• Journal of the Korean Professional Engineers Association
• /
• v.22 no.3
• /
• pp.5-13
• /
• 1989

A major problem of stopwatch time study is how to do for the accurate and consistent performance rating, which is one of the critical variables to determine the accuracy of work measurement and should be still dependent upon time observer's judgement. Therefore the time observer's ability for the performance rating is very important, and must be improved by correct training method and procedure. This paper developed a new benchmark and benchmark-observation method for the effective performance rating training of assembling and machining operations. The trainees' ability in the accuracy and consistency of the performance rating ,improved significantly after being trained by subject method. The percentage improvement in rating accuracy and consistency values was 34.7% and 49% respectively. In addition, benchmark-practice method for the performance rating training is not significant, so it is proofed that the skill of a certain operation is not important for the improvement of the rating ability.

• Phonetics and Speech Sciences
• /
• v.6 no.2
• /
• pp.29-34
• /
• 2014

In the Vector Taylor Series (VTS)-based noisy speech recognition methods, Hidden Markov Models (HMM) are usually trained with clean speech. However, better performance is expected by training the HMM with noisy speech. In a previous study, we could find that Minimum Mean Square Error (MMSE) estimation of the training noisy speech in the log-spectrum domain produce improved recognition results, but since the proposed algorithm was done in the log-spectrum domain, it could not be used for the HMM adaptation. In this paper, we modify the previous algorithm to derive a novel mathematical relation between test and training noisy speech in the cepstrum domain and the mean and covariance of the Multi-condition TRaining (MTR) trained noisy speech HMM are adapted. In the noisy speech recognition experiments on the Aurora 2 database, the proposed method produced 10.6% of relative improvement in Word Error Rates (WERs) over the MTR method while the previous MMSE estimation of the training noisy speech produced 4.3% of relative improvement, which shows the superiority of the proposed method.

• Hwankyungkyoyuk
• /
• v.14 no.2
• /
• pp.68-75
• /
• 2001

The purpose of study is to provide fundamental data for the improvement of the teacher inservice training for environmental education through analysis of current inservice training programmes. The subject of analysis are documents on training programmes which was conducted after 2000 by 10 training organizations. Based on the results of this study, inservice training programmes is classified with 6 organizations which consist of education training institute, education & scientific research institute, national · public organizations, colleges of an attached organizations, civil organizations, teacher research society. The strategies for improvement of proposed in this study can be summarized as follows: First,'60 hours training programmes for general competencies improvement of environmental teacher' have to reconsider about scarcity areas to analysis of programmes. Second, this training programmes need to establish in training programmes of nothing region for increase in training opportunity of teachers. Third,'the core training programmes'is continued to be complementing about this programmes and need to establish about training programmes of teaching method of environmental education, environmentally value and attitude, etc

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.7
• /
• pp.170-178
• /
• 1994

The nearest neighbor classification rule is widely used because it is not only simple but the error rate is asymptotically less than twice Bayes theoretical minimum error. But the method basically use the whole training patterns as the reference vectors. so that both storage and classification time increase as the number of training patterns increases. LVQ(Learning Vector Quantization) resolved this problem by training the reference vectors instead of just storing the whole training patterns. But it is a heuristic algorithm which has no theoretic background there is no terminating condition and it requires a lot of iterations to get to meaningful result. This paper is to propose a new training method of the reference vectors. which minimize the given error function. The nearest neighbor network,the network version of the nearest neighbor classification rule is proposed. The network is funtionally identical to the nearest neighbor classification rule is proposed. The network is funtionally identical to the nearest neighbor classification rule and the reference vectors are represented by the weights between the nodes. The network is trained to minimize the error function with respect to the weights by the steepest descent method. The learning algorithm is derived and it is shown that the proposed method can adjust more reference vectors than LVQ in each iteration. Experiment showed that the proposed method requires less iterations and the error rate is smaller than that of LVQ2.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.10
• /
• pp.142-149
• /
• 2013

In general, a color of an image is represented by using red, green, and blue channels in a RGB camera system. However, only information of three channels are limited to estimate a spectral reflectance of a real scene. Because of this, the RGB camera system can not accurately represent the color. To overcome this limitation and represent an accurate color, researches to estimate the spectral reflectance by using a multi-channel camera system are being actively proceeded. Recently, a reflectance estimation method adaptively constructing a similar training set from a traditional training set according to a camera response by using a spectral similarity was introduced. However, in this method, an accuracy of the similar training set is reduced because the spectral similarity based on an average and a maximum distances was applied. In this paper, a reflectance estimation method applied a spectral similarity based on a correlation coefficient is proposed to improve the accuracy of the similar training set. Firstly, the correlation coefficient between the similar training set and the spectral reflectance obtained by Wiener estimation method is calculated. Secondly, the similar training set is constructed from the traditional training set according to the correlation coefficient. Finally, Wiener estimation method applied the similar training set is performed to estimate the spectral reflectance. To evaluate a performance of the proposed method with previous methods, experimental results are compared. As a result, the proposed method showed the best performance.

