• The Journal of the Acoustical Society of Korea
• /
• v.20 no.7
• /
• pp.55-62
• /
• 2001

In this paper, we propose a new adaptative pruning algorithm which effectively reduces the search space during the recognition process. As maximum probabilities between neighbor frames are highly interrelated, an efficient pruning threshold value can be obtained from the maximum probabilities of previous frames. The main idea is to update threshold at the present frame by a combination of previous maximum probability and hypotheses probabilities. As present threshold is obtained in on-going recognition process, the algorithm does not need any pre-experiments to find threshold values even when recognition tasks are changed. In addition, the adaptively selected threshold allows an improvement of recognition speed under different environments. The proposed algorithm has been applied to a Korean Address recognition system. Experimental results show that the proposed algorithm reduces the search space of average 14.4% and 9.14% respectively while preserving the recognition accuracy, compared to the previous method of using fixed pruning threshold values and variable pruning threshold values.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.20 no.2
• /
• pp.165-179
• /
• 2007

For the purpose of optimum design of reinforced concrete structures, pre-determined section database of column and beam are constructed and arranged in order of the resisting capacity. Then, regression equations representing the relation between section number and section resisting capacity are derived. In advance, effective optimization algorithms which search optimized solution quickly using direct search method from these database are proposed. In practice, from the fact that engineers conduct member design close to capacity optimization rather than cost optimization, both capacity and cost optimization using proposed algorithms are performed, and the review for the obtained results are followed. Moreover, the investigation for the applicability and effectiveness of the Introduced design procedure is conducted through correlation study for example structures. Because of no restriction in constructing objective functions with very simple optimization processes and fast convergence, the introduced method can effectively be used in the preliminary design stage. Especially, selected solutions from database are directly applicable in practice because these sections already satisfy all the requirements in design codes and practical restrictions.

• Journal of Broadcast Engineering
• /
• v.9 no.4 s.25
• /
• pp.434-445
• /
• 2004

Motion Estimation is an essential module in video encoders based on international standards such as H.263 and MPEG. Many fast motion estimation algorithms have been proposed in order to reduce the computational complexity of a well-known full search algorithms(FS). However, these fast algorithms can not work efficiently in DSP processors recently developed for video processing. To solve for this. we propose an efficient motion estimation scheme optimized in the DSP processor like Philips TM1300. A motion vector predictor is pre-estimated and a small search range is chosen in the proposed scheme using strong motion vector correlation between a current macro block (MB) and its neighboring MB's to reduce computation time. An MPEG-4 SP@L3(Simple Profile at Level 3) encoding system is implemented in Philips TM1300 to verify the effectiveness of the proposed method. In that processor, we can achieve better performance using our method than other conventional ones while keeping visual quality as good as that of the FS.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.20 no.1
• /
• pp.43-48
• /
• 1984

The transfer matrix method is well-known and extensively used for finding solutions in vibration problems. At the final stage of this method natural frequencies are obtained by a trial and error search procedure. In this paper authors presented the method which needed only a few division number to yield an accurate solution and the most effective method to get an approximate solution in the case of beam vibrations. The methods which were presented by authors could be applied for the beam with nonuniform section and uniformly distributed load, and the values of numerical calculations by these methods have just agreed with those of experiments.

• Journal of the Korea Convergence Society
• /
• v.12 no.11
• /
• pp.69-80
• /
• 2021

Neural machine translation using deep neural network has emerged as a mainstream research, and an abundance of investment and studies on model structure and parallel language pair have been actively undertaken for the best performance. However, most recent neural machine translation studies pass along decoding strategy to future work, and have insufficient a variety of experiments and specific analysis on it for generating language to maximize quality in the decoding process. In machine translation, decoding strategies optimize navigation paths in the process of generating translation sentences and performance improvement is possible without model modifications or data expansion. This paper compares and analyzes the significant effects of the decoding strategy from classical greedy decoding to the latest Dynamic Beam Allocation (DBA) in neural machine translation using a sequence to sequence model.

