• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.6
• /
• pp.285-292
• /
• 2023

With the wide spread of smart farms and the advancements in IoT technology, it is easy to obtain additional data in addition to crop images. Consequently, deep learning-based crop disease diagnosis research utilizing multimodal data has become important. This study proposes a crop disease diagnosis method using multimodal supervised contrastive learning by expanding upon the multimodal self-supervised learning. RandAugment method was used to augment crop image and time series of environment data. These augmented data passed through encoder and projection head for each modality, yielding low-dimensional features. Subsequently, the proposed multimodal supervised contrastive loss helped features from the same class get closer while pushing apart those from different classes. Following this, the pretrained model was fine-tuned for crop disease diagnosis. The visualization of t-SNE result and comparative assessments of crop disease diagnosis performance substantiate that the proposed method has superior performance than multimodal self-supervised learning.

• Journal of the Korea Convergence Society
• /
• v.11 no.6
• /
• pp.23-28
• /
• 2020

A comment system is essential for communication on the Internet. However, there are also malicious comments such as inappropriate expression of others by exploiting anonymity online. In order to protect users from malicious comments, classification of malicious / normal comments is necessary, and this can be implemented as text classification. Text classification is one of the important topics in natural language processing, and studies using pre-trained models such as BERT and graph structures such as GCN and GAT have been actively conducted. In this study, we implemented a comment classification system using BERT, GCN, and GAT for actual published comments and compared the performance. In this study, the system using the graph-based model showed higher performance than the BERT.

• Journal of Korea Multimedia Society
• /
• v.24 no.8
• /
• pp.1000-1011
• /
• 2021

In recent years, using Deep Learning methods to apply for medical and biomedical image analysis has seen many advancements. In clinical, using Deep Learning-based approaches for cancer image analysis is one of the key applications for cancer detection and treatment. However, the scarcity and shortage of labeling images make the task of cancer detection and analysis difficult to reach high accuracy. In 2015, the Unet model was introduced and gained much attention from researchers in the field. The success of Unet model is the ability to produce high accuracy with very few input images. Since the development of Unet, there are many variants and modifications of Unet related architecture. This paper proposes a new approach of using Unet++ with pretrained EfficientNet as backbone architecture for breast tumor cell nuclei segmentation and uses the multi-organ transfer learning approach to segment nuclei of breast tumor cells. We attempt to experiment and evaluate the performance of the network on the MonuSeg training dataset and Triple Negative Breast Cancer (TNBC) testing dataset, both are Hematoxylin and Eosin (H & E)-stained images. The results have shown that EfficientUnet++ architecture and the multi-organ transfer learning approach had outperformed other techniques and produced notable accuracy for breast tumor cell nuclei segmentation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.10
• /
• pp.1300-1306
• /
• 2018

The purpose of this study is to see possibility of Char2Vec as alternative of Word2Vec that most famous word embedding model in Sentence Similarity Measure Problem by Deep-Learning. In experiment, we used the Siamese Ma-LSTM recurrent neural network architecture for measure similarity two random sentences. Siamese Ma-LSTM model was implemented with tensorflow. We train each model with 200 epoch on gpu environment and it took about 20 hours. Then we compared Word2Vec based model training result with Char2Vec based model training result. as a result, model of based with Char2Vec that initialized random weight record 75.1% validation dataset accuracy and model of based with Word2Vec that pretrained with 3 million words and phrase record 71.6% validation dataset accuracy. so Char2Vec is suitable alternate of Word2Vec to optimize high system memory requirements problem.

• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.191-204
• /
• 2021

Visual Question Answering (VQA) and image captioning are tasks that require understanding of the features of images and linguistic features of text. Therefore, co-attention may be the key to both tasks, which can connect image and text. In this paper, we propose a model to achieve high performance for VQA by image caption generated using a pretrained standard transformer model based on MSCOCO dataset. Captions unrelated to the question can rather interfere with answering, so some captions similar to the question were selected to use based on a similarity to the question. In addition, stopwords in the caption could not affect or interfere with answering, so the experiment was conducted after removing stopwords. Experiments were conducted on VQA-v2 data to compare the proposed model with the deep modular co-attention network (MCAN) model, which showed good performance by using co-attention between images and text. As a result, the proposed model outperformed the MCAN model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.6
• /
• pp.1516-1529
• /
• 2023

In this paper, a motor gearbox fault-detection system based on a hierarchical flow-based model is proposed. The proposed system is used for the anomaly detection of a motion sound-based actuator module. The proposed flow-based model, which is a generative model, learns by directly modeling a data distribution function. As the objective function is the maximum likelihood value of the input data, the training is stable and simple to use for anomaly detection. The operation sound of a car's side-view mirror motor is converted into a Mel-spectrogram image, consisting of a folding signal and an unfolding signal, and used as training data in this experiment. The proposed system is composed of an encoder and a decoder. The data extracted from the layer of the pretrained feature extractor are used as the decoder input data in the encoder. This information is used in the decoder by performing an interlayer cross-scale convolution operation. The experimental results indicate that the context information of various dimensions extracted from the interlayer hierarchical data improves the defect detection accuracy. This paper is notable because it uses acoustic data and a normalizing flow model to detect outliers based on the features of experimental data.

