• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.9
• /
• pp.49-57
• /
• 2009

In this paper, we propose a multiple voltage scheduling method which reduces energy consumption considering both timing constraints and resource constraints. In the other multiple voltage scheduling techniques, high voltage is assigned to operations in the longest path and low voltage is assigned to operations that are not on the longest path. However, in those methods, voltages are assigned to specific operations restrictively. We use a simulated annealing technique, in which several voltages are assigned to specific operations flexibly regardless of whether they are on the longest path. In this paper, a post processing algorithm is proposed to further reduce the energy consumption. In some cases, designers may want to reduce the level shifters. To make tradeoff between the total energy and the number (or energy) of level shifters weighted term can be added to the cost function. When the level shifter energy is weighted six times, for example, the number of level shifters is reduced by about 24% and their energy consumption is reduced by about 20%.

• Journal of KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1171-1179
• /
• 2004

We describe word boundary detection that extracts the boundary between speech and non-speech. The proposed method uses two features. One is the normalized root mean square of speech signal, which is insensitive to white noises and represents temporal information. The other is the normalized met-frequency band energy of voice signal, which is frequency information of the signal. Our method detects word boundaries using a recurrent fuzzy associative memory(RFAM) that extends FAM by adding recurrent nodes. Hebbian learning method is employed to establish the degree of association between an input and output. An error back-propagation algorithm is used for teaming the weights between the consequent layer and the recurrent layer. To confirm the effectiveness, we applied the suggested system to voice data obtained from KAIST.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.5A
• /
• pp.493-500
• /
• 2006

This paper presents the efficient structure and parameter optimization of downlink beamforming transmitter in OFDMA/TDD system. To design downlink beamforming transmitter for multiple transmit antennas, an efficient beamforming structure for multiple users and the choice of word-length of each block are critical in the aspect of its performance and hardware complexity. We propose an efficient beamforming scheme, which stores the weights of subcarriers into memory without user identification at the receiver of base station and calculates the weights for corresponding user in a subcarrier unit of IFFT input at high speed. Also, we obtain the word-length of main data path and other design parameters by fixed-point simulation analysis. The proposed architecture could reduce the memory size proportional to the maximum number of users per frame, and the processing time of an OFDM symbol at the receiver of base station without the need of additional processing time for calculating the weights at the transmitter.

• Journal of Digital Convergence
• /
• v.11 no.3
• /
• pp.389-395
• /
• 2013

More sales of food and beverage coupons have been made using SNS on social commerce recently. If one buys coupons on social commerce, he/she can enjoy products at a lower price; however, there are drawbacks that one must consider such as location, service hours, and discount rate. Thus, this paper suggests a system that recommends food and beverage coupons on social commerce for users that considers a user's personal context of location, time, and purchase history. In order to reflect a user's context awareness and continuous preference, this paper suggests a method based on the Bayesian network. In order to reflect personalized weighting on the standard of coupon selection to match a user's preference, a measurement and classification of weighting preferences is performed on the basis of AHP. 20 experiments in one month involving 12 students were carried out to verify the effectiveness of the system, resulting in an 80% satisfaction level.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.4 s.36
• /
• pp.385-396
• /
• 2005

In this paper, a new structural approach to on-line signature verification is presented. A primitive pattern is defined as a part segmented by a local minimal position of speed. And a structural description of signature is composed of subpatterns which are defined as such forms as rotation shape, cusp shape and bell shape, acquired by composition of the primitives regarding the directional changes. As the matching method to find identical parts between two signatures, a modified DP(dynamic programming) matching algorithm is presented. And also, variation and complexity of local parts are computed from the training samples, and reference model and decision boundary are derived from these. Error rate, execution time and memory usage are compared among the functional approach, the parametric approach and the proposed structural approach. It is found that the average error rate can be reduced from 14.2% to 4.05% when the local parts of a signature are weighted and the complexity is used as a factor of decision threshold. Though the error rate is similar to that of functional approaches. time consumption and memory usage of the proposed structural approach are shown to be very effective.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.363-363
• /
• 2019