• Imaging Science in Dentistry
• /
• v.53 no.2
• /
• pp.101-108
• /
• 2023

Purpose: The aim of this study was to conduct a scoping review and meta-analysis to provide overall estimates of the recall and precision of artificial intelligence for detection and segmentation using oral and maxillofacial cone-beam computed tomography (CBCT) scans. Materials and Methods: A literature search was done in Embase, PubMed, and Scopus through October 31, 2022 to identify studies that reported the recall and precision values of artificial intelligence systems using oral and maxillofacial CBCT images for the automatic detection or segmentation of anatomical landmarks or pathological lesions. Recall (sensitivity) indicates the percentage of certain structures that are correctly detected. Precision (positive predictive value) indicates the percentage of accurately identified structures out of all detected structures. The performance values were extracted and pooled, and the estimates were presented with 95% confidence intervals(CIs). Results: In total, 12 eligible studies were finally included. The overall pooled recall for artificial intelligence was 0.91 (95% CI: 0.87-0.94). In a subgroup analysis, the pooled recall was 0.88 (95% CI: 0.77-0.94) for detection and 0.92 (95% CI: 0.87-0.96) for segmentation. The overall pooled precision for artificial intelligence was 0.93 (95% CI: 0.88-0.95). A subgroup analysis showed that the pooled precision value was 0.90 (95% CI: 0.77-0.96) for detection and 0.94 (95% CI: 0.89-0.97) for segmentation. Conclusion: Excellent performance was found for artificial intelligence using oral and maxillofacial CBCT images.

• Imaging Science in Dentistry
• /
• v.53 no.2
• /
• pp.137-144
• /
• 2023

Purpose: This study aimed to evaluate the anatomic circle around the impacted lower third molar to show, document, and correlate essential findings that should be included in the routine radiographic assessment protocol as clinically meaningful factors in overall case evaluation and treatment planning. Materials and Methods: Cone-beam computed tomographic images of impacted lower third molars were selected according to specific inclusion criteria. Impacted teeth were classified according to their position before assessment. The adjacent second molars were assessed for distal caries, distal bone loss, and root resorption. The fourth finding was the presence of a retromolar canal distal to the impaction. Communication with the dentist responsible for each case was done to determine whether these findings were detected or undetected by them before communication. Results: Statistically significant correlations were found between impaction position, distal bone loss, and detected distal caries associated with the adjacent second molar. The greatest percentage of undetected findings was found in the evaluation of distal bone status, followed by missed detection of the retromolar canal. Conclusion: The radiographic assessment protocol for impacted third molars should consider a step-by-step evaluation for second molars, and clinicians should be aware of the high prevalence of second molar affection in horizontal and mesioangular impactions. They also should search for the retromolar canal due to its associated clinical considerations.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.25 no.2
• /
• pp.235-244
• /
• 2014

This paper describes the subarray channel configuration and calibration method for airborne AESA radar antenna. AESA radar demonstrator was designed and implemented for the digital beam-forming performance test of the 12 channel subarray structure. Magnitude and phase difference can be exist between the manufactured subarray channel. In this paper, calibration method for the subarray difference error was suggested. We measured digital monopulse slope in the subarray channel and verified the channel calibration effect. To verify the subarray channel operation, digital monopulse channel was compared with analog monopulse channel performance. AESA radar demonstrator was tested in the ground far field test range. Emulated single target was generated to test the detection and tracking performance of the demonstrator with the same waveform and search pattern. We verified that the detection and tracking performance of the 12 subarray digital monopulse channel was similar with the conventional analog monopulse channel. Also, ABF(Adaptive Beam-Forming) function for the sidelobe jammer was tested and effective operation was verified.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.5
• /
• pp.94-101
• /
• 1996

When objects are scanned spatially by a laser-beam and mechanical mirror scanners, spatial information can be obtained, and then it is improtant to accurately obtain the parameters relating the light source and camera positions, etc.. In this paper, a calibration technique is presented for correction of measuremtn errors in a three-dimensional laser scanner system with two galvanometers. First, a model of the systematic errors is developed based on the geometry of the scanning system. Calibration parameter values are then iteratively adjusted with coarse-fine search in order to minimize errors (evaluation function) between measured and computed distances. It is shown that this correction method results in measurement precision suitable for practical use.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.06c
• /
• pp.59-62
• /
• 1998

본 논문에서는 한국전자통신연구원의 Handy Combi 응용 도메인을 대상으로 한 화자독립 연속음성인식 시스템 개발에 관하여 기술한다. 불특정화자가 자연스럽게 발음한 연속음성을 인식하는 기술은 펜인식 등과 더불어 멀티모달 인터페이스의 핵심 요소로서, 이동 환경에서 사용자의 다양한 요구사항을 처리하는 지능형 에이전트에 구현을 위해 필수적으로 개발되어야 하는 기술이다. 본 논문에서는 연속확률분포를 가지는 Hidden Markov Model(HMM) 기반의 연속음성인식 시스템을 구현하였다. 개발된 시스템은 음성특징벡터로 MFCC를 사용하였으며, 음소 모델의 강인한 훈련을 위해 음성학적 지식에 기반을 둔 tree-based clustering 방식을 도입하였다. 인식단계에서는 인식속도를 개선시키기 위해 beam-search 기법을 적용하였다. 인식 실험 결과, 99.7%의 어절 인식률과 98.8%의 문장 인식률을 얻었으며, 최종적인 문장의 이해도는 99% 이상이었다.