• Journal of the Korea Convergence Society
• /
• v.12 no.10
• /
• pp.27-35
• /
• 2021

Automatic post editing is a research field that aims to automatically correct errors in machine translation results. This research is mainly being focus on high resource language pairs, such as English-German. Recent APE studies are mainly adopting transfer learning based research, where pre-training language models, or translation models generated through self-supervised learning methodologies are utilized. While translation based APE model shows superior performance in recent researches, as such researches are conducted on the high resource languages, the same perspective cannot be directly applied to the low resource languages. In this work, we apply two transfer learning strategies to Korean-English APE studies and show that transfer learning with translation model can significantly improves APE performance.

• Journal of Korea Multimedia Society
• /
• v.25 no.10
• /
• pp.1375-1385
• /
• 2022

In this paper, we propose a semi-supervised domain adaptation solution to deal with practical face recognition (FR) scenarios where a single face image for each target identity (to be recognized) is only available in the training phase. Main goal of the proposed method is to reduce the discrepancy between the target and the source domain face images, which ultimately improves FR performances. The proposed method is based on the Domain Adatation network (DAN) using an MMD loss function to reduce the discrepancy between domains. In order to train more effectively, we develop a novel loss function learning strategy in which MMD loss and cross-entropy loss functions are adopted by using different weights according to the progress of each epoch during the learning. The proposed weight adoptation focuses on the training of the source domain in the initial learning phase to learn facial feature information such as eyes, nose, and mouth. After the initial learning is completed, the resulting feature information is used to training a deep network using the target domain images. To evaluate the effectiveness of the proposed method, FR performances were evaluated with pretrained model trained only with CASIA-webface (source images) and fine-tuned model trained only with FERET's gallery (target images) under the same FR scenarios. The experimental results showed that the proposed semi-supervised domain adaptation can be improved by 24.78% compared to the pre-trained model and 28.42% compared to the fine-tuned model. In addition, the proposed method outperformed other state-of-the-arts domain adaptation approaches by 9.41%.

• ETRI Journal
• /
• v.44 no.5
• /
• pp.794-804
• /
• 2022

Transformer models have had a great impact on natural language processing (NLP) in recent years by realizing outstanding and efficient contextualized language models. Recent studies have used transformer-based language models for various NLP tasks, including Persian named entity recognition (NER). However, in complex tasks, for example, NER, it is difficult to determine which contextualized embedding will produce the best representation for the tasks. Considering the lack of comparative studies to investigate the use of different contextualized pretrained models with sequence modeling classifiers, we conducted a comparative study about using different classifiers and embedding models. In this paper, we use different transformer-based language models tuned with different classifiers, and we evaluate these models on the Persian NER task. We perform a comparative analysis to assess the impact of text representation and text classification methods on Persian NER performance. We train and evaluate the models on three different Persian NER datasets, that is, MoNa, Peyma, and Arman. Experimental results demonstrate that XLM-R with a linear layer and conditional random field (CRF) layer exhibited the best performance. This model achieved phrase-based F-measures of 70.04, 86.37, and 79.25 and word-based F scores of 78, 84.02, and 89.73 on the MoNa, Peyma, and Arman datasets, respectively. These results represent state-of-the-art performance on the Persian NER task.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.1
• /
• pp.19-24
• /
• 2023

In traditional fish farming way, the workers have to observe all of the pools every time and every day to feed at the right timing. This method causes tremendous stress on workers and wastes time. To solve this problem, we implemented an automatic detection system for feeding time using deep neural network. The detection system consists of two steps: classification of the presence or absence of feed and checking DO (Dissolved Oxygen) of the pool. For the classification, the pretrained ResNet18 model and transfer learning with custom dataset are used. DO is obtained from the DO sensor in the pool through HTTP in real time. For better accuracy, the next step, checking DO proceeds when the result of the classification is absence of feed several times in a row. DO is checked if it is higher than a DO reference value that is set by the workers. These actions are performed automatically in the UI programs developed with LabVIEW.