• The KIPS Transactions:PartA
• /
• v.18A no.1
• /
• pp.1-10
• /
• 2011

Physical modeling has been widely used for sound synthesis since it synthesizes high quality sound which is similar to real-sound for musical instruments. However, physical modeling requires a lot of parameters to synthesize a large number of sounds simultaneously for the musical instrument, preventing its real-time processing. To solve this problem, this paper proposes a single instruction, multiple data (SIMD) based multi-core processor that supports real-time processing of sound synthesis of gayageum which is a representative Korean traditional musical instrument. The proposed SIMD-base multi-core processor consists of 12 processing elements (PE) to control 12 strings of gayageum in which each PE supports modeling of the corresponding string. The proposed SIMD-based multi-core processor can generate synthesized sounds of 12 strings simultaneously after receiving excitation signals and parameters of each string as an input. Experimental results using a sampling reate 44.1 kHz and 16 bits quantization show that synthesis sound using the proposed multi-core processor was very similar to the original sound. In addition, the proposed multi-core processor outperforms commercial processors(TI's TMS320C6416, ARM926EJ-S, ARM1020E) in terms of execution time ($5.6{\sim}11.4{\times}$ better) and energy efficiency (about $553{\sim}1,424{\times}$ better).

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.8
• /
• pp.579-588
• /
• 2013

This paper proposes an effective watermarking approach for copyright protection of fire video images. The proposed watermarking approach efficiently utilizes the inherent characteristics of fire data with respect to color and texture by using a gray level co-occurrence matrix (GLCM) and fuzzy c-means (FCM) clustering. GLCM is used to generate a texture feature dataset by computing energy and homogeneity properties for each candidate fire image block. FCM is used to segment color of the fire image and to select fire texture blocks for embedding watermarks. Each selected block is then decomposed into a one-level wavelet structure with four subbands [LL, LH, HL, HH] using a discrete wavelet transform (DWT), and LH subband coefficients with a gain factor are selected for embedding watermark, where the visibility of the image does not affect. Experimental results show that the proposed watermarking approach achieves about 48 dB of high peak-signal-to-noise ratio (PSNR) and 1.6 to 2.0 of low M-singular value decomposition (M-SVD) values. In addition, the proposed approach outperforms conventional image watermarking approach in terms of normalized correlation (NC) values against several image processing attacks including noise addition, filtering, cropping, and JPEG compression.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.6
• /
• pp.703-709
• /
• 2019

This paper proposes speech recognition systems employing Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) structures combined with Hidden Markov Moldel (HMM) to effectively recognize the speech of VeloPharyngeal Insufficiency (VPI) patients, and compares the recognition performance of the systems to the Gaussian Mixture Model (GMM-HMM) and fully-connected Deep Neural Network (DNNHMM) based speech recognition systems. In this paper, the initial model is trained using normal speakers' speech and simulated VPI speech is used for generating a prior model for speaker adaptation. For VPI speaker adaptation, selected layers are trained in the CNN-HMM based model, and dropout regulatory technique is applied in the LSTM-HMM based model, showing 3.68 % improvement in recognition accuracy. The experimental results demonstrate that the proposed LSTM-HMM-based speech recognition system is effective for VPI speech with small-sized speech data, compared to conventional GMM-HMM and fully-connected DNN-HMM system.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.28 no.8
• /
• pp.628-635
• /
• 2017

It is important to prescribe and take medicines that are appropriate for symptoms, since medicines are closely related to human health and life. Moreover, it becomes more important to accurately classify genuine medicines with counterfeit, since the number of counterfeit increases worldwide. However, the number of high-quality experts who have enough experience to properly classify them is limited and there exists a need for the automatic technique to classify medicine tablets. In this paper, we propose a method to classify the tablets by using a handheld spectrometer which provides both Near Infra-Red (NIR) and visible light spectrums. We adopted Support Vector Machine(SVM) as a machine learning algorithm for tablet classification. As a result of the simulation, we could obtain the classification accuracy of 99.9 % on average by using both NIR and visible light spectrums. Also, we proposed a two-step SVM approach to discriminate the counterfeit tablets from the genuine ones. This method could improve both the accuracy and the processing time.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.41 no.9
• /
• pp.1044-1053
• /
• 2016

