• Proceedings of the KIEE Conference
• /
• 2002.11c
• /
• pp.521-524
• /
• 2002

This paper deals with the rule generation from data for control system and data mining using rough set. If the cores and reducts are searched for without consideration of the frequency of data belonging to the same equivalent class, the unnecessary attributes may not be discarded, and the resultant rules don't represent well the characteristics of the data. To improve this, we handle the inconsistent data with a probability measure defined by support, As a result the effect of uncertainty in knowledge reduction can be reduced to some extent. Also we construct the rule base in a hierarchical structure by applying core as the classification criteria at each level. If more than one core exist, the coverage degree is used to select an appropriate one among then to increase the classification rate. The proposed method gives more proper and effective rule base in compatibility and size. For some data mining example the simulations are performed to show the effectiveness of the proposed method.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.4 no.2
• /
• pp.205-210
• /
• 2004

This paper deals with rule extraction from data using rough set theory. We construct the rule base in a hierarchical granulation structure by applying core as a classification criteria at each level. When more than one core exist, the coverage is used for the selection of an appropriate one among them to increase the classification rate and accuracy. In Addition, a probabilistic approach is suggested so that the partially useful information included in inconsistent data can be contributed to knowledge reduction in order to decrease the effect of the uncertainty or vagueness of data. As a result, the proposed method yields more proper and efficient rule base in compatability and size. The simulation result shows that it gives a good performance in spite of very simple rules and short conditionals.

• Speech Sciences
• /
• v.8 no.1
• /
• pp.77-91
• /
• 2001

This paper investigates the timing of Korean spoken in a news-reading speech style in order to improve the naturalness of durations used in Korean speech synthesis. Each segment in a corpus of 671 read sentences was annotated with 69 segmental and prosodic features so that the measured duration could be correlated with the context in which it occurred. A CART model based on the features showed a correlation coefficient of 0.79 with an RMSE (root mean squared prediction error) of 23 ms between actual and predicted durations in reserved test data. These results are comparable with recent published results in Korean and similar to results found in other languages. An analysis of the classification tree shows that phrasal structure has the greatest effect on the segment duration, followed by syllable structure and the manner features of surrounding segments. The place features of surrounding segments only have small effects. The model has application in Korean speech synthesis systems.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.6
• /
• pp.10-16
• /
• 1994

In case of off-line character recognition with statistical method, the character recognition speed for Korean or Chinese characters is slow since the amount of calculation is huge. To improve this problem, we seperate the recognition steps into several functional stages and implement them with hardwares for each stage so that all the stages can be processed with pipline structure. In accordance with temporal parallel processing, a high speed character recognition system can be implemented. In this paper, we implement a classification hardware, which is one of the several functional stages, to improve the speed by parallel structure with multiple DSPs(Digital Signal Processors). Also, it is designed to be able to expand DSP boards in parallel to make processing faster as much as we wish. We implement the hardware as an add-on board in IBM-PC, and the result of experiment is that it can process about 47-times and 71-times faster with 2 DSPs and 3 DSPs respectively than the IBM-PC(486D$\times$2-66MHz). The effectiveness is proved by developing a high speed OCR(Optical Character Recognizer).

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.28 no.4
• /
• pp.89-101
• /
• 2020

In order to enhance the safety of the international aviation industry, the International Civil Aviation Organization has recommended establishing an operational foundation for systematic and integrated collection, storage, analysis and sharing of aviation safety data. Accordingly, the Korea aviation industry also needs to comprehensively manage the safety data which generated and collected by various stakeholders related to aviation safety, and through this, it is necessary to previously identify and remove hazards that may cause accident. For more effective data management and utilization, a standard structure should be established to enable integrated management and sharing of safety data. Therefore, this study aims to propose the framework about how to manage and integrate the aviation safety data for big data-based aviation safety management and shared platform.