활성화 함수(activation function)는 기계학습(machine learning)의 학습과정에 비선형성을 도입하여 심층적인 학습을 용이하게 하고 예측의 정확도를 높이는 중요한 요소 중 하나이다(Roy et al., 2019). 일반적으로 기계학습에서 사용되고 있는 활성화 함수의 종류에는 계단 함수(step function), 시그모이드 함수(sigmoid 함수), 쌍곡 탄젠트 함수(hyperbolic tangent function), ReLU 함수(Rectified Linear Unit function) 등이 있으며, 예측의 정확도 향상을 위하여 다양한 형태의 활성화 함수가 제시되고 있다. 본 연구에서는 기계학습을 통하여 수위예측 시 정확도 향상을 위하여 Hybrid 활성화 함수를 제안하였다. 연구대상지는 조수간만의 영향을 받는 한강을 대상으로 선정하였으며, 2009년 ~ 2018년까지 10년간의 수문자료를 활용하였다. 수위예측 알고리즘은 Python 내 Tensorflow의 RNN (Recurrent Neural Networks) 모델을 이용하였으며, 강수량, 수위, 조위, 댐 방류량, 하천 유량의 수문자료를 학습시켜 3시간 및 6시간 후의 수위를 예측하였다. 예측정확도 향상을 위하여 입력 데이터는 정규화(Normalization)를 시켰으며, 민감도 분석을 통하여 신경망모델의 은닉층 개수, 학습률의 최적 값을 도출하였다. Hybrid 활성화 함수는 쌍곡 탄젠트 함수와 ReLU 함수를 혼합한 형태로 각각의 가중치($w_1,w_2,w_1+w_2=1$)를 변경하여 정확도를 평가하였다. 그 결과 가중치의 비($w_1/w_2$)에 따라서 예측 결과의 RMSE(Roote Mean Square Error)가 최소가 되고 NSE (Nash-Sutcliffe model Efficiency coefficient)가 최대가 되는 지점과 Peak 수위의 예측정확도가 최대가 되는 지점을 확인할 수 있었다. 본 연구는 현재 Data modeling을 통한 수위예측의 정확도 향상을 위해 기초가 되는 연구이나, 향후 다양한 형태의 활성화 함수를 제안하여 정확도를 향상시킨다면 예측 결과를 통하여 침수예보에 대한 의사결정이 가능할 것으로 기대된다.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.5
• /
• pp.525-533
• /
• 2022

We propose an effective method for suppressing both side and grating lobes by applying 2-dimensional Fourier Transform to the received channel data during the receive focusing process of an ultrasound imaging system. When the signal from the image point is focused, the channel signals have the same DC value across the channels. However, even after echoes from outside an imaging point are focused, they are manifested as having different spatial frequencies depending on their incident angles. Therefore, after the receive focusing delay time is applied, 2-D Fourier Transform is performed on the time-channel data to separate the DC component and other frequency components in the spectral domain, and the weighting value is defined using the ratio of the two values. The side lobe and grating lobe were suppressed by multiplying the ultrasound image by a weighting value. Ultrasound images with a frequency of 5 MHz were simulated in a 64-channel linear array. The grating lobe appearing in the ultrasound image was completely removed by applying the proposed method. In addition, the side lobe was reduced and the lateral resolution was greatly increased. Results of computer simulation on a human organ mimicking image show that the proposed method can aid in better lesion diagnosis by increasing the image contrast.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 1999.06b
• /
• pp.35-40
• /
• 1999

이 논문은 디지털 TV의 압축 비트열을 전송할 때, 채널 오류로 인해 디지털 TV 디코더에서 화면을 재생할 때 발생할 수 있는 화질의 저하를 최소화시키는 방법을 기술한다. 텔레비전과 같은 단방향 방송 시스템에서 전송 도중에 오류가 발생하여 수신측에서 전송 오류를 제대로 정정하기 곤란할 때, 재생된 화면이 심각하게 훼손될 수 있다. 이때 오류로 인한 화면의 손상을 주변의 공간적 또는 시간적 상관 정보를 이용하여 화면을 복원하는 동작을 오류 은폐라고 한다. 최근 디지털 TV나 고선명 TV의 기본 부호화 방식으로 많이 사용하고 있는 MPEG 비디오 표준 방식은 공간적인 중복 정보를 압축하기 위해 DCT 변환을 수행하며, 시간적인 중복 정보를 압축하기 위해 움직임 예측과 움직임 보상 방법을 이용한다. 또한 MPEG 비디오 압축 방식은 일종의 차분 부호화 방법을 사용하기 때문에 현재 화면에서 오류가 발생하면 현재 화면뿐만 아니라, 시간적으로 나중에 부호화된 화면에도 오류가 전파된다. 본 논문에서는 MPEG 비트열의 채널 오류의 영향을 분석하여 화면간에 존재하는 시간적인 중복성을 이용하여 움직임 벡터를 추정하여 손상된 부분을 은폐시키는 방법을 제안한다. 기본적으로 손상된 매크로블록의 위와 아래로 인접한 화소값들을 움직임 벡터의 추정에 사용한다. 제안된 방법들 중에서 주변 움직임 벡터들의 가중치를 이용한 평균값 방법과 초기 움직임 벡터를 이용한 확장 영역 움직임 추정 방법이 우수한 결과를 보였다.

• KIPS Transactions on Computer and Communication Systems
• /
• v.5 no.3
• /
• pp.47-54
• /
• 2016

In this paper, we propose a new opportunistic routing scheme based on timed bargaining game. The proposed algorithm effectively formulates the opportunistic routing mechanism as an repeated bargaining model using timed learning method. Additionally, we formulate a new contention window adjusting scheme for reduce collision rate. Simulation results indicate that the proposed scheme has excellent performance than other existing schemes under widely diverse VANET environments.

• Journal of Internet Computing and Services
• /
• v.14 no.2
• /
• pp.15-24
• /
• 2013

In recent years, the number of social network system has grown rapidly. Among them, social bookmarking system(SBS) is one of the most popular systems. SBS provides network platform which users can share and manage various types of online resources by using tags. In SBS, it can be possible to reflect tag and time in order to enhance the quality of personalized recommendation. In this paper, we proposed recommender system which reflect tag and time at weight generation and similarity calculation. Also we adapted proposed method to real dataset and the result of experiment showed that the our method offers better performance when such information is integrated.