• Journal of Information Processing Systems
• /
• v.20 no.4
• /
• pp.501-513
• /
• 2024

In recent years, graph neural networks (GNNs) have been extensively used to analyze graph data across various domains because of their powerful capabilities in learning complex graph-structured data. However, recent research has focused on improving the performance of a single GNN with only two or three layers. This is because stacking layers deeply causes the over-smoothing problem of GNNs, which degrades the performance of GNNs significantly. On the other hand, ensemble methods combine individual weak models to obtain better generalization performance. Among them, gradient boosting is a powerful supervised learning algorithm that adds new weak models in the direction of reducing the errors of the previously created weak models. After repeating this process, gradient boosting combines the weak models to produce a strong model with better performance. Until now, most studies on GNNs have focused on improving the performance of a single GNN. In contrast, improving the performance of GNNs using multiple GNNs has not been studied much yet. In this paper, we propose gradient boosted graph neural networks (GBGNN) that combine multiple shallow GNNs with gradient boosting. We use shallow GNNs as weak models and create new weak models using the proposed gradient boosting-based loss function. Our empirical evaluations on three real-world datasets demonstrate that GBGNN performs much better than a single GNN. Specifically, in our experiments using graph convolutional network (GCN) and graph attention network (GAT) as weak models on the Cora dataset, GBGNN achieves performance improvements of 12.3%p and 6.1%p in node classification accuracy compared to a single GCN and a single GAT, respectively.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.5
• /
• pp.892-898
• /
• 2012

Rough classification methods are used to improving the recognition speed in many character recognition problems. In this case, some irreversible result can occur by an error in rough classification. Methods for duplicating each model in several classes are used in order to reduce this risk. But the errors by rough classfication can not be completely ruled out by these methods. In this paper, an recognition method is proposed to increase speed that matches models selectively without any increase in error. This method constructs a model graph using similarity between models. Then a search process begins from a particular point in the model graph. In this process, matching of unnecessary models are reduced that are not similar to the input pattern. In this paper, the proposed method is applied to the recognition problem of handwriting numbers and upper/lower cases of English alphabets. In the experiments, the proposed method was compared with the basic method that matches all models with input pattern. As a result, the same recognition rate, which has shown as the basic method, was obtained by controlling the out-degree of the model graph and the number of maintaining candidates during the search process thereby being increased the recognition speed to 2.45 times.

• The Journal of Korea Robotics Society
• /
• v.9 no.3
• /
• pp.178-184
• /
• 2014

Tendon driven robot mechanisms have many advantages such as allowing miniaturization and light-weight designs and/or enhancing flexibility in the design of structures. When designing or analyzing tendon driven mechanisms, it is important to determine how the tendons should be connected and whether the designed mechanism is easily controllable. Graph representation is useful to view and analyze such tendon driven mechanisms that are complicatedly interconnected between mechanical elements. In this paper, we propose a method of generalized graph representation that provides us with an intuitive analysis tool not only for tendon driven manipulators, but also various other kinds of mechanical systems which are combined with tendons. This method leads us to easily obtain structure matrix - which is the one of the most important steps in analyzing tendon driven mechanisms.

• Korean Journal of Computational Design and Engineering
• /
• v.3 no.4
• /
• pp.211-222
• /
• 1998

Parametric design is an important modeling paradigm in CAD/CAM applications, enabling efficient design modifications and variations. One of the major issues in parametric design is to develop a geometric constraint solver that can handle a large set of geometric configurations efficiently and robustly. In this appear, we propose a new approach to geometric constraint solving that employs a graph-based method to solve the ruler-and-compass constructible configurations and a numerical method to solve the ruler-and-compass non-constructible configurations, in a way that combines the advantages of both methods. The geometric constraint solving process consists of two phases: 1) planning phase and 2) execution phase. In the planning phase, a sequence of construction steps is generated by clustering the constrained geometric entities and reducing the constraint graph in sequence. in the execution phase, each construction step is evaluated to determine the geometric entities, using both approaches. By combining the advantages of the graph-based constructive approach with the universality of the numerical approach, the proposed approach can maximize the efficiency, robustness, and extensibility of geometric constraint solver.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.161-177
• /
• 2019

