• Journal of Biomedical Engineering Research
• /
• v.42 no.3
• /
• pp.80-85
• /
• 2021

We developed a model to classify the absence of cervical cancer using deep learning from the cervical image to which the histogram equalization algorithm was applied, and to compare the performance of each model. A total of 4259 images were used for this study, of which 1852 images were normal and 2407 were abnormal. And this paper applied Image Sharpening(IS), Histogram Equalization(HE), and Contrast Limited Adaptive Histogram Equalization(CLAHE) to the original image. Peak Signal-to-Noise Ratio(PSNR) and Structural Similarity index for Measuring image quality(SSIM) were used to assess the quality of images objectively. As a result of assessment, IS showed 81.75dB of PSNR and 0.96 of SSIM, showing the best image quality. CLAHE and HE showed the PSNR of 62.67dB and 62.60dB respectively, while SSIM of CLAHE was shown as 0.86, which is closer to 1 than HE of 0.75. Using ResNet-50 model with transfer learning, digital image-processed images are classified into normal and abnormal each. In conclusion, the classification accuracy of each model is as follows. 90.77% for IS, which shows the highest, 90.26% for CLAHE and 87.60% for HE. As this study shows, applying proper digital image processing which is for cervical images to Computer Aided Diagnosis(CAD) can help both screening and diagnosing.

• Nuclear Engineering and Technology
• /
• v.54 no.1
• /
• pp.36-48
• /
• 2022

As a valid numerical method to obtain a high-resolution result of a flow field, computational fluid dynamics (CFD) have been widely used to study coolant flow and heat transfer characteristics in fuel rod bundles. However, the time-consuming, iterative calculation of Navier-Stokes equations makes CFD unsuitable for the scenarios that require efficient simulation such as sensitivity analysis and uncertainty quantification. To solve this problem, a reduced-order model (ROM) based on proper orthogonal decomposition (POD) and machine learning (ML) is proposed to simulate the flow field efficiently. Firstly, a validated CFD model to output the flow field data set of the rod bundle is established. Secondly, based on the POD method, the modes and corresponding coefficients of the flow field were extracted. Then, an deep feed-forward neural network, due to its efficiency in approximating arbitrary functions and its ability to handle high-dimensional and strong nonlinear problems, is selected to build a model that maps the non-linear relationship between the mode coefficients and the boundary conditions. A trained surrogate model for modes coefficients prediction is obtained after a certain number of training iterations. Finally, the flow field is reconstructed by combining the product of the POD basis and coefficients. Based on the test dataset, an evaluation of the ROM is carried out. The evaluation results show that the proposed POD-ROM accurately describe the flow status of the fluid field in rod bundles with high resolution in only a few milliseconds.

• Imaging Science in Dentistry
• /
• v.52 no.4
• /
• pp.393-398
• /
• 2022

Purpose: This study proposed a generative adversarial network (GAN) model for T2-weighted image (WI) synthesis from proton density (PD)-WI in a temporomandibular joint(TMJ) magnetic resonance imaging (MRI) protocol. Materials and Methods: From January to November 2019, MRI scans for TMJ were reviewed and 308 imaging sets were collected. For training, 277 pairs of PD- and T2-WI sagittal TMJ images were used. Transfer learning of the pix2pix GAN model was utilized to generate T2-WI from PD-WI. Model performance was evaluated with the structural similarity index map (SSIM) and peak signal-to-noise ratio (PSNR) indices for 31 predicted T2-WI (pT2). The disc position was clinically diagnosed as anterior disc displacement with or without reduction, and joint effusion as present or absent. The true T2-WI-based diagnosis was regarded as the gold standard, to which pT2-based diagnoses were compared using Cohen's ĸ coefficient. Results: The mean SSIM and PSNR values were 0.4781(±0.0522) and 21.30(±1.51) dB, respectively. The pT2 protocol showed almost perfect agreement(ĸ=0.81) with the gold standard for disc position. The number of discordant cases was higher for normal disc position (17%) than for anterior displacement with reduction (2%) or without reduction (10%). The effusion diagnosis also showed almost perfect agreement(ĸ=0.88), with higher concordance for the presence (85%) than for the absence (77%) of effusion. Conclusion: The application of pT2 images for a TMJ MRI protocol useful for diagnosis, although the image quality of pT2 was not fully satisfactory. Further research is expected to enhance pT2 quality.

• Journal of the Korea Convergence Society
• /
• v.13 no.1
• /
• pp.13-21
• /
• 2022

Recent deep learning-based research shows excellent performance in most natural language processing (NLP) fields with pre-trained language models. In particular, the auto-encoder-based language model proves its excellent performance and usefulness in various fields of Korean language understanding. However, the decoder-based Korean generative model even suffers from generating simple sentences. Also, there is few detailed research and data for the field of conversation where generative models are most commonly utilized. Therefore, this paper constructs multi-turn dialogue data for a Korean generative model. In addition, we compare and analyze the performance by improving the dialogue ability of the generative model through transfer learning. In addition, we propose a method of supplementing the insufficient dialogue generation ability of the model by extracting recommended response candidates from external knowledge information through a retrival model.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.2
• /
• pp.343-350
• /
• 2022

