• International journal of advanced smart convergence
• /
• v.8 no.1
• /
• pp.141-146
• /
• 2019

These days, there are many applications using neural networks as parts of their system. On the other hand, adversarial examples have become an important issue concerining the security of neural networks. A classifier in neural networks can be fooled and make it miss-classified by adversarial examples. There are many research to encounter adversarial examples by using denoising methods. Some of them using GAN (Generative Adversarial Network) in order to remove adversarial noise from input images. By producing an image from generator network that is close enough to the original clean image, the adversarial examples effects can be reduced. However, there is a chance when adversarial noise can survive the approximation process because it is not like a normal noise. In this chance, we propose a research that utilizes high-level representation in the classifier by combining GAN network with a trained U-Net network. This approach focuses on minimizing the loss function on high representation terms, in order to minimize the difference between the high representation level of the clean data and the approximated output of the noisy data in the training dataset. Furthermore, the generated output is checked whether it shows minimum error compared to true label or not. U-Net network is trained with true label to make sure the generated output gives minimum error in the end. At last, the remaining adversarial noise that still exist after low-level approximation can be removed with the U-Net, because of the minimization on high representation terms.

• Korean Journal of Computational Design and Engineering
• /
• v.18 no.3
• /
• pp.234-242
• /
• 2013

Different CAD systems are used in ship outfitting design on different usage and purpose. Therefore, data exchanges between CAD systems are required from different formats. For data exchange, boundary representation standard formats such as IGES and ISO 10303 (STEP) are widely used. However, they present only B-rep representation. Because of different CAD systems have their own geometry format, data exchanges with design intend are difficult. Especially, Tribon and PDMS use primitives for express their geometry in ship outfitting design. However, Tribon primitives are represented their parameter by values that are non-parametric. Therefore, data size of catalogue library is bigger than different CAD system using parametric primitive representation. And that system has difficulty on data reprocessing. To solve that problem, we discuss about shape DB which contains design parameters of primitive for exchange Tribon primitives. And geometry data exchange between Tribon and Shape Database that defines based on PDMS scheme are specified using primitive mapping that can represent design intend.

• ETRI Journal
• /
• v.11 no.1
• /
• pp.109-122
• /
• 1989

We developed new quadtree and octree representation schemes which reduce the storage requirements from exponential to polynomial. The new schemes not only lessen the large storage requirements of the existing quadtree and octree representation schemes but guarantee an exact representation of the original object. These are made possible by adopting a new set of termination conditions that ensure finiteness of the quadtree and octree during the decomposition. These new data structures are analyzed theoretically and tested empirically. For space complexity, we analyzed its best case, worst case, and average case. Given an $n_e$-gon, we show that the expected number of nodes in our quadtree isO($$$n_e^1.292$) For a polyhedron with $n_f$ faces, the expected number of nodes in the new octree is O($$$n_f^1.667$). For time complexity, we again analyzed the best, worst, and average cases for constructing such quadtree and octree and find the average to be the same as those of the space complexity. Finally, random $n_e$- gons are generated as test data. Regression equations are fitted and are shown to support the claims on the average case performance.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.9
• /
• pp.3211-3230
• /
• 2014

Recently, sparse coding has achieved remarkable success in image representation tasks. In practice, the performance of clustering can be significantly improved if limited label information is incorporated into sparse coding. To this end, in this paper, a novel semi-supervised algorithm, called constrained sparse concept coding (CSCC), is proposed for image representation. CSCC considers limited label information into graph embedding as additional hard constraints, and hence obtains embedding results that are consistent with label information and manifold structure information of the original data. Therefore, CSCC can provide a sparse representation which explicitly utilizes the prior knowledge of the data to improve the discriminative power in clustering. Besides, a kernelized version of our proposed CSCC, namely kernel constrained sparse concept coding (KCSCC), is developed to deal with nonlinear data, which leads to more effective clustering performance. The experimental evaluations on the MNIST, PIE and Yale image sets show the effectiveness of our proposed algorithms.

• ETRI Journal
• /
• v.27 no.6
• /
• pp.818-821
• /
• 2005

Recently, MPEG has opened activity to standardize scene representation for lightweight applications such as in mobile phones. The standard is named lightweight applications scene representation (LASeR) and can be applied to improve and make efficient rich media applications and services on mobile devices. In this standard, we proposed an efficient type codec for point data to maximize the bit efficiency of LASeR. In this paper, we describe the new method and the test results of the proposed scheme.

