• The Journal of the Acoustical Society of Korea
• /
• v.23 no.3
• /
• pp.242-247
• /
• 2004

Conventional keyword spotting systems use the connected word recognition network consisted by keyword models and filler models in keyword spotting. This is why the system can not construct the language models of word appearance effectively for detecting keywords in large vocabulary continuous speech recognition system with large text data. In this paper to solve this problem, we propose a keyword spotting system using pseudo N-gram language model for detecting key-words and investigate the performance of the system upon the changes of the frequencies of appearances of both keywords and filler models. As the results, when the Unigram probability of keywords and filler models were set to 0.2, 0.8, the experimental results showed that CA (Correctly Accept for In-Vocabulary) and CR (Correctly Reject for Out-Of-Vocabulary) were 91.1% and 91.7% respectively, which means that our proposed system can get 14% of improved average CA-CR performance than conventional methods in ERR (Error Reduction Rate).

• The Journal of Korean Institute of Information Technology
• /
• v.17 no.5
• /
• pp.13-20
• /
• 2019

Recently, object detection technology based on CNN has been actively studied. Object detection technology is used as an important technology in autonomous vehicles, intelligent image analysis, and so on. In this paper, we propose a rotation change robust object detection system by applying MI-FL (Moment Invariant-Feature Layer) to SSD (Single Shot Multibox Detector) which is one of CNN-based object detectors. First, the features of the input image are extracted based on the VGG network. Then, a total of six feature layers are applied to generate bounding boxes by predicting the location and type of object. We then use the NMS algorithm to get the bounding box that is the most likely object. Once an object bounding box has been determined, the invariant moment feature of the corresponding region is extracted using MI-FL, and stored and learned in advance. In the detection process, it is possible to detect the rotated image more robust than the conventional method by using the previously stored moment invariant feature information. The performance improvement of about 4 ~ 5% was confirmed by comparing SSD with existing SSD and MI-FL.

• Journal of the Korean Society of Food Science and Nutrition
• /
• v.43 no.8
• /
• pp.1304-1316
• /
• 2014

The aims of this study were to estimate daily intakes of artificial sweeteners from beverages and liquid teas as well as evaluate their potential health risks in Korean children and adolescents (1 to 19 years old). Dietary intake assessment was conducted using actual levels of aspartame, acesulfame-K, and sucralose in non-alcoholic beverages (651 beverages and 87 liquid teas), and food consumption amounts were drawn from "The Fourth Korea National Health and Nutrition Examination Survey (2007~2009)". To estimate dietary intake of non-alcoholic beverages, a total of 6,082 children and adolescents (Scenario I) were compared to 1,704 non-alcoholic beverage consumption subjects (Scenario II). The estimated daily intake of artificial sweeteners was calculated based on point estimates and probabilistic estimates. The values of probabilistic artificial sweeteners intakes were presented by a Monte Carlo approach considering probabilistic density functions of variables. The level of safety for artificial sweeteners was evaluated by comparisons with acceptable daily intakes (ADI) of aspartame (0~40 mg/kg bw/day), acesulfame-K (0~15 mg/kg bw/day), and sucralose (0~15 mg/kg bw/day) set by the World Health Organization. For total children and adolescents (Scenario I), mean daily intakes of aspartame, acesulfame-K, and sucralose estimated by probabilistic estimates using Monte Carlo simulation were 0.09, 0.01, and 0.04 mg/kg bw/day, respectively, and 95th percentile daily intakes were 0.30, 0.02, and 0.13 mg/kg bw/day, respectively. For consumers-only (Scenario II), mean daily intakes of aspartame, acesulfame-K, and sucralose estimated by probabilistic estimates using Monte Carlo simulation were 0.52, 0.03, and 0.22 mg/kg bw/day, respectively, and 95th percentile daily intakes were 1.80, 0.12, and 0.75 mg/kg bw/day, respectively. For scenarios I and II, neither aspartame, acesulfame-K, nor sucralose had a mean and 95th percentile intake that exceeded 5.06% of ADI.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2010.07a
• /
• pp.53-56
• /
• 2010

본 논문은 Shannon의 정리에 따른 채널 용량에 근접한 성능을 보이는 것으로 알려진 터보 복호기 기반의 반복적인 검출과 복호화(Iterative Detection and Decoding) 기법에서 반복적인 복호화를 수행할 시에 제외되었던 리스트 구 복호기(List Sphere Decoder)에서 사전 정보(prior information)을 이용할 수 있도록 하여 수정된 IDD 기법을 제안하였다. 기존의 기법에서는 사후확률(A posteriori probability)을 계산하기 위하여 리스트 구 복호기를 사용하였으나 반복적인 복호화 수행 시에는 사전 정보를 이용하지 않는 특성으로 인하여 제외된다. 만약 잡음(noise) 등의 이유로 검출된 심볼 벡터 목록이 원래의 것과 매우 다른 경우라도 재 검출을 하지 않기 때문에 반복적인 복호화를 수행하더라도 원래의 정보에 근접하기 어렵게 된다. 본 논문에서는 이러한 기존의 기법에서 리스트 구 복호기를 터보 복호기의 Log Likelihood Ratio (LLR) 값을 사전 정보로 이용할 수 있도록 수정된 리스트 구 복호기를 제안하였다. 수정된 리스트 복호기는 반복적인 복호화를 수행 시 이전의 복호화에서 얻은 정보를 이용하여 새로이 검출된 심볼 벡터 목록을 제공하게 된다. 실제의 통신환경과 유사한 모델의 실험을 통해 수정된 IDD 기법이 기존의 IDD로 구성되는 내부 피드백에 RS 복호기 기반의 외부 피드백으로 구성된 형태로 피드백 회수가 증가할수록 기존의 IDD에 비해 성능이 개선됨을 확인하였다.