In this paper, we consider the secondary users (SUs) who are capable of harvesting energy from ambient radio frequency (RF) signals and are allowed to sequentially sense up to 2 different channels to find out idle channels not occupied by the primary users (PUs). The EH SUs are permitted to transmit data packets only if both idle channels and sufficient energy are available. Compared with traditional SUs, the EH SUs consume energy with data transmission and also harvest energy without additional energy supply. Consequently, the battery state is expected to be fluctuated due to energy consumption and harvesting, and therefore we develop a Markov battery model to provide energy variations at the 2-channel sensing EH SUs. With the proposed battery model, we derive the steady-state probability that the EH SUs completely run out of energy, and the achievable throughput of EH SUs is derived accordingly. To evaluate the proposed Markov battery model, the Monte-Carlo simulation was performed to validate the accuracy of energy outage probability and achievable throughput at the 2-channel sensing EH SUs.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.10C
• /
• pp.995-1006
• /
• 2003

It is well know that in the wireless communication systems the transmitted signals can suffer from multipath fading due to the wave propagation characteristics and the obstacles over the paths, resulting in serious reduction in the power of the received signals. However, it is possible to take advantage of the inherent diversity imposed in the multipath reception if the underlying channel can be properly estimated. One of the diversity reception methods in this case is Rake processing. In this paper we study the Rake receivers for the direct-sequence spread-spectrum communication systems utilizing PN (pseudo noise) sequences to achieve spread spectrum. A conventional Rake receiver can use the finite-duration impulse (FIR) filter followed by the PN sequence demodulator, where the FIR filter coefficients are the reverse-ordered complex conjugate values of the fading channel impulse response estimates. Here, we propose a new Rake processing method by replacing the aforementioned PN code sequence with a new set of optimum demodulator coefficients. More specifically, the concept of the new optimum Rake processing is first introduced and then the optimum demodulator coefficients are theoretically derived. The performance obtained using the new optimum Rake processing is also calculated. The analytical results are verified by computer simulation. As a result, it is shown that the new optimum Rake processing method improves the MSE performance more than 10 dB over the conventional one using the fixed PN sequence demodulator. It is also shown that the new optimum Rake processing method improves the MSE performance about 10 dB over the Adaptive Correlator that performs the combining of the multipath components and PN demodulation concurrently. And finally, the MSE performance of the optimum Rake demodulator is very close to the MSE performance of OPSK demodulator under the AWGN channel.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.469-469
• /
• 2017

기후 변화 및 환경오염으로 인하여 물부족 국가가 세계적으로 증가하고 있는 추세이며, 특히 집중형 강우의 형태가 많아짐에 따라 홍수피해 및 상수공급의 문제가 사회적으로 큰 이슈가 되고 있다. 최근 20여 년간의 급속한 경제성장과 도시화 과정에서 인도네시아는 인구와 산업의 과도한 도시집중으로 지난 1960-80년대 한국이 산업화 과정에서 겪었던 것보다 훨씬 심각한 환경문제에 직면하고 있으며, 자카르타와 반둥을 포함하는 광역 수도권 지역의 물 부족과 수질 오염, 환경문제가 이미 매우 위험한 수준에 도달하고 있는 실정이다. 특히, 찌따룸강 중상류에 위치한 인도네시아 3대 도시인 반둥시는 고질적인 용수부족 문제를 겪고 있다. 2010년 현재 약 일평균 15 CMS의 용수가 부족한 상황이며, 2030년에는 지속적인 인구증가로 약 23 CMS의 용수가 추가로 더 필요한 것으로 전망된다. 이러한 용수공급 문제 해결을 위해 반둥시 및 찌따룸강 유역관리청은 댐 및 지하수 개발, 유역 간 물이동 등의 구조적인 대책뿐만 아니라 비구조적인 대책으로써 기존 및 신규 저수지 연계운영을 통한 용수이용의 효율성을 높이는 방안을 모색하고 있다. 이에 따라 본 연구에서는 해당유역의 용수공급 부족 문제를 해소할 수 있는 비구조적인 대책의 일환으로써 다양한 댐 및 보, 소수력 발전, 취수장 등 유역 내 수리 시설물의 운영 최적화를 위한 지능형 물관리 시스템 적용 방안을 제시하고자 한다. 본 연구의 지능형 물관리 시스템은 센서 및 사물 인터넷(Internet of Things, IoT), 네트워크 기술을 바탕으로 시설물 및 운영자, 유관기관 간의 양방향 통신을 통해 유기적인 상호연계 체계를 제공 할 수 있다. 또한 유역의 수문상황과 시설물의 운영현황, 용수공급 및 수요 현황을 실시간으로 확인함으로써 수요에 따른 즉각적인 용수공급량의 조절이 가능하다. 또한, 빅데이터 분석 및 기계학습(Machine Learning)을 통해 개별 물관리 시설물에 대한 최적 운영룰을 업데이트할 수 있으며, 유역의 수문상황과 용수 수요 현황을 고려하여 최적의 용수공급 우선순위를 선정할 수 있다. 지능형 물관리 시스템 개발의 목적은 찌상쿠이 유역의 수문현황을 실시간으로 모니터링하고, 하천시설물의 운영을 분석하여 최적의 용수공급 및 배분을 통해 유역의 수자원 활용 효율성을 향상시키는 데 있다. 이를 위해 수문자료의 수집체계를 구축하고 기관간 정보공유체계를 수립함으로써 분석을 위한 기반 인프라를 구성하며, 이를 기반으로 유역 유출을 비롯한 저수지 운영, 물수지 분석을 수행하고, 분석 및 예측결과, 과거 운영 자료를 토대로 새로운 물관리 시설 운영룰 및 시설물 간 연계운영 방안, 용수공급 우선순위 의사결정 등을 지원하고자 한다. 본 연구의 지능형 물관리 시스템은 통합 DB를 기반으로 수리수문 현상의 모의 분석을 통해 하천 시설물 운영의 합리적 기준을 제시함으로써 다양한 관리주체들의 시설물운영에 대한 이견 및 분쟁을 해소하고, 한정된 수자원과 다양한 수요 간의 효율적이고 합리적인 분배 및 시설물 운영문제를 해결하기 위한 의사결정도구로써 활용할 수 있을 것으로 기대된다.