• Korean Journal of Computational Design and Engineering
• /
• v.11 no.6
• /
• pp.429-439
• /
• 2006

The purpose of this study is to support the integration from current 2D drawing-based design to future 3D model-based design. In this paper, an important theme is the combination between the STEP-based 2D drawing standards (i.e., AP202) and the IFC-based 3D building model standards. To achieve the purpose, two methodologies are proposed as follows: the development of IFC extension model for the 2D shape data representation by harmonizing ISO/STEP AP202; and the development of mapping solution between IFC 2D extension model and KOSDIC by constructing the exchange scenario for 2D shape data representation. It is expected that the proposed IFC2X2 2D extension model and mapping solution will offer the basis of development of the integrated standards model in AEC industry.

• Korean Journal of Computational Design and Engineering
• /
• v.1 no.1
• /
• pp.1-19
• /
• 1996

Non-manifold topological representations can provide a single unified representation for mixed dimensional models or cellular models and thus have a great potential to be applied in many application areas. Various boundary representations for non-manifold topology have been proposed in recent years. These representations are mainly interested in describing the sufficient adjacency relationships and too redundant as a result. A model stored in these representations occupies too much storage space and is hard to be manipulated. In this paper, we proposed a compact hierarchical non-manifold boundary representation that is extended from the half-edge data structure for solid models by introducing the partial topological entities to represent some non-manifold conditions around a vertex, edge or face. This representation allows to reduce the redundancy of the existing schemes while full topological adjacencies are still derived without the loss of efficiency. To verify the statement above, the storage size requirement of the representation is compared with other existing representations and present some main procedures for querying and traversing the representation. We have also implemented a set of the generalized Euler operators that satisfy the Euler-Poincare formula for non-manifold geometric models.

• Journal of Physiology & Pathology in Korean Medicine
• /
• v.16 no.5
• /
• pp.845-850
• /
• 2002

This paper suggest a methodology for representation of findings which can be called as signs and symptoms. A finding consists of unit signs and unit symptoms, and moreover findings which appear in one individual patient have so many different relationship each other. So, it is nat appropriate to list all of possible findings as medical standard or to fill findings as independent things in paper for medical record. We try to distinguish finding item from finding list, and suggest the methodology by which we can make finding list from finding items. That is, we suggest finding item[Concept], value types, relationship, logical operator, and syntax as a component of representation. And by using urinary symptom, we make the example for representation methodology. Finally, we mention the background knowledge, brief research process of related area.

• Journal of Korean Society of Transportation
• /
• v.13 no.3
• /
• pp.35-52
• /
• 1995

This paper reviews both network representation schemes of transportation network and algorithms for findings the shortest path between two points in the network. Two types of network representation schmes for considering U-Turn have been proposed along with some modifications of the vine-based Dijkstra shoretest path algorithm. The traditional Sioux-Fall network has been chosen and modified with the introduction of left-turn prohibitions and U-Turns for the purpose of evaluating the peroformance of two modified algorithms and network representation schemes(NA1 an d NA2 ). This type of modification in both network representation scheme (including network data) and algorithms is not only supposed to be needed for route guidance but supposed to contribute to finding more realistic path and estimating more accurate traffic voulume thorughout the entire network.

• Korean Journal of Computational Design and Engineering
• /
• v.11 no.4
• /
• pp.235-245
• /
• 2006

This paper proposes a CSG-based representation scheme for heterogeneous objects including multi-material objects and Functionally Graded Materials (FGMs). In particular, this scheme focuses on the construction of complicated heterogeneous objects guaranteeing desired material continuities at all the interfaces. In order to create various types of heterogeneous primitives, we first describe methods for specifying material composition functions such as geometry-independent, geometry-dependent functions. Constructive Material Composition (CMC) and corresponding heterogeneous Boolean Operators (e.g. material union, difference, intersection. and partition) are then proposed to illustrate how material continuities are dealt with. Finally, we describe the model hierarchy and data structure for computer representation. Even though the proposed scheme alone is sufficient for modeling all sorts of heterogeneous objects, the proposed scheme adopts a hybrid representation between CSG and decomposition. That is because hybrid representation can avoid the unnecessary growth of binary trees.