• International journal of advanced smart convergence
• /
• 제7권3호
• /
• pp.15-22
• /
• 2018

Nowadays multiple control methods are used in robot control systems. A model, predictor or error estimator is often used as feedback controller to control a robot. While robots have become more and more intensive with algorithms capable to acquiring independent knowledge from raw data. This paper represents experimental results of real time machine learning control that does not require explicit knowledge about the plant. The controller can be applied on a broad range of tasks with different dynamic characteristics. We tested our controller on the balancing problem of a single propeller pendulum. Experimental results show that the use of a supervised machine learning algorithm in a single propeller pendulum allows the stable swing of a given angle.

• 대한산업공학회지
• /
• 제39권6호
• /
• pp.535-545
• /
• 2013

The goal of stock investment is earning high rate or return with stability. To accomplish this goal, using a portfolio that distributes stocks with high rate of return with less variability and a stock price prediction model with high accuracy is required. In this paper, three methods are suggested to require these conditions. First of all, in portfolio re-balance part, Max-Return and Min-Risk (MRMR) model is suggested to earn the largest rate of return with stability. Secondly, Entering/Leaving Rule (E/L) is suggested to upgrade portfolio when particular stock's rate of return is low. Finally, to use outstanding stock price prediction model, a model based on Semi-Supervised Learning (SSL) which was suggested in last research was applied. The suggested methods were validated and applied on stocks which are listed in KOSPI200 from January 2007 to August 2008.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권7호
• /
• pp.2359-2376
• /
• 2022

With the exponential growth of medical image big data represented by high-resolution CT images(CTI), the high-resolution CTI data is of great importance for clinical research and diagnosis. The paper takes lung CTI as an example to study. Retrieving answer CTIs similar to the input one from the large-scale lung CTI database can effectively assist physicians to diagnose. Compared with the conventional content-based image retrieval(CBIR) methods, the CBIR for lung CTIs demands higher retrieval accuracy in both the contour shape and the internal details of the organ. In traditional supervised deep learning networks, the learning of the network relies on the labeling of CTIs which is a very time-consuming task. To address this issue, the paper proposes a Weakly Supervised Similarity Evaluation Network (WSSENet) for efficiently support similarity analysis of lung CTIs. We conducted extensive experiments to verify the effectiveness of the WSSENet based on which the CBIR is performed.

• International Journal of Computer Science & Network Security
• /
• 제22권9호
• /
• pp.258-270
• /
• 2022

Cancer has become a common disease for the past two decades throughout the globe and there is significant increase of cancer among women. Breast cancer and ovarian cancers are more prevalent among women. Majority of the patients approach the physicians only during their final stage of the disease. Early diagnosis of cancer remains a great challenge for the researchers. Although several drugs are being synthesized very often, their multi-benefits are less investigated. With millions of drugs synthesized and their data are accessible through open repositories. Drug repurposing can be done using machine learning techniques. We propose a feature selection technique in this paper, which is novel that generates multiple populations for the grey wolf algorithm and classifies breast cancer drugs efficiently. Leukemia drug dataset is also investigated and Multilayer perceptron achieved 96% prediction accuracy. Three supervised machine learning algorithms namely Random Forest classifier, Multilayer Perceptron and Support Vector Machine models were applied and Multilayer perceptron had higher accuracy rate of 97.7% for breast cancer drug classification.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2024년도 제69차 동계학술대회논문집 32권1호
• /
• pp.49-50
• /
• 2024

본 논문에서는 웹쉘 악성코드를 정확하게 분류하고, 빠른시간안에 자동으로 웹쉘 분류 및 분석을 통하여 웹쉘을 탐지하기 위하여 인공지능 머신러닝 기반의 Supervised AI ML 및 Classification 알고리즘을 적용하여 빠른 시간안에 분류, 정확한 분석을 통하여 자동화된 탐지시스템인 MWSDS를 제안하고 웹쉘 실험 데이터를 통하여 실증하였다. 본제안의 경우 웹쉘악성코드 공격에 대한 대응뿐만아니라 관리적인 정보보호 체계수립을 통하여 보다 효과적이며, 지속적으로 대응할 수 있을 것으로 전망된다.

• 한국인터넷방송통신학회논문지
• /
• 제18권5호
• /
• pp.171-177
• /
• 2018

최근 많은 이미지 데이터 셋들은 일반적인 특성을 추출하기 위한 다양한 데이터 클래스와 특징을 가지고 있다. 하지만 이러한 다양한 데이터 클래스와 특징으로 인해 해당 데이터 셋으로 훈련된 물체 검출 딥러닝 모델은 데이터 특성이 다른 환경에서 좋은 성능을 내지 못하는 단점을 보인다. 이 논문에서는 하위 카테고리 기반 물체 검출 방법과 오픈셋 물체 검출 방법을 이용하여 이를 극복하고, 강인한 물체 검출 딥러닝 모델을 훈련하기 위해 능동 준지도 학습 (Active Semi-Supervised Learning)을 이용한 다중 분기 트리 구조를 제안한다. 우리는 이 구조를 이용함으로써 데이터 특성이 다른 환경에서 적응할 수 있는 모델을 가질 수 있고, 나아가 이 모델을 이용하여 이전의 모델보다 높은 성능을 확보 할 수 있다.

