• Imaging Science in Dentistry
• /
• v.52 no.4
• /
• pp.351-357
• /
• 2022

Purpose: Convolutional neural networks (CNNs) have rapidly emerged as one of the most promising artificial intelligence methods in the field of medical and dental research. CNNs can provide an effective diagnostic methodology allowing for the detection of early-staged diseases. Therefore, this study aimed to evaluate the performance of a deep CNN algorithm for apical lesion segmentation from panoramic radiographs. Materials and Methods: A total of 1000 panoramic images showing apical lesions were separated into training (n=800, 80%), validation (n=100, 10%), and test (n=100, 10%) datasets. The performance of identifying apical lesions was evaluated by calculating the precision, recall, and F1-score. Results: In the test group of 180 apical lesions, 147 lesions were segmented from panoramic radiographs with an intersection over union (IoU) threshold of 0.3. The F1-score values, as a measure of performance, were 0.828, 0.815, and 0.742, respectively, with IoU thresholds of 0.3, 0.4, and 0.5. Conclusion: This study showed the potential utility of a deep learning-guided approach for the segmentation of apical lesions. The deep CNN algorithm using U-Net demonstrated considerably high performance in detecting apical lesions.

• Journal of Biomedical Engineering Research
• /
• v.44 no.2
• /
• pp.139-146
• /
• 2023

Osteoporosis is a disease in which the risk of bone fractures increases due to a decrease in bone density caused by aging. Osteoporosis is diagnosed by measuring bone density in the total hip, femoral neck, and lumbar spine. To accurately measure bone density in the lumbar spine, the vertebral region must be segmented from the lumbar X-ray image. Deep learning-based automatic spinal segmentation methods can provide fast and precise information about the vertebral region. In this study, we used 695 lumbar spine images as training and test datasets for a deep learning segmentation model. We proposed a lumbar automatic segmentation model, CM-Net, which combines the center point of the spine and the modified U-Net network. As a result, the average Dice Similarity Coefficient(DSC) was 0.974, precision was 0.916, recall was 0.906, accuracy was 0.998, and Area under the Precision-Recall Curve (AUPRC) was 0.912. This study demonstrates a high-performance automatic segmentation model for lumbar X-ray images, which overcomes noise such as spinal fractures and implants. Furthermore, we can perform accurate measurement of bone density on lumbar X-ray images using an automatic segmentation methodology for the spine, which can prevent the risk of compression fractures at an early stage and improve the accuracy and efficiency of osteoporosis diagnosis.

• Journal of Internet of Things and Convergence
• /
• v.9 no.1
• /
• pp.9-17
• /
• 2023

Deep learning with large amount of computations is difficult to implement on micro-sized IoT devices or moblie devices. Recently, lightweight deep learning technologies have been introduced to make sure that deep learning can be implemented even on small devices by reducing the amount of computation of the model. Quantization is one of lightweight techniques that can be efficiently used to reduce the memory and size of the model by expressing parameter values with continuous distribution as discrete values of fixed bits. However, the accuracy of the model is reduced due to discrete value representation in quantization. In this paper, we introduce various quantization techniques to correct the accuracy. We selected APoT and EWGS from existing quantization techniques, and comparatively analyzed the results through experimentations The selected techniques were trained and tested with CIFAR-10 or CIFAR-100 datasets in the ResNet model. We found out problems with them through experimental results analysis and presented directions for future research.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.85-92
• /
• 2023

Deep learning applications are increasingly being leveraged for disease detection tasks in medical imaging modalities such as X-ray, Computed Tomography (CT), and Magnetic Resonance Imaging (MRI). Most data-centric deep learning challenges necessitate the use of supervised learning methodologies to attain high accuracy and to facilitate performance evaluation through comparison with the ground truth. Supervised learning mandates a substantial amount of image and label sets, however, procuring an adequate volume of medical imaging data for training is a formidable task. Various data augmentation strategies can mitigate the underfitting issue inherent in supervised learning-based models that are trained on limited medical image and label sets. This research investigates the enhancement of a deep learning-based rib fracture segmentation model and the efficacy of data augmentation techniques such as left-right flipping, rotation, and scaling. Augmented dataset with L/R flipping and rotations(30°, 60°) increased model performance, however, dataset with rotation(90°) and ⨯0.5 rescaling decreased model performance. This indicates the usage of appropriate data augmentation methods depending on datasets and tasks.

• Smart Media Journal
• /
• v.11 no.2
• /
• pp.31-38
• /
• 2022

Promptly predicting changes in the salinity in rivers is an important task to predict the damage to agriculture and ecosystems caused by salinity infiltration and to establish disaster prevention measures. Because machine learning(ML) methods show much less computation cost than physics-based hydraulic models, they can predict the river salinity in a relatively short time. Due to shorter training time, ML methods have been studied as a complementary technique to physics-based hydraulic model. Many studies on salinity prediction based on machine learning have been studied actively around the world, but there are few studies in South Korea. With a massive number of datasets available publicly, we evaluated the performance of various kinds of machine learning techniques that predict the salinity of the Nakdong River Estuary Basin. As a result, LightGBM algorithm shows average 0.37 in RMSE as prediction performance and 2-20 times faster learning speed than other algorithms. This indicates that machine learning techniques can be applied to predict the salinity of rivers in Korea.