It is worth verifying the effectiveness of data integration between data with different features. This study investigated whether the data integration affects the accuracy of deep neural network (DNN), and which integration method shows the best improvement. This study used two different public datasets. One public dataset was taken in an actual farm in India. And another was taken in a laboratory environment in Korea. Leaf images were selected from two different public datasets to have five classes which includes normal and four different types of plant diseases. DNN used pre-trained VGG16 as a feature extractor and multi-layer perceptron as a classifier. Data were integrated into three different ways to be used for the training process. DNN was trained in a supervised manner via the integrated data. The trained DNN was evaluated by using a test dataset taken in an actual farm. DNN shows the best accuracy for the test dataset when DNN was first trained by images taken in the laboratory environment and then trained by images taken in the actual farm. The results show that data integration between plant images taken in a different environment helps improve the performance of deep neural networks. And the results also confirmed that independent use of plant images taken in different environments during the training process is more effective in improving the performance of DNN.

• International Journal of Computer Science & Network Security
• /
• v.22 no.9
• /
• pp.195-207
• /
• 2022

COVID-19 has remained one of the most serious health crises in recent history, resulting in the tragic loss of lives and significant economic impacts on the entire world. The difficulty of controlling COVID-19 poses a threat to the global health sector. Considering that Artificial Intelligence (AI) has contributed to improving research methods and solving problems facing diverse fields of study, AI algorithms have also proven effective in disease detection and early diagnosis. Specifically, acoustic features offer a promising prospect for the early detection of respiratory diseases. Motivated by these observations, this study conceptualized a speech-based diagnostic model to aid in COVID-19 diagnosis. The proposed methodology uses speech signals from confirmed positive and negative cases of COVID-19 to extract features through the pre-trained Visual Geometry Group (VGG-16) model based on Mel spectrogram images. This is used in addition to the K-means algorithm that determines effective features, followed by a Genetic Algorithm-Support Vector Machine (GA-SVM) classifier to classify cases. The experimental findings indicate the proposed methodology's capability to classify COVID-19 and NOT COVID-19 of varying ages and speaking different languages, as demonstrated in the simulations. The proposed methodology depends on deep features, followed by the dimension reduction technique for features to detect COVID-19. As a result, it produces better and more consistent performance than handcrafted features used in previous studies.

• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.81-94
• /
• 2022

Real estate and artworks were considered challenging investment targets for individual investors because of their relatively high average transaction price despite their long investment history. Recently, the so-called fractional investment, generally known as investing in a share of the ownership right for real-life assets, etc., and most investors perceive that they actually own a piece (fraction) of the ownership right through their investments, is gaining popularity. Founded in 2016, Musicow started the first service that allows users to invest in copyright fees related to music distribution. Using the LSTM algorithm, one of the deep learning algorithms, this research predict the price of right to participate in copyright fees traded in Musicow. In addition to variables related to claims such as transfer price, transaction volume of claims, and copyright fees, comprehensive indicators indicating the market conditions for music copyright fees participation, exchange rates reflecting economic conditions, KTB interest rates, and Korea Composite Stock Index were also used as variables. As a result, it was confirmed that the LSTM algorithm accurately predicts the transaction price even in the case of fractional investment which has a relatively low transaction volume.

• Imaging Science in Dentistry
• /
• v.53 no.4
• /
• pp.271-281
• /
• 2023

Purpose: The objective of this scoping review was to investigate the applicability and performance of various convolutional neural network (CNN) models in tooth numbering on panoramic radiographs, achieved through classification, detection, and segmentation tasks. Materials and Methods: An online search was performed of the PubMed, Science Direct, and Scopus databases. Based on the selection process, 12 studies were included in this review. Results: Eleven studies utilized a CNN model for detection tasks, 5 for classification tasks, and 3 for segmentation tasks in the context of tooth numbering on panoramic radiographs. Most of these studies revealed high performance of various CNN models in automating tooth numbering. However, several studies also highlighted limitations of CNNs, such as the presence of false positives and false negatives in identifying decayed teeth, teeth with crown prosthetics, teeth adjacent to edentulous areas, dental implants, root remnants, wisdom teeth, and root canal-treated teeth. These limitations can be overcome by ensuring both the quality and quantity of datasets, as well as optimizing the CNN architecture. Conclusion: CNNs have demonstrated high performance in automated tooth numbering on panoramic radiographs. Future development of CNN-based models for this purpose should also consider different stages of dentition, such as the primary and mixed dentition stages, as well as the presence of various tooth conditions. Ultimately, an optimized CNN architecture can serve as the foundation for an automated tooth numbering system and for further artificial intelligence research on panoramic radiographs for a variety of purposes.

• Journal of Intelligence and Information Systems
• /
• v.29 no.4
• /
• pp.257-269
• /
• 2023

There is currently a burgeoning demand for image synthesis from photos and videos using deep learning models. Existing video synthesis models solely extract motion information from the provided video to generate animation effects on photos. However, these synthesis models encounter challenges in achieving accurate lip synchronization with the audio and maintaining the image quality of the synthesized output. To tackle these issues, this paper introduces a novel framework based on an image animation approach. Within this framework, upon receiving a photo, a video, and audio input, it produces an output that not only retains the unique characteristics of the individuals in the photo but also synchronizes their movements with the provided video, achieving lip synchronization with the audio. Furthermore, a super-resolution model is employed to enhance the quality and resolution of the synthesized output.