• Convergence Security Journal
• /
• v.21 no.3
• /
• pp.49-56
• /
• 2021

Recently, artificial intelligence is regarded as an essential technology in our society. In particular, the invasion of privacy in artificial intelligence has become a serious problem in modern society. Split learning, proposed at MIT in 2019 for privacy protection, is a type of federated learning technique that does not share any raw data. In this study, we studied a safe and accurate segmentation learning model using known differential privacy to safely manage data. In addition, we trained SVHN and GTSRB on a split learning model to which 15 different types of differential privacy are applied, and checked whether the learning is stable. By conducting a learning data extraction attack, a differential privacy budget that prevents attacks is quantitatively derived through MSE.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.9
• /
• pp.1049-1059
• /
• 2019

We propose a learning-based image segmentation algorithm. Starting from super-pixels, our method learns the probability of merging two regions based on the ground truth made by humans. The learned information is used in determining whether the two regions should be merged or not in a segmentation stage. Unlike exiting learning-based algorithms, we use both local and object information. The local information represents features computed from super-pixels and the object information represent high level information available only in the learning process. The object information is considered as privileged information, and we can use a framework that utilize the privileged information such as SVM+. In experiments on the Berkeley Segmentation Dataset and Benchmark (BSDS 500) and PASCAL Visual Object Classes Challenge (VOC 2012) data set, out model exhibited the best performance with a relatively small training data set and also showed competitive results with a sufficiently large training data set.

• Journal of the Korean School Mathematics Society
• /
• v.14 no.4
• /
• pp.459-476
• /
• 2011

In this study, as a teaching way of classes by achievement level, low level students have taken five-minute tests that could be efficient in remedial feedback, and we have investigated the students' improvements of their assessment in mathematics through each of the unit tests. The results show that the five-minute tests with low level classes helped them develop mathematics problem-solving skills and also form positive attitudes about mathematics. More studies of the various methods must be done so that low level students can develop their abilities to solve mathematical problems skillfully.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.211-212
• /
• 2018

머신 러닝 환경에서 많은 양의 데이터를 한꺼번에 학습하게 되면 데이터 트래픽이 증가함에 따라 흐름 정체가 발생하고 학습 품질이 저하되며 학습속도 지연 등의 문제가 발생한다. 본 연구는 머신러닝 환경에서 빅 데이터 학습 데이터 분할을 위한 핵심 목표인 Greedy 알고리즘에 대해 설명하고 간단한 Greedy 알고리즘을 사용하여 각각의 데이터 파티션을 생성하여 학습 속도의 효율성을 향상시키는 방법을 제안한다.

• Tunnel and Underground Space
• /
• v.34 no.3
• /
• pp.218-230
• /
• 2024

Recently, many studies have been conducted for safety management in construction sites by incorporating computer vision. Anchor box parameters are used in state-of-the-art deep learning-based object detection and segmentation, and the optimized parameters are critical in the training process to ensure consistent accuracy. Those parameters are generally tuned by fixing the shape and size by the user's heuristic method, and a single parameter controls the training rate in the model. However, the anchor box parameters are sensitive depending on the type of object and the size of the object, and as the number of training data increases. There is a limit to reflecting all the characteristics of the training data with a single parameter. Therefore, this paper suggests a method of applying multiple parameters optimized through data split to solve the above-mentioned problem. Criteria for efficiently segmenting integrated training data according to object size, number of objects, and shape of objects were established, and the effectiveness of the proposed data split method was verified through a comparative study of conventional scheme and proposed methods.

• Journal of Internet Computing and Services
• /
• v.20 no.3
• /
• pp.33-41
• /
• 2019

In this paper, we propose a method to learn the redesigned class with background and object for semantic segmentation of indoor scene image. Semantic image segmentation is a technique that divides meaningful parts of an image, such as walls and beds, into pixels. Previous work of semantic image segmentation has proposed methods of learning various object classes of images through neural networks, and it has been pointed out that there is insufficient accuracy compared to long learning time. However, in the problem of separating objects and backgrounds, there is no need to learn various object classes. So we concentrate on separating objects and backgrounds, and propose method to learn after class simplification. The accuracy of the proposed learning method is about 5 ~ 12% higher than the existing methods. In addition, the learning time is reduced by about 14 ~ 60 minutes when the class is configured differently In the same environment, and it shows that it is possible to efficiently learn about the problem of separating the object and the background.