• International Journal of Internet, Broadcasting and Communication
• /
• v.12 no.2
• /
• pp.45-50
• /
• 2020

A Language model with neural networks commonly trained with likelihood loss. Such that the model can learn the sequence of human text. State-of-the-art results achieved in various language generation tasks, e.g., text summarization, dialogue response generation, and text generation, by utilizing the language model's next token output probabilities. Monotonous and boring outputs are a well-known problem of this model, yet only a few solutions proposed to address this problem. Several decoding techniques proposed to suppress repetitive tokens. Unlikelihood training approached this problem by penalizing candidate tokens probabilities if the tokens already seen in previous steps. While the method successfully showed a less repetitive generated token, the method has a large memory consumption because of the training need a big vocabulary size. We effectively reduced memory footprint by encoding words as sequences of subword units. Finally, we report competitive results with token level unlikelihood training in several automatic evaluations compared to the previous work.

• Journal of The Korean Society of Integrative Medicine
• /
• v.1 no.2
• /
• pp.13-22
• /
• 2013

Purpose : The purpose of the research is that get a cut above clinical practice effect through satisfaction of clinical training, practical training, content, oversight of training and evaluation system. Clinical training consists of part of university in Gwang Ju and Jeon nam. Method : The target of training student was studying at physiotherapy a tree or four-year-course collage in Gwang ju and Jean nam. Data collection period is from 21 November 2012 to 1 February. We explained how to do a means of collecting data and get students consent fill in questionnaire. Data collection prossed by using spss 10.1 program also independent proofs, descriptive statistics, crosstabulation, regression analysis and frequency analysis. Results : The subjects average age is 24 in general characteristic. A school system of subjects was a tree-year-course students. They were 58people(39.1%). A school system of subjects was a four-year-course students. They were 90people(60.9%).The male was 72(48.6%) and the female was 76(51.4%). We researched to know about satisfaction of clinical training, practical training, content, environment of practical establishment, trainee manage and evaluation method. All-round satisfaction of clinical training average was 1.90 Satisfaction of clinical training period and content average was 1.83Satisfaction of environment of practical establishment average was 1.88 Satisfaction of clinical training establishments' trainee manage and evaluation average was 1.94 Conclusion : It is important that student can get specific their future and can do at clinical throught clinical training after their graduation improving satisfaction of clinical training would give to impact a physical therapist reserve.

• Proceedings of the KSPS conference
• /
• 2005.11a
• /
• pp.159-162
• /
• 2005

Identification is the process automatically identify who is speaking on the basis of information obtained from speech waves. In training phase, each speaker models are trained using each speaker's speech data. GMMs (Gaussian Mixture Models), which have been successfully applied to speaker modeling in text-independent speaker identification, are not efficient in insufficient training data environment. This paper proposes speaker modeling method using MLLR (Maximum Likelihood Linear Regression) method which is used for speaker adaptation in speech recognition. We make SD-like model using MLLR adaptation method instead of speaker dependent model (SD). Proposed system outperforms the GMMs in small training data environment.

• The Journal of the Korea Contents Association
• /
• v.12 no.4
• /
• pp.263-272
• /
• 2012

In an attempt to analyze the policy priority on civil servant's education training policy as human resource management in this study, a positive analysis with the experts using AHP method was conducted. Summarizing the outcome of the study; First, in terms of relative priority of the evaluation elements by sector, the education training operation system, among education training program, education training evaluation, and education infrastructure, was analyzed as the most important element. Second, as a result of analyzing the priority of detail sectors of civil servant's education training, Action learning education program was proved to be the top priority project education training program aspect and education training operation system was also evaluated as the top priority project in education training agency budget expansion aspect, education training and personnel administration link was evaluated as the top priority project education training evaluation, and trainer secure was proved to be the top priority education infrastructure. Such outcome of the project is expected to make commitment for evaluating the civil servant's education training policy.

• Journal of IKEEE
• /
• v.22 no.3
• /
• pp.816-821
• /
• 2018

Recently, machine learning algorithms based on artificial neural networks started to be used widely as classifiers in the field of EEG research for emotion analysis and disease diagnosis. When a machine learning model is used to classify EEG data, if training data is composed of only data having similar characteristics, classification performance may be deteriorated when applied to data of another group. In this paper, we propose a method to construct training data set by selecting several groups of data using semi-supervised learning algorithm to improve these problems. We then compared the performance of the two models by training the model with a training data set consisting of data with similar characteristics to the training data set constructed using the proposed method.