• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.11
• /
• pp.2255-2263
• /
• 1994

Protocol conformance testing consists of procedures to observe an output and to check a transition state of the Implementation Under Test considered as a black box by applying an input. There are several methods to check the transition state such as Unique Input/Output(UIO) sequence. Distinguishing Sequence(DS) and Characterization Set(CS). Particularly, as a test method for a state having no UIO sequence, Partial UIO sequence method can be considered. In this paper, three properties which can be found among Partial UIO sequence and a modified algorithm using these properties are suggested.

• 한국어정보학회:학술대회논문집
• /
• 2016.10a
• /
• pp.56-61
• /
• 2016

본 논문에서는 Sequence-to-sequence 모델을 생성요약의 방법으로 한국어 문서요약에 적용하였으며, copy mechanism과 input feeding을 적용한 RNN search 모델을 사용하여 시스템의 성능을 높였다. 인터넷 신문기사를 수집하여 구축한 한국어 문서요약 데이터 셋(train set 30291 문서, development set 3786 문서, test set 3705문서)으로 실험한 결과, input feeding과 copy mechanism을 포함한 모델이 형태소 기준으로 ROUGE-1 35.92, ROUGE-2 15.37, ROUGE-L 29.45로 가장 높은 성능을 보였다.

• Proceedings of the IEEK Conference
• /
• 2000.07a
• /
• pp.367-370
• /
• 2000

It is shown that $I_{DDQ}$ testing is very useful for shipping fault-free CMOS ICs. However, test time of $I_{DDQ}$ testing is extremely larger than one of logic testing. In this paper, a new test input sequence generation methodology is proposed to reduce the test time of $I_{DDQ}$ testing. At first, it is Shown that $I_{DDQ}$ test time Will be denominated by charge supply current for load capacitance of gates whose output logic values are changed by test input vector application and the charge current depends on input sequence of test vectors. After that, a test input sequence generation methodology is proposed. The feasibility is checked by some experiments.riments.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.476-482
• /
• 2020

Sequence-to-sequence deep neural networks with attention mechanisms have shown superior performance across various domains, where the sizes of the input and the output sequences may differ. However, if the input sequences are much longer than the output sequences, and the characteristic of the input sequence changes within a single output token, the conventional attention mechanisms are inappropriate, because only a single context vector is used for each output token. In this paper, we propose a double-attention mechanism to handle this problem by using two context vectors that cover the left and the right parts of the input focus separately. The effectiveness of the proposed method is evaluated using speech recognition experiments on the TIMIT corpus.

• 한국어정보학회:학술대회논문집
• /
• 2016.10a
• /
• pp.20-24
• /
• 2016

Sequence-to-sequence 모델은 입력열을 길이가 다른 출력열로 변환하는 모델로, 단일 신경망 구조만을 사용하는 End-to-end 방식의 모델이다. 본 논문에서는 Sequence-to-sequence 모델을 한국어 구구조 구문 분석에 적용한다. 이를 위해 구구조 구문 트리를 괄호와 구문 태그 및 어절로 이루어진 출력열의 형태로 만들고 어절들을 단일 기호 'XX'로 치환하여 출력 단어 사전의 수를 줄였다. 그리고 최근 기계번역의 성능을 높이기 위해 연구된 Attention mechanism과 Input-feeding을 적용하였다. 실험 결과, 세종말뭉치의 구구조 구문 분석 데이터에 대해 기존의 연구보다 높은 F1 89.03%의 성능을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2016.10a
• /
• pp.20-24
• /
• 2016

Sequence-to-sequence 모델은 입력열을 길이가 다른 출력열로 변환하는 모델로, 단일 신경망 구조만을 사용하는 End-to-end 방식의 모델이다. 본 논문에서는 Sequence-to-sequence 모델을 한국어 구구조 구문 분석에 적용한다. 이를 위해 구구조 구문 트리를 괄호와 구문 태그 및 어절로 이루어진 출력열의 형태로 만들고 어절들을 단일 기호 'XX'로 치환하여 출력 단어 사전의 수를 줄였다. 그리고 최근 기계번역의 성능을 높이기 위해 연구된 Attention mechanism과 Input-feeding을 적용하였다. 실험 결과, 세종말뭉치의 구구조 구문 분석 데이터에 대해 기존의 연구보다 높은 F1 89.03%의 성능을 보였다.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.15-16
• /
• 2006

