• KIISE Transactions on Computing Practices
• /
• v.22 no.7
• /
• pp.315-319
• /
• 2016

Recurrent neural network based language models (RNN LM) have shown improved results in language model researches. The RNN LMs are limited to post processing sessions, such as the N-best rescoring step of the wFST based speech recognition. However, it has considerable vocabulary problems that require large computing powers for the LM training. In this paper, we try to find the 1st pass N-gram model using word embedding, which is the simplified deep neural network. The class based language model (LM) can be a way to approach to this issue. We have built class based vocabulary through word embedding, by combining the class LM with word N-gram LM to evaluate the performance of LMs. In addition, we propose that part-of-speech (POS) tagging based LM shows an improvement of perplexity in all types of the LM tests.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.40 no.12
• /
• pp.1077-1085
• /
• 2016

Linear roller bearing (LRB) is an important mechanical element that is widely used in precise positioning systems that are subjected to large loads. This paper presents a new model for estimating the displacement of an LRB subjected to external forces. For this purpose, assuming that the linear motion block (LM block) is rigid, the equilibrium conditions for the LRB were obtained by solving the equilibrium equations of the rollers and the rigid LM block using the iterative Newton-Raphson method. The contact loads between the rollers and raceways were determined considering the profiled rollers. Then, the structural deformations of the LM block, subjected to the contact loads from the rigid LM block model, were computed using a finite element model for the LM block. The displacements of the LRB were then determined by superposition of the rigid LM block displacements on the induced displacements due to the structural deformations of the LM block. The proposed method was verified through comparison with a program by the bearing manufacturer.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.43 no.2 s.308
• /
• pp.11-19
• /
• 2006

In this paper, we propose a method to estimate camera extrinsic parameter based on projective and permutation invariance point features. Because feature informations in previous research is variant to c.:men viewpoint, extraction of correspondent point is difficult. Therefore, in this paper, we propose the extracting method of invariant point features, and new matching method using similarity evaluation function and Graham search method for reducing time complexity and finding correspondent points accurately. In the calculation of camera extrinsic parameter stage, we also propose two-stage motion parameter estimation method for enhancing convergent degree of LM algorithm. In the experiment, we compare and analyse the proposed method with existing method by using various indoor images to demonstrate the superiority of the proposed algorithms.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1302-1308
• /
• 2019

This paper presents an approach for classifying image made by acquired position data from a 2D Lidar sensor with a convolutional neural network (CNN). Lidar sensor has been widely used for unmanned devices owing to advantages in term of data accuracy, robustness against geometry distortion and light variations. A CNN algorithm consists of one or more convolutional and pooling layers and has shown a satisfactory performance for image classification. In this paper, different types of CNN architectures based on training methods, Gradient Descent(GD) and Levenberg-arquardt(LM), are implemented. The LM method has two types based on the frequency of approximating Hessian matrix, one of the factors to update training parameters. Simulation results of the LM algorithms show better classification performance of the image data than that of the GD algorithm. In addition, the LM algorithm with more frequent Hessian matrix approximation shows a smaller error than the other type of LM algorithm.

• Korean Journal of Organic Agriculture
• /
• v.26 no.2
• /
• pp.269-279
• /
• 2018

The present study was conducted to evaluate fertigation effects of balanced manure nutrients from combined fertilizers on the growth of eggplant (Solanum melongena L.) and soil chemical properties in greenhouse cultivation. There are 4 treatments for fertigation; animal liquid manure (LM), LM supplemented with chemical ferilizer (LM+CF) to make nutrient-balanced ferilizer, LM supplemented with chemical ferilizer and phosphorus acid (LM+CF+PA), and conventional chemical fertilzer (CCF). Fruit lenth, width, and weight for LM treatment were significantly lower than LM+CF, LM+CF+PA, and CCF treatments. There was no significant differences in fruit length, width, and weight among LM+CF, LM+CF+PA, and CCF treatments. The yield of eggplants per plant for LM treatment (26.2 kg) was decreased by 6% compared to CCF treatment (27.9 kg). The growth and yield of eggplants were not significant different among CCF, LM+CF, and LM+CF+PA treatments. The marketable yield for LM treatment was lowest among 4 treatments and that for LM+CF+PA treatment (26.6 kg) was increased by 8% compared to CCF treatment (24.1 kg). The chemical properties of the soil for CCF treatment, in general, tended to be higher compared to the other treatments. In conclusion, the present study demonstrates that the application of balanced liquid manure combined with mineral fertilizer to soil is considered as a good management practices because it improves, eggplant quality and soil properties.

