• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.7
• /
• pp.800-804
• /
• 2005

This paper presents a fuzzy learning rule which is the fuzzified version of LVQ(Learning Vector Quantization). This fuzzy learning rule 3 uses fuzzy learning rates. instead of the traditional learning rates. LVQ uses the same learning rate regardless of correctness of classification. But, the new fuzzy learning rule uses the different learning rates depending on whether classification is correct or not. The new fuzzy learning rule is integrated into the improved IAFC(Integrated Adaptive Fuzzy Clustering) neural network. The improved IAFC neural network is both stable and plastic. The iris data set is used to compare the performance of the supervised IAFC neural network 3 with the performance of backprogation neural network. The results show that the supervised IAFC neural network 3 is better than backpropagation neural network.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.23 no.9
• /
• pp.814-822
• /
• 2013

For the fault diagnosis of a mechanical system, pattern recognition methods have being used frequently in recent research. Hidden Markov model(HMM) and artificial neural network(ANN) are typical examples of pattern recognition methods employed for the fault diagnosis of a mechanical system. In this paper, a hybrid method that combines HMM and ANN for the fault diagnosis of a mechanical system is introduced. A rotating blade which is used for a wind turbine is employed for the fault diagnosis. Using the HMM/ANN hybrid model along with the numerical model of the rotating blade, the location and depth of a crack as well as its presence are identified. Also the effect of signal to noise ratio, crack location and crack size on the success rate of the identification is investigated.

• Journal of Korean Society of Transportation
• /
• v.14 no.4
• /
• pp.49-76
• /
• 1996

Until recently, the inductive loop detecters(ILD) have been used to collect a traffic information in a part of traffic manangment and control. The ILD is able to collect a various traffic data such as a occupancy time and non-occupancy time, traffic volume, etc. The occupancy time of these is very important information for traffic control algorithms, which is required a high accuracy. This accuracy may be improved by classifying a vehicle type with ILD. To classify a vehicle type based on a Analog Digital Converted data collect form ILD, this study used a typical and modifyed statistic method and General Learning Vector Quantization unsuperviser neural network model and a hybrid model of GLVQ and statistic method, As a result, the hybrid model of GLVQ neural network model is superior to the other methods.

• Journal of Digital Convergence
• /
• v.11 no.2
• /
• pp.323-330
• /
• 2013

The trend of the use of open softwares is increasing daily. Despite these increases in the use of open softwares, the importance of the problem of their quality is not considered enough. Also, if we look at the existing evaluation model which is used as the evaluation model of the existing open software, the level of the quantization is insufficient, the evaluator can show some subjectivity and as a clear evaluation method is not published, there are many problems to apply it in practice. Therefore, in this study, we have built an evaluation basis that can show consistent quantization without the influence of the evaluator and that can improve the evaluation speed with the automatization of the evaluation information acquisition. The built quality evaluation model will take an important role to evaluate and to improve the open software.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.4
• /
• pp.561-567
• /
• 2013

Some of the feature vectors for speech recognition are rare and unusual. These patterns lead to overfitting for the parameters of the speech recognition system and, as a result, cause structural risks in the system that hinder the good performance in recognition. In this paper, as a method of removing these unusual patterns, we try to exclude vectors whose norms are larger than a specified cutoff value and then train the speech recognition system. The objective of this study is to exclude as many unusual feature vectors under the condition of no significant degradation in the speech recognition error rate. For this purpose, we introduce a cutoff parameter and investigate the resultant effect on the speaker-independent speech recognition of isolated words by using FVQ(Fuzzy Vector Quantization)/HMM(Hidden Markov Model). Experimental results showed that roughly 3%~6% of the feature vectors might be considered as unusual, and therefore be excluded without deteriorating the speech recognition accuracy.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.9 no.4
• /
• pp.964-969
• /
• 2008

