• Proceedings of the Korean Information Science Society Conference
• /
• 2005.11b
• /
• pp.661-663
• /
• 2005

상업적인 추천 시스템에서 폭넓게 사용되고 있는 사용자 기반의 협력적 여과 방법 (User-Based Collaborative Filtering)은 확장성과 실시간 성능에 관련된 많은 제약을 갖는다. 이와 같은 맹점을 해결하기 위해 제안된 모델 기반의 협력적 여과 방법 (Model-Based Collaborative Filtering)은 추천은 매우 빠르지만, 모델을 구축하는 데 많은 시간이 소요되며, 사용자 기반의 협력적 여과 방법에 비해 추천의 질이 떨어지는 경향이 있다. 또한, 과거에 추천되있던 히스토리를 바탕으로 한 신뢰도 정보를 고려하는 추천 시스템은 추천의 정확도를 향상시키기 위한 다양한 연구 가운데 하나이다. 본 논문에서는 사용자 기반의 협력적 여과 방법의 문제점을 개선하고 추천의 정확도를 높이기 위해, 유사한 아이템의 모델을 미리 구축하는 아이템 기반의 협력적 여과 방법 (Item-Based Collaborative Filtering)에 각 아이템의 추천에 대한 신뢰도를 고려하여 보다 효율적인 추천 시스템을 제안하고자 한다. 또한, 기존 추천 시스템과의 성능 비교 실험을 통해 제안한 방법의 타당성을 제시한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10e
• /
• pp.310-312
• /
• 2002

인터넷 보급의 확산 및 네트웍 성능의 향상으로 멀티캐스트 기술을 응용한 서비스가 일대다의 정보 전달 서비스 위주에서 다중 사용자 네트웍 게임 (MUG) 등과 같이 인터넷을 사용한 다자간 상호작용 서비스로 그 응용 서비스 범위를 넓혀가게 되었다. 본 논문에서는 이러한 다자간 상호작용을 지원하기 위한 다대다 멀티캐스트 세션을 위한 신뢰성 있는 멀티 캐스트 프로토콜에서 발생하는 비용을 기술하고, 기존의 일대다 멀티캐스트 세션을 위해 제안된 신뢰성 있는 멀티캐스트 프로토콜들(RMTP, TMTP)과 다대다 멀티캐스트 세션을 위해 제안된 프로토콜들(SRM, Lorax. GAM)에 대 한 시뮬레이션을 통하여 에러 복구 기법의 처리비용을 측정함으로써 다대다 멀티캐스트 세션에 적합한 신뢰성 있는 멀티캐스트프로토콜을 제안한다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.170-175
• /
• 2022

일반적으로 대규모 언어 모델들은 다량의 데이터를 오랜시간 사전학습하면서 레이블을 예측하기 위한 성능을 높여왔다. 최근 언어 모델의 레이블 예측에 대한 정확도가 높아지면서, 언어 모델이 왜 해당 결정을 내렸는지 이해하기 위한 신뢰도 높은 Natural Language Explanation(NLE) 을 생성하는 것이 시간이 지남에 따라 주요 요소로 자리잡고 있다. 본 논문에서는 높은 레이블 정확도를 유지하면서 동시에 언어 모델의 예측에 대한 신뢰도 높은 explanation 을 생성하는 참신한 자연어 추론 시스템을 제시한 Natural-language Inference over Label-specific Explanations(NILE)[1] 을 소개하고 한국어 데이터셋을 이용해 NILE 과 NLE 를 활용하지 않는 일반적인 자연어 추론 태스크의 성능을 비교한다.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.5
• /
• pp.410-419
• /
• 2004

In this paper. we propose a method of confidence measure fusion under Bayesian framework for speech recognition. Centralized and distributed schemes are considered for confidence measure fusion. Centralized fusion is feature level fusion which combines the values of individual confidence scores and makes a final decision. In contrast. distributed fusion is decision level fusion which combines the individual decision makings made by each individual confidence measuring method. Optimal Bayesian fusion rules for centralized and distributed cases are presented. In isolated word Out-of-Vocabulary (OOV) rejection experiments. centralized Bayesian fusion shows over 13% relative equal error rate (EER) reduction compared with the individual confidence measure methods. In contrast. the distributed Bayesian fusion shows no significant performance increase.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.4
• /
• pp.16-22
• /
• 2000