In this paper, we study the performance improvement of the answer extraction in Question-Answering system by using sentence dependency parsing result. The Question-Answering (QA) system consists of query analysis, which is a method of analyzing the user's query, and answer extraction, which is a method to extract appropriate answers in the document. And various studies have been conducted on two methods. In order to improve the performance of answer extraction, it is necessary to accurately reflect the grammatical information of sentences. In Korean, because word order structure is free and omission of sentence components is frequent, dependency parsing is a good way to analyze Korean syntax. Therefore, in this study, we improved the performance of the answer extraction by adding the features generated by dependency parsing analysis to the inputs of the answer extraction model (Bidirectional LSTM-CRF). The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. In this study, we compared the performance of the answer extraction model when inputting basic word features generated without the dependency parsing and the performance of the model when inputting the addition of the Eojeol tag feature and dependency graph embedding feature. Since dependency parsing is performed on a basic unit of an Eojeol, which is a component of sentences separated by a space, the tag information of the Eojeol can be obtained as a result of the dependency parsing. The Eojeol tag feature means the tag information of the Eojeol. The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. From the dependency parsing result, a graph is generated from the Eojeol to the node, the dependency between the Eojeol to the edge, and the Eojeol tag to the node label. In this process, an undirected graph is generated or a directed graph is generated according to whether or not the dependency relation direction is considered. To obtain the embedding of the graph, we used Graph2Vec, which is a method of finding the embedding of the graph by the subgraphs constituting a graph. We can specify the maximum path length between nodes in the process of finding subgraphs of a graph. If the maximum path length between nodes is 1, graph embedding is generated only by direct dependency between Eojeol, and graph embedding is generated including indirect dependencies as the maximum path length between nodes becomes larger. In the experiment, the maximum path length between nodes is adjusted differently from 1 to 3 depending on whether direction of dependency is considered or not, and the performance of answer extraction is measured. Experimental results show that both Eojeol tag feature and dependency graph embedding feature improve the performance of answer extraction. In particular, considering the direction of the dependency relation and extracting the dependency graph generated with the maximum path length of 1 in the subgraph extraction process in Graph2Vec as the input of the model, the highest answer extraction performance was shown. As a result of these experiments, we concluded that it is better to take into account the direction of dependence and to consider only the direct connection rather than the indirect dependence between the words. The significance of this study is as follows. First, we improved the performance of answer extraction by adding features using dependency parsing results, taking into account the characteristics of Korean, which is free of word order structure and omission of sentence components. Second, we generated feature of dependency parsing result by learning - based graph embedding method without defining the pattern of dependency between Eojeol. Future research directions are as follows. In this study, the features generated as a result of the dependency parsing are applied only to the answer extraction model in order to grasp the meaning. However, in the future, if the performance is confirmed by applying the features to various natural language processing models such as sentiment analysis or name entity recognition, the validity of the features can be verified more accurately.

• Journal of Korea Society of Industrial Information Systems
• /
• v.10 no.2
• /
• pp.26-31
• /
• 2005

In this paper, we present the document summarizers which are simpler and more condense than the existing ones generally used in the web search engines. This method is a statistic-based summarization method using the concept of the complete graph. We suppose that each sentence as a vertex and the similarity between two sentences as a link of the graph. We compare this summarizer with those of Clustering and MMR techniques which are well-known as the good summarization methods. For the comparison, we use FScore using the summarization results generated by human subjects. Our experimental results verify the accuracy of this method, being about $30\%$ better than the others.

• Journal of Korean Institute of Industrial Engineers
• /
• v.23 no.4
• /
• pp.709-717
• /
• 1997

For an undirected graph G=(V, E), the Rural Postman Problem (RPP) is a problem that finds a minimum cost tour that must pass edges in E'($\subseteq$ E) at least once. RPP, such as Traveling Salesman Problem (TSP), is known as an NP. Complete problem. In the previous study of RPP, he structure of the chromosome is constructed by E' and the direction of the edge. Hence, the larger the size of IE' I is, the larger the size of the chromosome and the size of the solution space are. In this paper, we transform the RPP into a Hamiltonian graph and use a genetic algorithm to solve the transformed problem using restructured chromosomes. In the simulations, we analyze our method and the previous study. From the simulation results, it is found that the results of the proposed method is better than those of the previous method and the proposed method also obtains the near optimal solution in earlier generations than the previous study.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2001.07a
• /
• pp.1069-1072
• /
• 2001

We described a method resonant mode identification in dielectric-rod loaded cylindrical cavity resonators. Resonant frequency of dielectric loaded cavity is calculated by analyzing the characteristic equation. The characteristic equation is solved by using the ContourPlot graph of Mathematica. As the result of comparing calculation value and experimental value of resonant frequencies, we know that the field representation of non-decaying mode is exact. The contour graph method is not a method using approximated representation of electromagnetic field variation at the outer area of dielectric in the resonators but a method using exact representation.

• Proceedings of the KIEE Conference
• /
• 2006.10c
• /
• pp.184-186
• /
• 2006

We propose a new method for shortest path planning of mobile robots. The topological information of the graph is obtained by thinning method to generate the collision-free path of robot. And the travelling path is determined through hierarchical planning stages. The first stage generates an initial path by use of Dijkstra's algorithm. The second stage then generates the final path by use of dynamic programming (DP). The DP adjusts the intial path to reduce the total travelling distance of robot. Simulation results are presented to verify the performance of the proposed method.

• Journal of the Korean housing association
• /
• v.18 no.2
• /
• pp.121-128
• /
• 2007

The investigation of the apartment housing as a dwelling type has become the most important and popular research subject to understand the housing culture in Korea. In their methods of typological analysis, it is found that most studies represent each unit plan as a simplified architectural drawing. This type of typology, however, has difficulties in processing a large scale of data set because each representation of a plan contain too many informations. To deal with this problem, this study suggests a new graph-theoretical method by which apartment plans can be represented in a more simple and effective way. This new method is also tested against the sample plans from Gangnam-gu area in Seoul to reveal the design logic hidden in plan configuration. Through a series of analyses, it is verified that there exist a design strategy that guides the particular pattern of zoning and allocation of each room in the plan.