• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.10
• /
• pp.73-81
• /
• 2015

The paper proposes a prototype selection method and evaluates the generalization performance of standard algorithms and prototype based classification learning. The proposed prototype classifier defines multidimensional spheres with variable radii within class areas and generates a small set of training data. The nearest-neighbor classifier uses the new training set for predicting the class of test data. By decomposing bias and variance of the mean expected error value, we compare the generalization errors of k-nearest neighbor, Bayesian classifier, prototype selection using fixed radius and the proposed prototype selection method. In experiments, the bias-variance changing trends of the proposed prototype classifier are similar to those of nearest neighbor classifiers with all training data and the prototype selection rates are under 27.0% on average.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.5B
• /
• pp.443-448
• /
• 2009

A boundary value of the velocity of data gathering node (DGN) and a critical value for training overhead beyond which the scheme will not be feasible for a Multiple Input Multiple Output (MIMO) based cooperative communication for energy-limited wireless sensor networks is proposed in this paper. The performance in terms of energy efficiency and delay for a combination of two transmitting and two receiving antennas is analyzed. The results show that a set of critical value of velocity and training overhead pair is present for the long haul communication from the sensors to the data gathering node. Finally a relation between training overhead and velocity is simulated.

• Journal of Broadcast Engineering
• /
• v.27 no.2
• /
• pp.185-197
• /
• 2022

Recently, deep learning networks with high performance for object recognition are emerging. In the case of object recognition using deep learning, it is important to build a training data set to improve performance. To build a data set, we need to collect and label the images. This process requires a lot of time and manpower. For this reason, open data sets are used. However, there are objects that do not have large open data sets. One of them is data required for license plate detection and recognition. Therefore, in this paper, we propose an artificial license plate generator system that can create large data sets by minimizing images. In addition, the detection rate according to the artificial license plate arrangement structure was analyzed. As a result of the analysis, the best layout structure was FVC_III and B, and the most suitable network was D2Det. Although the artificial data set performance was 2-3% lower than that of the actual data set, the time to build the artificial data was about 11 times faster than the time to build the actual data set, proving that it is a time-efficient data set building system.

• Proceedings of the Korean Society of Agricultural Engineers Conference
• /
• 2001.10a
• /
• pp.156-161
• /
• 2001

Currently, agricultural facilities are evaluated using either basic inspections or detailed analysis. However, conventional analyses as well as methods based on fuzzy logic and rule of thumb have not been very successful in providing a clear relationship between rating and real state of agricultural facilities, because they can't provide exactly acceptable reliability of degraded structures with manager or supervisor. Therefore, in this stage, we must define probabilistic variables for representing degradation of structures being given damages during a survival time. This paper describes the application of neural network system in developing the relation between subjective ratings and parameters of agricultural reservoir as well as that between subjective and analytical ratings. It is shown that neural networks can be trained and used successfully in estimating a rating based on several parameters. The specific application problem for agricultural reservoir in the rural area of Korea is presented and database is constructed to maintain training data set, the information of inspection and facilities. This study showed that a successful training of a neural network could be useful, especially if the input data set for target problem contains parameters with a diverse combination of inter-correlation coefficients. And the networks had a prediction rating of about $^{\ast}^{\ast}^{\ast}%$. The neural network system is expected to show high performance fairly in estimate than statistical method to use equation that is consisted of very lowly interrelated variables.

• Journal of The Korean Society of Emergency Medicine
• /
• v.29 no.5
• /
• pp.455-464
• /
• 2018

Objective: Machine learning is not yet widely used in the medical field. Therefore, this study was conducted to compare the performance of preexisting severity prediction models and machine learning based models (random forest [RF], gradient boosting [GB]) for mortality prediction in pneumonia patients. Methods: We retrospectively collected data from patients who visited the emergency department of a tertiary training hospital in Seoul, Korea from January to March of 2015. The Pneumonia Severity Index (PSI) and Sequential Organ Failure Assessment (SOFA) scores were calculated for both groups and the area under the curve (AUC) for mortality prediction was computed. For the RF and GB models, data were divided into a test set and a validation set by the random split method. The training set was learned in RF and GB models and the AUC was obtained from the validation set. The mean AUC was compared with the other two AUCs. Results: Of the 536 investigated patients, 395 were enrolled and 41 of them died. The AUC values of PSI and SOFA scores were 0.799 (0.737-0.862) and 0.865 (0.811-0.918), respectively. The mean AUC values obtained by the RF and GB models were 0.928 (0.899-0.957) and 0.919 (0.886-0.952), respectively. There were significant differences between preexisting severity prediction models and machine learning based models (P<0.001). Conclusion: Classification through machine learning may help predict the mortality of pneumonia patients visiting the emergency department.