• The Journal of Korean Association of Computer Education
• /
• v.6 no.1
• /
• pp.19-28
• /
• 2003

In this paper, we propose a development methodology based on the component architecture to develop the LMS that follows ADL's SCORM standard, which is interested internally and externally in the e-learning contents standardization research. In like manner, we designed the SCORM core module and implemented the EJB component based on J2EE through the sugested methodology for the conversion of SCORM Conformance LMS, from iOneLMS which was developed based on distributed object component like JSP, EJB, etc. The SCORM component modules were implemented by three modules - content importing, course registering, and taking course and gathering the learning data. In the process of implementation, we used the part of the class library that was supported by the ADL and AICC again.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.47 no.3
• /
• pp.76-82
• /
• 2010

In this paper, we develop Linguistic Model (LM) based on information granules as a systematic approach to generating fuzzy if-then rules from a given input-output data. The LM introduced by Pedrycz is performed by fuzzy information granulation obtained from Context-based Fuzzy Clustering(CFC). This clustering estimates clusters by preserving the homogeneity of the clustered patterns associated with the input and output data. Although the effectiveness of LM has been demonstrated in the previous works, it needs to improve in the sense of performance. Therefore, we focus on the automatic generation of linguistic contexts, addition of bias term, and the transformed form of consequent parameter to improve both approximation and generalization capability of the conventional LM. The experimental results revealed that the improved LM yielded a better performance in comparison with LM and the conventional works for automobile MPG(miles per gallon) predication and Boston housing data.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.4 s.36
• /
• pp.191-201
• /
• 2005

In this paper, we propose a method to calculate camera motion parameter, which is based on efficient invariant features irrelevant to the camera veiwpoint. As feature information in previous research is variant to camera viewpoint. information content is increased, therefore, extraction of accurate features is difficult. LM(Levenberg-Marquardt) method for camera extrinsic parameter converges on the goat value exactly, but it has also drawback to take long time because of minimization process by small step size. Therefore, in this paper, we propose the extracting method of invariant features to camera viewpoint and two-stage calculation method of camera motion parameter which enhances accuracy and convergent degree by using camera motion parameter by 2D homography to the initial value of LM method. The proposed method are composed of features extraction stage, matching stage and calculation stage of motion parameter. In the experiments, we compare and analyse the proposed method with existing methods by using various indoor images to demonstrate the superiority of the proposed algorithm.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.11
• /
• pp.2121-2130
• /
• 2015

In this paper, we analyze LM-MIMO (load-modulation multiple-input multiple-output) system with single RF chain. And then, we confirm that load modulation technique can support generation of high-order m-PSK modulation and m-QAM modulation in LM-MIMO system. Finally we evaluate performance of LM-MIMO system with load modulation. Conventional MIMO system requires a number of RF chains for expansion of MIMO dimension. A number of RF chains can cause various problems. On the other hand, although LM-MIMO system is expanded, LM-MIMO system requires single RF chain only. Therefore, LM-MIMO system has low-complexity and low power consumption. As results, we can confirm that load modulation of T-model can modulate high-order m-PSK and m-QAM singal. Also, we can confirm that $4{\times}4$ LM-MIMO system using load modulation has a similar performance to conventional $4{\times}4$ MIMO system.

• The Journal of the Korea Contents Association
• /
• v.12 no.2
• /
• pp.339-348
• /
• 2012

Acute promyelocytic leukemia (APL) is a cancer of the blood and bone marrow. Although all-trans retionic acid (ATRA) is the agents for ALP therapy, there are various side effects. For overcome this problem, we need the development of new therapeutic agents for APL. A number of bacteria produce various virulence factors with cytotoxic effects on human cancer cells. To understand the anti-cancer effect of Listeria monocytogenes on APL, we examined alteration of the cell viability, apoptosis and cell cycle arrest of the human promyelocytic leukemia cell line, HL-60 cells. The cell supernatant (LmSup) and the extract of L. monocytogenes (LmE) inhibited the cell viability and induced apoptosis of HL-60 cells. These cytotoxic effect of LmSup and LmE mediated by modulation of cell cycle and ROS production. These results indicate that released or included bacterial molecules from L. monocytogenes have a cytotoxicity in HL-60 cells. Therefore, LmSup and LmE may be used as the potential target for the treatment of cancer induced by HL-60 cells.