• Annual Conference on Human and Language Technology
• /
• 2002.10e
• /
• pp.33-39
• /
• 2002

중국어의 품사 태깅(part-of-speech tagging)을 위해서는 중국어 문장들은 내부 단어간의 명확한 분리가 없기 때문에 단어 분할(word segmentation)과 품사 태깅을 동시에 처리해야 한다. 본 논문은 규칙 기반(rule base)과 사전 기반(dictionary base) 기법을 혼합하여 구현한 단어 분할 시스템을 사용하여 입력 문장을 단어 단위로 분할하고, HMM(hidden Markov model) 기반 통계적 품사 태깅 기법을 사용한다. 특히, 본 논문에서는 주어진 말뭉치(corpus)로부터 자동 학습(automatic training)을 통해 품사 사전을 구축하여 구현된 시스템과 말뭉치간의 독립성을 유지한다. 말뭉치는 중국어 간체와 번체 모두를 대상으로 하고, 각 말뭉치로부터 자동 학습을 통해 얻어진 품사 사전으로 단어 분할과 품사 태깅을 한다. 실험결과들은 간체, 번체 각각의 단어 분할 성능과 품사 태깅 성능을 보여준다.

• The Journal of the Korea Contents Association
• /
• v.13 no.12
• /
• pp.1048-1057
• /
• 2013

This study aims to evaluate the effect on the academic achievement and satisfaction of the learner's prior knowledge level and segmenting time of video lectures in an learning environment using smartphones. Depending on the level of prior knowledge, learners were divided into two groups of the upper 35% and the lower 35%. Each group was offered video lectures by a 5-min, 10-min, 15-min, and 20-min length. As a result, a high level of prior knowledge only had a positive effect on the academic achievement. With respect to the segmentation time of video lectures, 10-min, 15-min lectures were effective to the academic achievement and 15-min, 20-min lectures influenced positively the learners' satisfaction. Moreover, the interaction between the level of learners' prior knowledge and the segmentation time of video lectures only had an impact on their academic achievement. The results of the simple effect analysis conducted to examine the effect of interaction carefully show that 15-min, 20-min video lectures are more effective for the upper 35% in prior knowledge and 10-min ones are better for its' lower 35%. In a nutshell, these findings suggest that the high-prior knowledge groups could be provided with a longer video lectures, and furthermore, 5-min video lectures are not adequate in a mobile learning environment with smartphones.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.1
• /
• pp.187-194
• /
• 1996

In this paper, we propose regression neural networks based on regression trees. We map regression trees into three layered feedforward networks. We put multi feature split functions in the first layer so that the networks have a better chance to get optimal partitions of input space. We suggest two supervised learning algorithms for the network training and test both in single feature split and multifeature split functions. In experiments, the proposed regression neural networks is proved to have the better learning performance than those of the single feature split regression trees and the single feature split regression networks. Furthermore, we shows that the proposed learning schemes have an effect to prune an over-grown tree without degrading the learning performance.

• Annual Conference on Human and Language Technology
• /
• 1995.10a
• /
• pp.30-34
• /
• 1995

신경망을 이용하는 연속 음성 인식에서 학습이라 함은 인위적으로 분할된 음성 데이타를 토대로 진행되는 것이 지배적이었다. 그러나 분할된 음성데이타를 마련하기 위해서는 많은 시간과 노력, 숙련 등을 요구할 뿐만아니라 그 자체가 인식도메인의 변화나 확장을 어렵게 하는 하나의 요인 되기도 한다. 그래서 분할된 음성데이타의 사용을 가급적 피하고 그러면서도 성능을 떨어뜨리지 않는 신경망 학습법들이 나타나고 있다. 본 논문에서는 학습된 인식기를 이용하여 자동으로 한국어 음성데이타를 분할한 후 그 분할된 데이타를 이용하여 다시 인식기를 재학습시켜나가는 반복 과정을 소개하고자 한다. 여기에는 TDNN이 인식기로 사용되며 인식단위는 음소이다. 학습은 cross-validation 기법을 이용하여 제어된다.