• Journal of the Korean Society for information Management
• /
• v.39 no.1
• /
• pp.69-90
• /
• 2022

As basic data that can systematically support and evaluate R&D activities as well as set current and future research directions by grasping specific trends in domestic academic research, I sought efficient ways to assign standardized subject categories (control keywords) to individual journal papers. To this end, I conducted various experiments on major factors affecting the performance of automatic classification, focusing on feature selection techniques, for the purpose of automatically allocating the classification categories on the National Research Foundation of Korea's Academic Research Classification Scheme to domestic journal papers. As a result, the automatic classification of domestic journal papers, which are imbalanced datasets of the real environment, showed that a fairly good level of performance can be expected using more simple classifiers, feature selection techniques, and relatively small training sets.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.66-71
• /
• 2021

본 연구는 한국어 자질 기반 감성분석(Feature-based Sentiment Analysis: FbSA)을 위한 대규모의 학습데이터 구축에 있어 반자동 언어데이터 증강 기법(SSP: Semi-automatic Symbolic Propagation)에 입각한 자질-감성 주석 데이터셋 FeSAD(Feature-Sentiment-Annotated Dataset)의 개발 과정과 성능 평가를 소개하는 것을 목표로 한다. FeSAD는 언어자원을 활용한 SSP 1단계 주석 이후, 작업자의 주석이 2단계에서 이루어지는 2-STEP 주석 과정을 통해 구축된다. SSP 주석을 위한 언어자원에는 부분 문법 그래프(Local Grammar Graph: LGG) 스키마와 한국어 기계가독형 전자사전 DECO(Dictionnaire Electronique du COréen)가 활용되며, 본 연구에서는 7개의 도메인(코스메틱, IT제품, 패션/의류, 푸드/배달음식, 가구/인테리어, 핀테크앱, KPOP)에 대해, 오피니언 트리플이 주석된 FeSAD 데이터셋을 구축하는 프로세싱을 소개하였다. 코스메틱(COS)과 푸드/배달음식(FOO) 두 도메인에 대해, 언어자원을 활용한 1단계 SSP 주석 성능을 평가한 결과, 각각 F1-score 0.93과 0.90의 성능을 보였으며, 이를 통해 FbSA용 학습데이터 주석을 위한 작업자의 작업이 기존 작업의 10% 이하의 비중으로 감소함으로써, 학습데이터 구축을 위한 프로세싱의 소요시간과 품질이 획기적으로 개선될 수 있음을 확인하였다.

• Computers and Concrete
• /
• v.32 no.2
• /
• pp.217-232
• /
• 2023

Numerous studies have been performed on the behavior of pile foundations in cold regions. This study first attempted to employ artificial neural networks (ANN) to predict pile-bearing capacity focusing on pile data recorded primarily on cold regions. As the ANN technique has disadvantages such as finding global minima or slower convergence rates, this study in the second phase deals with the development of an ANN-based predictive model improved with an Elephant herding optimizer (EHO), Dragonfly Algorithm (DA), Genetic Algorithm (GA), and Evolution Strategy (ES) methods for predicting the piles' bearing capacity. The network inputs included the pile geometrical features, pile area (m²), pile length (m), internal friction angle along the pile body and pile tip (Ø°), and effective vertical stress. The MLP model pile's output was the ultimate bearing capacity. A sensitivity analysis was performed to determine the optimum parameters to select the best predictive model. A trial-and-error technique was also used to find the optimum network architecture and the number of hidden nodes. According to the results, there is a good consistency between the pile-bearing DA-MLP-predicted capacities and the measured bearing capacities. Based on the R2 and determination coefficient as 0.90364 and 0.8643 for testing and training datasets, respectively, it is suggested that the DA-MLP model can be effectively implemented with higher reliability, efficiency, and practicability to predict the bearing capacity of piles.

• Journal of Advanced Navigation Technology
• /
• v.25 no.6
• /
• pp.576-581
• /
• 2021

Compared to a low concentration, a high concentration clearly entails limitations in terms of predictive performance owing to differences in its frequency and environment of occurrence. To resolve this problem, in this study, an artificial intelligence neural network algorithm was used to classify low and high concentrations; furthermore, two prediction models trained using the characteristics of the classified concentration types were used for prediction. To this end, we constructed training datasets using weather and air pollutant data collected over a decade in the Cheonan region. We designed a DNN-based classification model to classify low and high concentrations; further, we designed low- and high-concentration prediction models to reflect characteristics by concentration type based on the low and high concentrations classified through the classification model. According to the results of the performance assessment of the prediction model by concentration type, the low- and high-concentration prediction accuracies were 90.38% and 96.37%, respectively.

• Journal of Internet of Things and Convergence
• /
• v.9 no.6
• /
• pp.119-124
• /
• 2023

Deep learning has recently demonstrated conspicuous efficacy across diverse domains than traditional machine learning techniques, as the most popular approach for pattern recognition. The classification problems for tabular data, however, are remain for the area of traditional machine learning. This paper introduces a novel network module designed to tabular data into high-dimensional tensors. The module is integrated into conventional deep learning networks and subsequently applied to the classification of structured data. The proposed method undergoes training and validation on four datasets, culminating in an average accuracy of 90.22%. Notably, this performance surpasses that of the contemporary deep learning model, TabNet, by 2.55%p. The proposed approach acquires significance by virtue of its capacity to harness diverse network architectures, renowned for their superior performance in the domain of computer vision, for the analysis of tabular data.