic or sound noise has been one of the major problems in handling patients, mainly due to the fast gradient pulsings in interaction with the main magnetic field. The sound noise is also proportionally louder as the magnetic field strength becomes larger. In this article, we have described a new imaging technique using a mechanically rotating DC gradient coil as an approach toward silent MR imaging, i.e., a mechanically rotated DC gradient effectively replaces both the phase encoding as well as the readout gradient pulsings and data obtained in this manner provides a set of project ion data which later can be used or the projection reconstructionorwithsomeinterpolation techniques one can also perform conventional 2-D FFT (Fast Fourier Transform) image reconstruction. We found, with this new technique, that the sound noise intensity compared with the conventional imaging technique, such as spin echo sequence, is reduced down to -20.7 dB or about 117.5 times. The experimental pulse sequence and its principle are described and images obtained by the new silent MR imaging technique are reported.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2001.11b
• /
• pp.1162-1165
• /
• 2001

Operational Modal Analysis also known as Output Only Modal Analysis has in the recent years been used for extracting modal parameters of civil engineering structures and is now becoming popular for mechanical structures. The advantage of the method is that no artificial excitation need to be applied to the structure or force signals to be measured. All the parameter estimation is based upon the response signals, thereby minimising the work of preparation for the test. This test case is a controlled lab set-up enabling different parameter estimation methods techniques to be used and compared to the Operational Modal Analysis. For Operational Modal Analysis two different estimation techniques are used： a non-parametric technique based on Frequency Domain Decomposition (FDD), and a parametric technique working on the raw data in time domain, a data driven Stochastic Subspace Identification (SS!) algorithm. These are compared to other methods such as traditional Modal Analysis.

• Proceedings of the Korea Database Society Conference
• /
• 1995.12a
• /
• pp.3-19
• /
• 1995

시스템 Technology ★ Server Technology - 운영환경구축 ★ Network 구성설계 - ATM, FDDI, NMS ★ Client/Server시스템 구성별 Bench Marking ★ Windows 메뉴 및 GUI 설계 ★다기능 PC 운영환경 설정 시스템 Technology ★ Data Base Technology - DB Administration - BB Performance Tuning ★ System Integration Technology - Application Integration - System Flow Control - Task Control - Applicational Interface - S/W Down Load 시스템 Technology ★ Memory Optimization ★ IBM/Facom Host API ★ 영상전화 Customizing - Intel Proshare ★ Auto Dialing - CTI Link ★ IC-Card Interface 시스템 Technology ★ Sound 처리 - Voice Mail - 음절 처리 ★ Image 처리 ★도움말 처리 - Hyper Text 시스템 Technology ★ Socket Programming - 긴급메일 - Peer to peer message switching ★ Set Up Programming -Install Shield ★ DB Access Programming - DB-Library ★ TCP/IP Programming(중략)

• Journal of Korean Society of Transportation
• /
• v.20 no.6
• /
• pp.129-140
• /
• 2002

This paper develops a dynamic structural equation model, which captures relationships among socio-demographics, activity participation(i.e., time use) and travel behavior in consideration with time variation effects. The data used in developing the model are two waves(the year 1991 and 1992) from Puget Sound Transportation Panel (PSTP). which is surveyed in Puget Sound Region in United States. The PSTP is widely used in transportation behavior analysis and includes various information of traveler's socio-economic, travel patterns, and activity participation. In the model, we use 10 endogenous variables including activity participations and travel behaviors and 10 exogenous variables composed of time variant and invariant traveler's socio-demographic variables. The empirical model shows that strong relationships exist not only between socio-demographics and travel behavior, but between waves. We also confirm needs of panel data set to identify and understand time variation effects and travel behaviors.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.22 no.5
• /
• pp.437-443
• /
• 2012

In this paper, the hybrid method to identify the exciting forces and radiated noise generated from the reciprocating compressor was presented. In order to identify the exciting force, both the acceleration data measured at the compressor shell and numerical finite element model for the full set of compressor were used simultaneously. Applying the identified exciting forces to the numerical model, the velocity responses of all nodes at the shell were predicted. Finally the radiated noises from the vibrating shell were predicted by using the direct boundary element acoustic analysis. For precise numerical modeling, the stiffness of rubber mounts and body springs were identified experimentally from the natural frequencies measured by impact testing. The error of over-all sound pressure level between predicted noise and measured noise was about 2.9 dB.