• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.1768-1769
• /
• 2011

The nonlinear estimation and the pseudo-linear estimation are used to treat the target localization in sensor network which provides range difference of arrival (RDOA) measurements. It is known that the nonlinear estimation has sensitive problem for the initial estimate and the pseudo-linear estimation has a large estimation error. The QCLS method is the typical estimator of the methods for pseudo-linear estimation. However the estimate by using the QCLS method includes the estimation error because the first stage of two estimation processes of the QCLS method causes the biased estimation error. Therefore we propose a instrumental variables(IV) method for minimizing the estimation error of the first stage. The simulation shows that the performance of the proposed method is superior to the QCLS method.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.29B no.10
• /
• pp.90-99
• /
• 1992

In this paper, we propose a neural network model for speech recognition. The model consists of feature extraction parts and recognition parts. The interconnection model based on ${\Delta}^2$G operator was used for frequency analysis. Two features, global feature and local feature, were extracted from this model. Recognition parts consist of global grouping stage and local grouping stage. When the input pattern was coded by slope method, the recognition rate of speakers, A and B, was 100%. When the test was performed with the data of 9 speakers, the recognition rate of 91.4% was obtained.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.7B
• /
• pp.573-582
• /
• 2006

In this paper, we propose a hybrid two-stage AWG-based WDM-PON architecture in order to overcome the limitations of the existing PONs and single AWG-based WDM-PONs as well as to accomodate the new services and the expandability of network. The proposed architecture employs two-stage AWG for downstream transmission and single AWG and combiners for upstream one at RN. It also employs the separated fiber with multi-wavelength for both direction. It leads to high scalability, low cost, and high capacity for transmission. In downstream transmission, the transparency can be guaranteed since the traffic is transmitted to ONU through each channel. However, several ONUs share the channel for upstream one by using WDM/TDMA scheme because the asymmetrical feature of networks is considered. The performance of the proposed one is evaluated and compared with other architectures in terms of cost, network capacity and up/downstream bandwidth.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10b
• /
• pp.1961-1965
• /
• 1991

This paper deals with an automatic and precision alignment technique for proximity printing in x-ray lithography, using two pairs of moire gratings, with moire signals from each pair being 180.deg. out of phase with each other. We constructed an automatic and precision alignment experimental system which could measure both transmitted moire signals and reflected moire signals at the same time. The automatic alignment was achieved using transmitted moire signals and also reflected moire signals as a control signal for a stage driver. The alignment position of the system was monitored not only by a control signal but also by a non-control signal. The effect of transmitted and reflected moire signals upon alignment accuracy was discussed. We concluded that the technique using diffracted moire signals is a viable automatic and precision alignment technique.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2000.11a
• /
• pp.313-318
• /
• 2000

Image segmentation is to divide an image into similar parts or objects. This paper presents a segmentation system which consists of a fuzzy neural network and a set of image processing filters. The fuzzy neural network does not need initialization of weights. Therefore it does not have the underutilization problem. This fuzzy neural network controls the size and number of clusters by the vigilance parameter instead of fixing the number of clusters at the initial stage. This fuzzy neural network does not require large amount of memory as in Fuzzy c-Means algorithm. Two satellite images were segmented using the proposed system. The segmented results show that the proposed system is better on segmenting images.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.

• International Journal of Computer Science & Network Security
• /
• v.21 no.9
• /
• pp.179-184
• /
• 2021

In our study we performed a sentiments analysis from the images. For this purpose, we used 153 images that contain: people, animals, buildings, landscapes, cakes and objects that we divided into two categories: images that suggesting a positive or a negative emotion. In order to classify the images using the two categories, we created two models. The SAI-G model was created with Google's AutoML Vision service. The SAI-C model was created on the Clarifai platform. The data were labeled in a preprocessing stage, and for the SAI-C model we created the concepts POSITIVE (POZITIV) AND NEGATIVE (NEGATIV). In order to evaluate the performances of the two models, we used a series of evaluation metrics such as: Precision, Recall, ROC (Receiver Operating Characteristic) curve, Precision-Recall curve, Confusion Matrix, Accuracy Score and Average precision. Precision and Recall for the SAI-G model is 0.875, at a confidence threshold of 0.5, while for the SAI-C model we obtained much lower scores, respectively Precision = 0.727 and Recall = 0.571 for the same confidence threshold. The results indicate a lower classification performance of the SAI-C model compared to the SAI-G model. The exception is the value of Precision for the POSITIVE concept, which is 1,000.

• International Journal of Computer Science & Network Security
• /
• v.23 no.8
• /
• pp.40-48
• /
• 2023

Speech Emotions recognition has become the active research theme in speech processing and in applications based on human-machine interaction. In this work, our system is a two-stage approach, namely feature extraction and classification engine. Firstly, two sets of feature are investigated which are: the first one is extracting only 13 Mel-frequency Cepstral Coefficient (MFCC) from emotional speech samples and the second one is applying features fusions between the three features: Zero Crossing Rate (ZCR), Teager Energy Operator (TEO), and Harmonic to Noise Rate (HNR) and MFCC features. Secondly, we use two types of classification techniques which are: the Support Vector Machines (SVM) and the k-Nearest Neighbor (k-NN) to show the performance between them. Besides that, we investigate the importance of the recent advances in machine learning including the deep kernel learning. A large set of experiments are conducted on Surrey Audio-Visual Expressed Emotion (SAVEE) dataset for seven emotions. The results of our experiments showed given good accuracy compared with the previous studies.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.19 no.1
• /
• pp.19-27
• /
• 2007

Analytical investigation on the performance of a two stage twin rotary compressor for $CO_2$ heat pump water heater system has been carried out. A computer simulation program was made based on analytical models for gas compression in control volumes, leakages among neighboring volumes, and dynamics of moving elements of the compressor. Calculated cooling capacity, compressor input, and COP were well compared to those of experiments over the compressor speeds tested. For the operating condition of suction pressure of 3 MPa, and discharge pressure of 9 MPa, and compressor inlet temperature of $35^{\circ}C$, the compressor efficiency was calculated to be 80.2%: volumetric, adiabatic, and mechanical efficiencies were 88.3%, 93.2%, and 92.7%, respectively. For the present compressor model, volumetric and adiabatic efficiencies of the second stage cylinder were lower by about $6{\sim}7%$ than those of the first stage mainly due to the smaller discharge port at the second stage. Parametric study on the discharge port size showed that the compressor performance could be improved by 3.5% just by increasing the discharge port diameter by 20%.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.288-295
• /
• 1998

This paper proposes a new data classification method based on the tolerant rough set that extends the existing equivalent rough set. Similarity measure between two data is described by a distance function of all constituent attributes and they are defined to be tolerant when their similarity measure exceeds a similarity threshold value. The determination of optimal similarity theshold value is very important for the accurate classification. So, we determine it optimally by using the genetic algorithm (GA), where the goal of evolution is to balance two requirements such that (1) some tolerant objects are required to be included in the same class as many as possible. After finding the optimal similarity threshold value, a tolerant set of each object is obtained and the data set is grounded into the lower and upper approximation set depending on the coincidence of their classes. We propose a two-stage classification method that all data are classified by using the lower approxi ation at the first stage and then the non-classified data at the first stage are classified again by using the rough membership functions obtained from the upper approximation set. We apply the proposed classification method to the handwritten numeral character classification. problem and compare its classification performance and learning time with those of the feed forward neural network's back propagation algorithm.