• Korean Journal of Computational Design and Engineering
• /
• v.1 no.1
• /
• pp.76-83
• /
• 1996

A three dimensional FE mesh generation scheme based on mapping approach is proposed in this study. A volume in Eucledian space is represented by composite hyperpatches which are piecewise cubic functions in parameters u, v, w. A key idea in the proposed approach is that I sampled grid data points lying only on the boundary surfaces are needed for the shape representation. Inner points which are necessary to form a hyperpatch are internally generated by Coons patches. This approach is most appropriate for the shapes which are compositions of hexahedronlike shapes and also severely curved.

• Journal of the Korean Society for Precision Engineering
• /
• v.13 no.1
• /
• pp.98-107
• /
• 1996

The use of composite hyperpatch model is proposed to predict a machine tool positional error over the entire work space. This is an appropriate representation of the distorted work space. This model is valid for any configuration of 3-axis machine tool. Tool position, which is given NC data or CL data, contains error vector in actual work space. In this study, off-line compensation scheme was investigated for tool position error due to inaccuracy in machine tool structure. The error vector in actual work space is corrected by the error model using Newton-Raphson method. The proposed error compensation method shows the possibility of improving machine accuracy at a low cost.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.27 no.1
• /
• pp.8-14
• /
• 2023

The existing self-supervised learning-based CutPaste generated synthetic data by cutting and attaching specific patches from normal images and then performed anomaly detection. However, this method has a problem in that there is a clear difference in the boundary of the patch. NSA for solving these problems have achieved higher anomaly detection performance by generating natural synthetic data through Poisson Blending. However, NSA has the disadvantage of having many hyperparameters that need to be adjusted for each class. In this paper, synthetic data similar to normal were generated by a simple method of making the size of the synthetic patch very small. At this time, since the patches are so locally synthesized, models that learn local features can easily overfit synthetic data. Therefore, we performed anomaly detection using gMLP, which learns global features, and even with simple synthesis methods, we were able to achieve higher performance than conventional self-supervised learning techniques.

• Journal of the Korean Society for Precision Engineering
• /
• v.18 no.11
• /
• pp.138-147
• /
• 2001

Alignment systems are frequently used under various semiconductor manufacturing environment. Particularly in PDP(Plasma Display Panel) manufacturing process, the alignment system is applied to the combining and sealing processes of the upper and lower glass panels of PDP, where these processes are performed in the vacuum chamber of high vacuum and high temperature. In this paper, the XYΘ-alignment stage is developed to align PDP panels. Because of high vacuum and high temperature environment, the alignment chamber has been designed to isolate the inner part of the alignment chamber from the outer environment of high vacuum and high temperature, in which every part of the alignment stage is inserted. As it is difficult to attach feedback sensors to the alignment stage in the alignment chamber, the alignment stage is implemented with the open loop algorithm, where the parallel link structure has been designed using step-motors and ball-screws for structural simplicity. The kinematic analysis is performed to drive the parallel link structure, based on the experiments of actuation-compensation of the alignment stage. For the error compensation, the hyperpatch model has been used to model the errors. From the experiments, the positional accuracy of the alignment stage can be improved significantly.

• Journal of KIISE
• /
• v.42 no.4
• /
• pp.451-458
• /
• 2015

Previous multimodal learning methods focus on problem-solving aspects, such as image and video search and tagging, rather than on knowledge acquisition via content modeling. In this paper, we propose the Multimodal Concept Hierarchy (MuCH), which is a content modeling method that uses a cartoon video dataset and a character-based subtitle generation method from the learned model. The MuCH model has a multimodal hypernetwork layer, in which the patterns of the words and image patches are represented, and a concept layer, in which each concept variable is represented by a probability distribution of the words and the image patches. The model can learn the characteristics of the characters as concepts from the video subtitles and scene images by using a Bayesian learning method and can also generate character-based subtitles from the learned model if text queries are provided. As an experiment, the MuCH model learned concepts from 'Pororo' cartoon videos with a total of 268 minutes in length and generated character-based subtitles. Finally, we compare the results with those of other multimodal learning models. The Experimental results indicate that given the same text query, our model generates more accurate and more character-specific subtitles than other models.