• Phonetics and Speech Sciences
• /
• v.11 no.3
• /
• pp.39-47
• /
• 2019

We propose a method to improve the performance of spontaneous speech recognizers by extending their phone set using speech data. In the proposed method, we first extract variable-length phoneme-level segments from broadcast speech signals, and convert them to fixed-length latent vectors using an long short-term memory (LSTM) classifier. We then cluster acoustically similar latent vectors and build a new phone set by choosing the number of clusters with the lowest Davies-Bouldin index. We also update the lexicon of the speech recognizer by choosing the pronunciation sequence of each word with the highest conditional probability. In order to analyze the acoustic characteristics of the new phone set, we visualize its spectral patterns and segment duration. Through speech recognition experiments using a larger training data set than our own previous work, we confirm that the new phone set yields better performance than the conventional phoneme-based and grapheme-based units in both spontaneous speech recognition and read speech recognition.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.45 no.3
• /
• pp.57-65
• /
• 2022

DEA(data envelopment analysis) is a technique for evaluation of relative efficiency of decision making units (DMUs) that have multiple input and output. A DEA model measures the efficiency of a DMU by the relative position of the DMU's input and output in the production possibility set defined by the input and output of the DMUs being compared. In this paper, we proposed several DEA models measuring the multi-period efficiency of a DMU. First, we defined the input and output data that make a production possibility set as the spanning set. We proposed several spanning sets containing input and output of entire periods for measuring the multi-period efficiency of a DMU. We defined the production possibility sets with the proposed spanning sets and gave DEA models under the production possibility sets. Some models measure the efficiency score of each period of a DMU and others measure the integrated efficiency score of the DMU over the entire period. For the test, we applied the models to the sample data set from a long term university student training project. The results show that the suggested models may have the better discrimination power than CCR based results while the ranking of DMUs is not different.

• Journal of the Korean Society of Physical Medicine
• /
• v.13 no.2
• /
• pp.137-145
• /
• 2018

PURPOSE: The purpose of this study was to investigate the effect of action observational physical training with rhythmic auditory stimulation on muscle activity and gait ability in patients with stroke. METHODS: Twenty-six chronic stroke patients participated in this study were assigned into three groups, experimental group 1 (10% faster tempo rhythmic auditory stimulation with action observation training) n=8, experimental group 2 (average tempo rhythmic auditory stimulation with action observation training) n=9, and control group (action observation training) n=9. In this experiment, the corresponding exercise were applied into the subjects of three group for 30 minute a day, 3 time a week during 4 weeks. All participants were measured to muscle activity of lower limb, 10 meter walking test, Figure of 8 walk test, Dynamic gait Index. The collected data were analyzed by using SPSS (version 18.0 for window) and verified that each data was a normal distribution based on Shapiro-Wilk test. Between-group and within-group comparison was analyzed by using One-way ANOVA test, Paired t-test respectively. In all statistical analyses, significance level, ${\alpha}$ was set by .05. RESULTS: The above results revealed that the all experimental group 1 and experimental group 2 and control group were all effective to improve the lower limb muscle activities, gait ability. However more positive effects shown action observational physical training with rhythmic auditory stimulation experimental group. CONCLUSION: This study suggest that action observation physical training with rhythmic auditory stimulation is effective intervention for improvement of muscle activity and walking ability in chronic stroke patients.

• Journal of Broadcast Engineering
• /
• v.26 no.3
• /
• pp.296-307
• /
• 2021

A digital hologram (DH) is an ultra-high value-added video content that includes 3D information in 2D data. Therefore, its intellectual property rights must be protected for its distribution. For this, this paper proposes a watermarking method of DH using a deep neural network. This method is a watermark (WM) invisibility, attack robustness, and blind watermarking method that does not use host information in WM extraction. The proposed network consists of four sub-networks: pre-processing for each of the host and WM, WM embedding watermark, and WM extracting watermark. This network expand the WM data to the host instead of shrinking host data to WM and concatenate it to the host to insert the WM by considering the characteristics of a DH having a strong high frequency component. In addition, in the training of this network, the difference in performance according to the data distribution property of DH is identified, and a method of selecting a training data set with the best performance in all types of DH is presented. The proposed method is tested for various types and strengths of attacks to show its performance. It also shows that this method has high practicality as it operates independently of the resolution of the host DH and WM data.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.5
• /
• pp.881-893
• /
• 2012

This paper proposes to develop system trading strategy using rough set, decision tree in futures market. While there is a great deal of literature about the analysis of data mining, there is relatively little work on developing trading strategies in futures markets. There are three objectives in this paper. The first objective is to analysis performance of decision tree in rule-based system trading. The second objective is to find proper profitable trading interval. The last objective is to find optimized training period of trading rule training. The results of this study show that proposed model is useful trading strategy in foreign exchange market and can be desirable solution which gives lots of investors an important investment information.