• The Journal of Society for e-Business Studies
• /
• v.26 no.3
• /
• pp.97-117
• /
• 2021

Predicting term deposit subscriptions is one of representative financial marketing in banks, and banks can build a prediction model using various customer information. In order to improve the classification accuracy for term deposit subscriptions, many studies have been conducted based on machine learning techniques. However, even if these models can achieve satisfactory performance, utilizing them is not an easy task in the industry when their decision-making process is not adequately explained. To address this issue, this paper proposes an explainable scheme for term deposit subscription forecasting. For this, we first construct several classification models using decision tree-based ensemble learning methods, which yield excellent performance in tabular data, such as random forest, gradient boosting machine (GBM), extreme gradient boosting (XGB), and light gradient boosting machine (LightGBM). We then analyze their classification performance in depth through 10-fold cross-validation. After that, we provide the rationale for interpreting the influence of customer information and the decision-making process by applying Shapley additive explanation (SHAP), an explainable artificial intelligence technique, to the best classification model. To verify the practicality and validity of our scheme, experiments were conducted with the bank marketing dataset provided by Kaggle; we applied the SHAP to the GBM and LightGBM models, respectively, according to different dataset configurations and then performed their analysis and visualization for explainable term deposit subscriptions.

• Journal of the Korea Organic Resources Recycling Association
• /
• v.28 no.4
• /
• pp.43-52
• /
• 2020

As the membrane gas separation technology grows, various models were developed by numerous researchers to describe the separation process. In this work, the counter-current model was compared thoroughly with experimental data. Experimentally, hollow fiber membrane using CA module was prepared for the separation of biogas. The pure gas permeation properties of membrane module for methane, nitrogen, oxygen, and carbon dioxide were measured. The permeance of CO2 and CH4 were 25.82 GPU and 0.65 GPU, respectively. The high CO2/CH4 selectivity of 39.7 was obtained. the separation test for three different simulated mixed gases were carried out after pure gas test, and the gas concentration of the permeate at various stage-cut were measured from CA membrane module. Results showed that the experimental data agreed with the numerical simulation. A mathematical model has implemented in this study for the separation of biogas using a membrane module. The finite difference method (FDM) is applied to calculate the membrane biogas separation behaviors. Futhermore, the counter-current model can be considered as a convenient model for biogas separation process.

• The Korean Journal of Applied Statistics
• /
• v.34 no.3
• /
• pp.309-327
• /
• 2021

Dynamic treatment regimes (DTRs) are decision-making rules designed to provide personalized treatment to individuals in multi-stage randomized trials. Unlike classical methods, in which all individuals are prescribed the same type of treatment, DTRs prescribe patient-tailored treatments which take into account individual characteristics that may change over time. The Q-learning method, one of regression-based algorithms to figure out optimal treatment rules, becomes more popular as it can be easily implemented. However, the performance of the Q-learning algorithm heavily relies on the correct specification of the Q-function for response, especially in observational studies. In this article, we examine a number of double-robust weighted least-squares estimating methods for Q-learning in high-dimensional settings, where treatment models for propensity score and penalization for sparse estimation are also investigated. We further consider flexible ensemble machine learning methods for the treatment model to achieve double-robustness, so that optimal decision rule can be correctly estimated as long as at least one of the outcome model or treatment model is correct. Extensive simulation studies show that the proposed methods work well with practical sample sizes. The practical utility of the proposed methods is proven with real data example.