• 한국컴퓨터정보학회논문지
• /
• 제22권7호
• /
• pp.17-29
• /
• 2017

IRecently, researches on the recognition of indoor user situations through various sensors in a smart home environment are under way. In this paper, the case study was conducted to determine the operation of the robot vacuum cleaner by inferring the user 's indoor situation through the operation of home appliances, because the indoor situation greatly affects the operation of home appliances. In order to collect learning data for indoor situation awareness model learning, we received feedbacks from user when there was a mistake about the cleaning situation. In this paper, we propose a semi-supervised learning method using user feedback data. When we receive a user feedback, we search for the labels of unlabeled data that most fit the feedbacks collected through genetic algorithm, and use this data to learn the model. In order to verify the performance of the proposed algorithm, we performed a comparison experiments with other learning algorithms in the same environment and confirmed that the performance of the proposed algorithm is better than the other algorithms.

• Journal of information and communication convergence engineering
• /
• 제22권2호
• /
• pp.165-171
• /
• 2024

In this study, we present a novel approach for enhancing chest X-ray image classification (normal, Covid-19, edema, mass nodules, and pneumothorax) by combining contrastive learning and machine learning algorithms. A vast amount of unlabeled data was leveraged to learn representations so that data efficiency is improved as a means of addressing the limited availability of labeled data in X-ray images. Our approach involves training classification algorithms using the extracted features from a linear fine-tuned Momentum Contrast (MoCo) model. The MoCo architecture with a Resnet34, Resnet50, or Resnet101 backbone is trained to learn features from unlabeled data. Instead of only fine-tuning the linear classifier layer on the MoCopretrained model, we propose training nonlinear classifiers as substitutes for softmax in deep networks. The empirical results show that while the linear fine-tuned ImageNet-pretrained models achieved the highest accuracy of only 82.9% and the linear fine-tuned MoCo-pretrained models an increased highest accuracy of 84.8%, our proposed method offered a significant improvement and achieved the highest accuracy of 87.9%.

• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.365-381
• /
• 2023

The existing vision-based techniques for inspection and condition assessment of civil infrastructure are mostly manual and consequently time-consuming, expensive, subjective, and risky. As a viable alternative, researchers in the past resorted to deep learning-based autonomous damage detection algorithms for expedited post-disaster reconnaissance of structures. Although a number of automatic damage detection algorithms have been proposed, the scarcity of labeled training data remains a major concern. To address this issue, this study proposed a semi-supervised learning (SSL) framework based on consistency regularization and cross-supervision. Image data from post-earthquake reconnaissance, that contains cracks, spalling, and exposed rebars are used to evaluate the proposed solution. Experiments are carried out under different data partition protocols, and it is shown that the proposed SSL method can make use of unlabeled images to enhance the segmentation performance when limited amount of ground truth labels are provided. This study also proposes DeepLab-AASPP and modified versions of U-Net++ based on channel-wise attention mechanism to better segment the components and damage areas from images of reinforced concrete buildings. The channel-wise attention mechanism can effectively improve the performance of the network by dynamically scaling the feature maps so that the networks can focus on more informative feature maps in the concatenation layer. The proposed DeepLab-AASPP achieves the best performance on component segmentation and damage state segmentation tasks with mIoU scores of 0.9850 and 0.7032, respectively. For crack, spalling, and rebar segmentation tasks, modified U-Net++ obtains the best performance with Igou scores (excluding the background pixels) of 0.5449, 0.9375, and 0.5018, respectively. The proposed architectures win the second place in IC-SHM2021 competition in all five tasks of Project 2.

• Composites Research
• /
• 제36권6호
• /
• pp.395-401
• /
• 2023

본 연구에서는 지도학습 기반 분할기법을 이용하여 단층 촬영된 단방향 복합재료의 유한요소모델링을 실시하였다. 우선, 단방향 복합재료의 형상 정보를 얻기 위해 Micro-CT 스캔을 수행하여 단방향 복합재료의 순수 체적(raw volume)을 획득하였고 여기에 몇 개의 단면을 선택하여 재료의 마이크로 구조인 섬유의 형상을 라벨링하였다. 이후 재료의 단면 이미지와 라벨링한 이미지를 각각 입출력으로 U-net 모델을 훈련시켰다. 이를 사용하여 선택되지 않은 단층촬영 이미지를 섬유형상을 구분하는 분할을 수행하였고 이렇게 생성된 3차원 정보를 이용해서 유한요소모델을 생성하였다. 최종적으로 단방향 복합재료 시편과 유한요소모델의 섬유체적비를 비교하여 제안된 방법의 적절성을 확인하였다.