• ETRI Journal
• /
• v.43 no.5
• /
• pp.775-786
• /
• 2021

To translate in real time, a simultaneous translation system should determine when to stop reading source tokens and generate target tokens corresponding to a partial source sentence read up to that point. However, conventional attention-based neural machine translation (NMT) models cannot produce translations with adequate latency in online scenarios because they wait until a source sentence is completed to compute alignment between the source and target tokens. To address this issue, we propose a reinforced learning (RL)-based attention mechanism, the reinforced attention mechanism, which allows a neural translation model to jointly train the stopping criterion and a partial translation model. The proposed attention mechanism comprises two modules, one to ensure translation quality and the other to address latency. Different from previous RL-based simultaneous translation systems, which learn the stopping criterion from a fixed NMT model, the modules can be trained jointly with a novel reward function. In our experiments, the proposed model has better translation quality and comparable latency compared to previous models.

• Communications for Statistical Applications and Methods
• /
• v.20 no.4
• /
• pp.241-246
• /
• 2013

In this study, we propose a nonparametric simultaneous test procedure for the location translation and scale parameters. We consider the Wilcoxon rank sum test for the location translation parameter and the Mood test for the scale parameter with the quadratic and maximal types of combining functions. Then we derive the limiting null distributions of the combining functions. We illustrate our procedure with an example and compare efficiency by obtaining the empirical powers through a simulation study. Finally, we discuss some interesting features related to the nonparametric simultaneous tests.

• Journal of the Korea Convergence Society
• /
• v.13 no.3
• /
• pp.43-54
• /
• 2022

Simultaneous translation is online decoding to translates with only subsentence. The goal of simultaneous translation research is to improve translation performance against delay. For this reason, most studies find trade-off performance between delays. We studied the experiments of the fixed policy-based simultaneous translation in Korean. Our experiments suggest that Korean tokenization causes many fragments, resulting in delay compared to other languages. We suggest follow-up studies such as n-gram tokenization to solve the problems.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.8 no.7
• /
• pp.759-768
• /
• 2018

As the number of exchange in various fields between countries increases, the number of international conference increases. Many students study simultaneous interpretation due to the increased demand of simultaneous interpretation. Since simultaneous interpretation requires a lot of learning time, students majoring in translation perform the self learning. The paper based sight translation training system is a representative self learning method, but backtracking decreases the efficiency of self learning and it requires the help of the partner. To improve the learning efficiency, computer based sight translation training system has been proposed. However, since students uses the computer based sight translation training system only in a fixed area due to low mobility of computer, the utilization of the system decreases. In this paper, smart device based sight translation training system has been proposed to increase the utilization of the proposed system. Since smart device has lower computing capabilities than the computer, we have proposed algorithms to deal with the low performance. We implement and evaluate the functionalities of the proposed training system.

• Journal of applied mathematics & informatics
• /
• v.26 no.1_2
• /
• pp.317-324
• /
• 2008

This paper shows the degree of simultaneous neural network approximation for a target function in $C^r$[-1, 1] and its first derivative. We use the Jackson's theorem for differentiable functions to get a degree of approximation to a target function by algebraic polynomials and trigonometric polynomials. We also make use of the de La Vall$\grave{e}$e Poussin sum to get an approximation order by algebraic polynomials to the derivative of a target function. By showing that the divided difference with a generalized translation network can be arbitrarily closed to algebraic polynomials on [-1, 1], we obtain the degree of simultaneous approximation.

• Journal of the Korean Professional Engineers Association
• /
• v.33 no.4
• /
• pp.53-58
• /
• 2000

Within 10years, it will be possible to integrate 10B transistors on a single chip microprocessor which wilt operate far beyond GHZ, and it will execute about 20-200 instructions per clock cycle from widely variable instruction streams leveraging SMT(Simultaneous Multithreading) technology . Also it will decouple the current legacy X86 binary compatibility by translation layer such as code morphing technology.

• Journal of Korean Neurosurgical Society
• /
• v.57 no.3
• /
• pp.178-184
• /
• 2015

Objective : Posterior vertebral translation as a type of spondylolisthesis, retrolisthesis is observed commonly in patients with degenerative spinal problems. Nevertheless, there is insufficient literature on retrolisthesis compared to anterolisthesis. The purpose of this study is to clarify the clinical features of retrolisthesis, and its developmental mechanism associated with a compensatory role in sagittal imbalance of the lumbar spine. Methods : From 2003 to 2012, 230 Korean patients who underwent spinal surgery in our department under the impression of degenerative lumbar spinal disease were enrolled. All participants were divided into four groups : 35 patients with retrolisthesis (group R), 32 patients with simultaneous retrolisthesis and anterolisthesis (group R+A), 76 patients with anterolisthesis (group A), and 87 patients with non-translation (group N). The clinical features and the sagittal parameters related to retrolisthesis were retrospectively analyzed based on the patients' medical records. Results : There were different clinical features and developmental mechanisms between retrolisthesis and anterolisthesis. The location of retrolisthesis was affected by the presence of simultaneous anterolisthesis, even though it predominantly manifest in L3. The relative lower pelvic incidence, pelvic tilt, and lumbar lordosis compared to anterolisthesis were related to the generation of retrolisthesis, with the opposite observations of patients with anterolisthesis. Conclusion : Retrolisthesis acts as a compensatory mechanism for moving the gravity axis posteriorly for sagittal imbalance in the lumbar spine under low pelvic incidence and insufficient intra-spinal compensation.

• Journal for History of Mathematics
• /
• v.34 no.5
• /
• pp.165-191
• /
• 2021

The simple simultaneous linear congruence equation solution in 《Sunzi Suanjing孫子算經》 has developed into a systematic general solution in 《Shuxue Jiuzhang數書九章》. The most important parts of it is the creation of the Dayan Qiuyishu大衍求一術. The reason why 《Sanhak Jeongeui算學正義》 deals with the Dayanshu大衍術 which has lost its practicality in the calculation of astronomical calendar, is that one of the best achievement of traditional mathematics cannot be omitted. It is written with an emphasis on the disclosure of the calculation principle. It shows that Nam Byeong-gil南秉吉 and Lee Sang-hyeok李尙爀 had a fairly structural understanding of the Dayan Qiuyishu.

• 제어로봇시스템학회:학술대회논문집
• /
• 1994.10a
• /
• pp.232-237
• /
• 1994

An extended study of optimal thruster combination for simultaneous attitude and orbital maneuvers of a jet-controled spacecraft is conducted. In this case, the spacecraft has not enough number of thrusters to control the rotation and translation separately. Therefore, thrusters are employed by combining to eliminate their coupling effects. The combinations are determined to minimize the fuel consumption. The redundancy study for some thruster failure cases is also presented.

• ETRI Journal
• /
• v.40 no.2
• /
• pp.257-265
• /
• 2018

Pronouns are frequently dropped in Korean sentences, especially in text messages in the mobile phone environment. Restoring dropped pronouns can be a beneficial preprocessing task for machine translation, information extraction, spoken dialog systems, and many other applications. In this work, we address the problem of dropped pronoun recovery by resolving two simultaneous subtasks: detecting zero-pronoun sentences and determining the type of dropped pronouns. The problems are statistically modeled by encoding the sentence and classifying types of dropped pronouns using a recurrent neural network (RNN) architecture. Various RNN-based encoding architectures were investigated, and the stacked RNN was shown to be the best model for Korean zero-pronoun recovery. The proposed method does not require any manual features to be implemented; nevertheless, it shows good performance.