• The Journal of the Korea Contents Association
• /
• v.16 no.5
• /
• pp.104-112
• /
• 2016

The Adaboost chooses a good set of features in rounds. On each round, it chooses the optimal feature and its threshold value by minimizing the weighted error of classification. The involved process of classification performs a hard decision. In this paper, we expand the process of classification to a soft fuzzy decision. We believe this expansion could allow some flexibility to the Adaboost algorithm as well as a good performance especially when the size of a training data set is not large enough. The typical Adaboost algorithm assigns a same weight to each training datum on the first round of a training process. We propose a new algorithm to assign different initial weights based on some statistical properties of involved features. In experimental results, we assess that the proposed method shows higher performance than the traditional one.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.13 no.6
• /
• pp.87-92
• /
• 2013

This paper focuses on analyzing the effects of primary user (PU) signal arrival on the local spectrum sensing performance. The probability for signal arrival at a sample in the sensing time is uniformly distributed in the sensing time. We first analyze the main factor that causes the degradation in the detection results in the conventional energy detection (CED) under the uniformly random arrival of the PU-signal. Thus we propose an approach in order to enhance the detection performance, in which an estimator which detects the arrival of the PU signal cooperates with a composite energy detection. The mathematical analysis and numerical simulation has validated the outperformance of the proposed approach compared to the CED.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.10 no.3
• /
• pp.79-84
• /
• 2011

In this paper, we analyzed performance of vehicle detection for ITS (Intelligent Transport System) applications. We simulated the vehicle detection at Hi-Pass System is based on DSRC (Dedicated Short Range Communication). DSRC is a wireless network using ITS, including GPS (Global Positioning System) satellites in conjunction with the national transportation system. The system performance is evaluated in terms of bit error probability. In the simulation, the vehicle speed is set at 60 km/h and carrier frequency is 5.8 GHz. Wireless channel is modeled as the Rician fading channel. In the transmitter, the ASK (amplitude shift keying) modulation scheme is applied. From simulation results, we confirmed that performance of applied Alamouti scheme is better than other systems.

• The KIPS Transactions:PartB
• /
• v.12B no.7 s.103
• /
• pp.721-728
• /
• 2005

This paper suggests effective object detection system for moving objects with specified color and motion information. The proposed detection system includes the object extraction and definition process which uses MCN(Multiple Color Normalization) and MCWUPC(Moving Color Weighted Unmatched Pixel Count) computation to decide the existence of moving object and object segmentation technique using signature information is used to exactly extract the objects with high probability. Finally, real time detection system is implemented to verify the effectiveness of the technique and experiments show that the success rate of object tracking is more than $89\%$ of total 120 image frames.

• Korean Journal of Optics and Photonics
• /
• v.12 no.2
• /
• pp.71-76
• /
• 2001

We studied quantum cryptography based on the quantum nature of light. We must reduce the intensity of the light pulse to the single photon regime for quantum cryptographic communication. Considering the noise and the quantum efficiency of the detector, however, we have to fmd a criterion for which we are able to distinguish the error caused by eavesdropping from other system noises. By changing the bias voltage of the detector and the threshold of the signal voltage, we find the safe region for which we can distribute the quantum key with positive proof of no-eavesdropping. The quantum key we used is a four state quantum key (BB84). BB84).

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.836-839
• /
• 2012

본 논문에서는 운전자의 안전을 보장하기 위해 영상 처리 기술을 기반으로 도로 정보를 검출해 운전자에게 알려주고, 버튼을 직접 손으로 눌러야 하는 물리적 인터페이스를 대체할 차세대 인터페이스 기술을 제안한다. 제안된 기술은 카메라 한대에서 입력 받은 영상 정보를 제안된 알고리즘을 통해 앞차와의 거리, 차선, 교통 표지판을 검출하고 차량 내부를 주시하는 카메라와 운전자의 음성을 인식할 마이크를 기반으로 음성인식과 동작 인식이 결합된 인터페이스를 제공한다. 본 논문에서 개발된 기술을 통해 설제 테스트를 실시해 본 결과 표지판인식, 차선검출, 앞차와의 거리 검출 등의 인식률이 약 90% 이상이었으며, 이러한 기술적 요소들은 운전자가 인지하지 못하는 상황 등에서도 적절한 정보를 운전자에게 제공해 줌으로써 교통사고 확률을 크게 낮출 수 있을 것으로 기대된다.