This paper proposes a HMM-based recognition method using DMSVQ(Dynamic Multi-Section Vector Quantization) codebook by DMS(Dynamic Multi-Section) model and fuzzy concept, as a study for speaker- independent speech recognition. In this proposed recognition method, training data are divided into several dynamic section and multi-observation sequences which are given proper probabilities by fuzzy rule according to order of short distance from DMSVQ codebook per each section are obtained. Thereafter, the HMM using this multi-observation sequences is generated, and in case of recognition, a word that has the most highest probability is selected as a recognized word. Other experiments to compare with the results of recognition experiments using proposed method are implemented as a data by the various conventional recognition methods under the equivalent environment. Through the experiment results, it is proved that the proposed method in this study is superior to the conventional recognition methods.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.3
• /
• pp.21-29
• /
• 2011

This paper presents a method that estimates distance and velocity of an object with reliability regardless of maneuver status of the target in stereo vision system. A stereo vision system can calculate a distance with disparity from left and right images. However, the distance estimation error may occur due to quantization error of image pixel. A sub-pixel interpolation method minimizes the quantization error and estimates accurate disparity with real value. Extended Kalman filter (EKF) was used to minimize the error covariance and estimate the object's velocity. However, divergence problem occurs due to model uncertainty when a target maneuvers highly, which makes the estimation error increase. In this paper, we propose a virtual model extended Kalman filter (VMEKF) method that minimizes the processing time and provides reliable estimation ability regardless of maneuver status. Computer simulations and experimental results in real road environments demonstrate that the proposed method gives a reliable estimation performance and reduces processing time under various maneuver status while comparing other estimation filters.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.6
• /
• pp.63-72
• /
• 2005

In this paper, we propose the bit-rate control algorithm for the block based image compression like H.263, H.263+ or MPEG-4. The proposed algorithm is designed to identify the quantization parameter set through the Lagrangian optimization technique based on the well-known linear source model. We set the Lagrangian cost function with the rates and distortion calculated from the linear source model. We calculate the quantization parameter set using the Vitervi algorithm to solve the Lagrangian optimization problem considering the Dquant method of H.263 and MPEG-4. The proposed algorithm improves the video quality by up to 1.5 dB compared with the TMN8 scheme, and is more effective in the video sources with dynamic activities than the consistent quality approaches.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.10
• /
• pp.2071-2078
• /
• 2009

The first frame is encoded in intra mode which generates a larger number of bits. In addition, the first frame is used for the inter mode encoding of the following frames. Thus the intial QP (Quantization Parameter) for the first frame affects the first frame as well as the following frames. Traditionally, the initial QP is determined among four constant values only depending on the bpp. In the case of low bit rate video coding, the initial QP value is fixed to 35 regardless of the output bandwidth. Although this initialization scheme is simple, yet it is not accurate enough. An accurate intial QP prediction scheme should not only depends on bpp but also on the complexity of the video sequence and the output bandwidth. In the proposed scheme, we use a linear model because there is a linear inverse proportional relationship between the output bandwidth and the optimal intial QP. Model parameters of the model are determined depending on the spatial complexity of the first frame. It is shown by experimental results that the new algorithm can predict the optimal initial QP more accurately and generate the PSNR performance better than that of the existing JM algorithm.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1330-1339
• /
• 2022

3D object detection generally aims to detect relatively large data such as automobiles, buses, persons, furniture, etc, so it is vulnerable to small object detection. In addition, in an environment with limited resources such as embedded devices, it is difficult to apply the model because of the huge amount of computation. In this paper, the accuracy of small object detection was improved by focusing on local features using only one layer, and the inference speed was improved through the proposed knowledge distillation method from large pre-trained network to small network and adaptive quantization method according to the parameter size. The proposed model was evaluated using SUN RGB-D Val and self-made apple tree data set. Finally, it achieved the accuracy performance of 62.04% at mAP@0.25 and 47.1% at mAP@0.5, and the inference speed was 120.5 scenes per sec, showing a fast real-time processing speed.