In this paper, we propose an adaptive SLM scheme based on peak observation for PAPR reduction of OFDM signals. The proposed scheme is composed of three steps: peak scaling, sequence selection, and SLM procedures. In the first step, the peak signal samples in the IFFT outputs of the original input sequence are scaled down. In the second step, the sub-carrier positions where power difference between the original input sequence and the FFT outputs of the scaled signal is large, are identified. Then, the phase sequences which have the maximum number of phase-reversed sequence words only for these positions, are selected. Finally, only using the selected phase sequences, the generic SLM procedure is performed for the original input sequence. Simulation results reveal that the proposed adaptive SLM remarkably reduces the complexity in terms of IFFT and PAPR calculations than the conventional SLM, while maintaining the PAPR reduction performance.

• Proceedings of the KIPE Conference
• /
• 2018.07a
• /
• pp.114-116
• /
• 2018

This paper presents a control strategy to achieve the balanced sinusoidal output currents, as well as sinusoidal input currents for the matrix converter (MC) under unbalanced input voltages. By regulating the modulation index of the converter according to the instantaneous input voltages, the output currents are kept balanced and sinusoidal. In order to obtain sinusoidal input currents, the input power factor angle should be dynamically calculated based on the positive and negative sequence components of the input voltages. This paper proposes a simple method to construct the expected input power factor angle without the complicated sequence component extraction of input voltages. Simulation results are given to validate the effectiveness of the proposed control strategy.

• Journal of Korea Multimedia Society
• /
• v.22 no.6
• /
• pp.655-664
• /
• 2019

In this paper, we propose a novel way of combining multiple deep convolutional neural network (DCNN) architectures which work well for accurate video face identification by adopting a serial combination of 3D and 2D DCNNs. The proposed method first divides an input video sequence (to be recognized) into a number of sub-video sequences. The resulting sub-video sequences are used as input to the 3D DCNN so as to obtain the class-confidence scores for a given input video sequence by considering both temporal and spatial face feature characteristics of input video sequence. The class-confidence scores obtained from corresponding sub-video sequences is combined by forming our proposed class-confidence matrix. The resulting class-confidence matrix is then used as an input for learning 2D DCNN learning which is serially linked to 3D DCNN. Finally, fine-tuned, serially combined DCNN framework is applied for recognizing the identity present in a given test video sequence. To verify the effectiveness of our proposed method, extensive and comparative experiments have been conducted to evaluate our method on COX face databases with their standard face identification protocols. Experimental results showed that our method can achieve better or comparable identification rate compared to other state-of-the-art video FR methods.

• Journal of applied mathematics & informatics
• /
• v.4 no.2
• /
• pp.513-528
• /
• 1997

A sort sequence $S_n$ is a sequence of all unordered pairs of indices in $I_n\;=\;{1,\;2,v...,\;n}$. With a sort sequence Sn we assicuate a sorting algorithm ($AS_n$) to sort input set $X\;=\;{x_1,\;x_2,\;...,\;x_n}$ as follows. An execution of the algorithm performs pairwise comparisons of elements in the input set X as defined by the sort sequence $S_n$, except that the comparisons whose outcomes can be inferred from the outcomes of the previous comparisons are not performed. Let $X(S_n)$ denote the acverage number of comparisons required by the algorithm $AS_n$ assuming all input orderings are equally likely. Let $X^{\ast}(n)\;and\;X^{\circ}(n)$ denote the minimum and maximum value respectively of $X(S_n)$ over all sort sequences $S_n$. Exact determination of $X^{\ast}(n),\;X^{\circ}(n)$ and associated extremal sort sequenes seems difficult. Here, we obtain bounds on $X^{\ast}(n)\;and\;X^{\circ}(n)$.