In order to predict the performance and the usefulness of the narrow-band underwater acoustic measurement system at design stage, whose error variance is not clearly described, in this study a boundary equation to estimate the measurement accuracy is proposed based on the confidency as SNR variation. The boundary is presented as a function of SNR and the number of samples. In this paper, the measurement performance for narrow-band signal is simulated by the proposed boundary equation and the results are reviewed in the biased noise condition and separately in the background noise rejected condition.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.6A
• /
• pp.607-613
• /
• 2008

In this paper, we investigate the performance of SMP-assisted 3GPP GERAN evolution system employing high order modulation and turbo coding. When applying the SMP which maps systematic bits into highly reliable bit positions, it is confirmed that there is the performance gain for the modulation and coding schemes of 16QAM(DAS-8) as well as 32QAM(DAS-11) by link level simulation.

• Journal of KIISE:Computing Practices and Letters
• /
• v.12 no.4
• /
• pp.229-240
• /
• 2006

The advanced Alarm System (AS) employs modem digital technology to implement the alarm functions of the NPP(Nuclear Power Plant). The use of modem digital technology can provide advanced alarm processing in which new algorithms such as a signal validation, advanced alarm processing logic and other features are applied to improve the control room man-machine interfaces. This paper will describe the design process of the AS of NPP, improving the system reliability and availability using the reliability prediction tool, design strategies regarding the human performance topics associated with a computer-based AS and the results of the performance analysis using a prototype of the AS.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.12
• /
• pp.125-132
• /
• 2010

In vocabulary recognition has unseen tri-phone appeared when recognition training. This system has not been created beginning estimation figure of model parameter. It's bad points could not be created that model for phoneme data. Therefore it's could not be secured accuracy of Gaussian model. To improve suggested Gaussian model to optimized method of model parameter using probability distribution. To improved of confidence that Gaussian model to optimized of probability distribution to offer by accuracy and to support searching of phoneme data. This paper suggested system performance comparison as a result of recognition improve represent 1.7% by out-of vocabulary rejection algorithm using normalization confidence.

• Journal of Aerospace System Engineering
• /
• v.18 no.1
• /
• pp.11-20
• /
• 2024

In the recent era of NewSpace, unlike high-reliability satellites of the past, low-reliability satellites are being developed and mass-produced at a lower cost to launch constellations satellites. To achieve cost-effective cluster satellite development, satellite users and developers need to assess the feasibility of maintaining mission performance over the expected lifespan when cluster satellites are launched. Plans for replacements due to random failures should also be established to maintain performance. This study proposed a method for assessing system reliability and availability to maintain mission performance and establish replacement strategies for Earth observation constellation satellites. In this study, a constellation reliability and availability model considering mission performance required for a satellite constellation, situations of satellite backup, and additional ground backups was established. The reliability model was structured based on the concept of a k-out-of-n system and the availability model used a Markov chain model. Based on the proposed reliability model, the minimum number of satellites required to meet mission requirements was defined and satellites needed in orbit during the required mission period to satisfy mission reliability were calculated. This research also analyzed the number of spare satellites in orbit and on the ground required to meet the desired availability during required service period through availability analysis.

• Journal of Korea Multimedia Society
• /
• v.14 no.1
• /
• pp.85-93
• /
• 2011

Previous existing vocabulary recognition programs calculate general vector values from a database, so they can not process phonemes that form during a search. And because they can not create a model for phoneme data, the accuracy of the Gaussian model can not secure. Therefore, in this paper, we recommend use of the Bhattacharyya distance measurement method based on the features of the phoneme-thus allowing us to improve the recognition rate by picking up accurate phonemes and minimizing recognition of similar and erroneous phonemes. We test the Gaussian model optimization through share continuous probability distribution, and we confirm the heighten recognition rate. The Bhattacharyya distance measurement method suggest in this paper reflect an average 1.9% improvement in performance compare to previous methods, and it has average 2.9% improvement based on reliability in recognition rate.