• Communications for Statistical Applications and Methods
• /
• v.29 no.2
• /
• pp.161-176
• /
• 2022

There has been active research in image classification using deep learning convolutional neural network (CNN) models. ImageNet large-scale visual recognition challenge (ILSVRC) (2010-2017) was one of the most important competitions that boosted the development of efficient deep learning algorithms. This paper introduces and compares six monumental models that achieved high prediction accuracy in ILSVRC. First, we provide a review of the models to illustrate their unique structure and characteristics of the models. We then compare those models under a unified framework. For this reason, additional devices that are not crucial to the structure are excluded. Four popular data sets with different characteristics are then considered to measure the prediction accuracy. By investigating the characteristics of the data sets and the models being compared, we provide some insight into the architectural features of the models.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.1
• /
• pp.1-15
• /
• 2022

The MobileNetV3 is specially designed for mobile devices with limited memory and computing power. To reduce the network parameters and improve the network inference speed, a new lightweight network is proposed based on MobileNetV3. Firstly, to reduce the computation of residual blocks, a partial residual structure is designed by dividing the input feature maps into two parts. The designed partial residual structure is used to replace the residual block in MobileNetV3. Secondly, a dual-path feature extraction structure is designed to further reduce the computation of MobileNetV3. Different convolution kernel sizes are used in the two paths to extract feature maps with different sizes. Besides, a transition layer is also designed for fusing features to reduce the influence of the new structure on accuracy. The CIFAR-100 dataset and Image Net dataset are used to test the performance of the proposed partial residual structure. The ResNet based on the proposed partial residual structure has smaller parameters and FLOPs than the original ResNet. The performance of improved MobileNetV3 is tested on CIFAR-10, CIFAR-100 and ImageNet image classification task dataset. Comparing MobileNetV3, GhostNet and MobileNetV2, the improved MobileNetV3 has smaller parameters and FLOPs. Besides, the improved MobileNetV3 is also tested on CPU and Raspberry Pi. It is faster than other networks

• Journal of Korean Association for Spatial Structures
• /
• v.3 no.3 s.9
• /
• pp.111-120
• /
• 2003

Tension members is a type of effective structural member, which is often used in large span structures. The structure systems composed with tension members are combined in various way and specific formations. So, there are need to research into the formations of tension structure and the type of adaptation in tension structure architectures. The structure systems with tension members were considered as tension main system, vector system and tension supported bending system, comprehensively. And tension structures were classified into the formation of tension structure with uniaxial or multiaxial line tension member, with surface member, with hybrid member of line and surface, concerning the flow of tension force. In each the formation of tension structure, the typical adaptations to architecture were also investigated through architecture examples. The type of the formation can be used to plan an architecture with respect to the flow of tension force and structural feature.

• International Journal of Contents
• /
• v.14 no.1
• /
• pp.7-11
• /
• 2018

Digital technology has been changing everyday life of ordinary people let alone the structure of world industry. The elderly care service is also going through changes influenced by the unavoidable impact from torrents of digital technologies. There are numerous reports and news about the digital technologies increasing the efficiency and effectiveness of care service yet lacking systematic understanding of the sources of such improvement. This study aims to present a new classification framework for digital elderly care service innovation to fully utilize the power of digital technologies drawing on insights from innovation studies and service studies. First, 4 features of digital technologies are identified as sources of new value in service innovation. The co-creation of value by users and producers in service and technology development is discussed to illuminate users' contributions to service innovation. Communication of needs and ideas with producers and application of new technologies into everyday practice of life are identified as the source of new value which can be attributed to the elderly. Customization along with efficiency gains is the key to digital elderly care service innovation. The classification framework, thus, incorporates the needs of the elderly as one axis of criteria in the conventional technology-centered framework. The new classification framework would help give due weight to user-driven or demand-driven innovation in the elderly care service R&D activities.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.6
• /
• pp.2806-2825
• /
• 2018

Unsupervised learning has shown good performance on image, video and audio classification tasks, and much progress has been made so far. It studies how systems can learn to represent particular input patterns in a way that reflects the statistical structure of the overall collection of input patterns. Many promising deep learning systems are commonly trained by the greedy layerwise unsupervised learning manner. The performance of these deep learning architectures benefits from the unsupervised learning ability to disentangling the abstractions and picking out the useful features. However, the existing unsupervised learning algorithms are often difficult to train partly because of the requirement of extensive hyperparameters. The tuning of these hyperparameters is a laborious task that requires expert knowledge, rules of thumb or extensive search. In this paper, we propose a simple and effective unsupervised feature learning algorithm for image classification, which exploits an explicit optimizing way for population and lifetime sparsity. Firstly, a sparse target matrix is built by the competitive rules. Then, the sparse features are optimized by means of minimizing the Euclidean norm ($L_2$) error between the sparse target and the competitive layer outputs. Finally, a classifier is trained using the obtained sparse features. Experimental results show that the proposed method achieves good performance for image classification, and provides discriminative features that generalize well.