• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.479-487
• /
• 2021

In this paper, we propose an Sound Event Detection (SED) model using self-training based on a noisy student model. The proposed SED model consists of two stages. In the first stage, a mean-teacher model based on an Residual Convolutional Recurrent Neural Network (RCRNN) is constructed to provide target labels regarding weakly labeled or unlabeled data. In the second stage, a self-training-based noisy student model is constructed by applying different noise types. That is, feature noises, such as time-frequency shift, mixup, SpecAugment, and dropout-based model noise are used here. In addition, a semi-supervised loss function is applied to train the noisy student model, which acts as label noise injection. The performance of the proposed SED model is evaluated on the validation set of the Detection and Classification of Acoustic Scenes and Events (DCASE) 2020 Challenge Task 4. The experiments show that the single model and ensemble model of the proposed SED based on the noisy student model improve F1-score by 4.6 % and 3.4 % compared to the top-ranked model in DCASE 2020 challenge Task 4, respectively.

• Database Research
• /
• v.34 no.3
• /
• pp.137-147
• /
• 2018

A key to a well-performing human activity recognition (HAR) system through machine learning technique is the availability of a substantial amount of labeled data. Collecting sufficient labeled data is an expensive and time-consuming task. To build a HAR system in a new environment (i.e., the target domain) with very limited labeled data, it is unfavorable to naively exploit the data or trained classifier model from the existing environment (i.e., the source domain) as it is due to the domain difference. While traditional machine learning approaches are unable to address such distribution mismatch, transfer learning approach leverages the utilization of knowledge from existing well-established source domains that help to build an accurate classifier in the target domain. In this work, we propose a transfer learning approach to create an accurate HAR classifier with very limited data through the multitask neural network. The classifier loss function minimization for source and target domain are treated as two different tasks. The knowledge transfer is performed by simultaneously minimizing the loss function of both tasks using a single neural network model. Furthermore, we utilize the unlabeled data in an unsupervised manner to help the model training. The experiment result shows that the proposed work consistently outperforms existing approaches.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.139-147
• /
• 2021

This paper introduces and implements a Sound Event Detection (SED) system based on weakly-supervised learning where only part of the data is labeled, and analyzes the effect of parameters. The SED system estimates the classes and onset/offset times of events in the acoustic signal. In order to train the model, all information on the event class and onset/offset times must be provided. Unfortunately, the onset/offset times are hard to be labeled exactly. Therefore, in the weakly-supervised task, the SED model is trained by "strongly labeled data" including the event class and activations, "weakly labeled data" including the event class, and "unlabeled data" without any label. Recently, the SED systems using the mean-teacher model are widely used for the task with several parameters. These parameters should be chosen carefully because they may affect the performance. In this paper, performance analysis was performed on parameters, such as the feature, moving average parameter, weight of the consistency cost function, ramp-up length, and maximum learning rate, using the data of DCASE 2020 Task 4. Effects and the optimal values of the parameters were discussed.

• Journal of Food Hygiene and Safety
• /
• v.36 no.3
• /
• pp.257-263
• /
• 2021

For this survey, PCR (polymerase chain reaction) testing was conducted using 14 species-specific primers to monitor the labeling of allergy-causing substances in various foods. Sixty samples from stationary stores near elementary schools and imported confectionery shops were tested, including snacks, candies, and chocolate. Allergens of milk, wheat, eggs, tomatoes, almonds and peanuts were detected in 30 cases (50.0%). In addition, many products were detected as either containing unlabeled substances or not showing allergen-related information and labeling in Korean. In order to ensure that consumers are able to purchase products safely and securely, a system for thorough guidance and monitoring of allergen-related labeling by domestic manufacturing and processing companies and import-related companies is required.

• Journal of Radiopharmaceuticals and Molecular Probes
• /
• v.8 no.2
• /
• pp.53-61
• /
• 2022

This study aimed to increase the targeting ability against PSMA in cell therapy using metabolic glycoengineering and biorthogonal chemistry and to visualize cell trafficking using PET imaging. Cellular membranes of THP-1 cells were decorated with azide(-N₃) using Ac₄ManNAz by metabolic glycoengineering. Engineered THP-1 cells were conjugated with DBCO-bearing fluorophore (ADIBO-Cy5.5) for 1 h at different concentrations and analyzed by confocal fluorescence microscopy and flow cytometry. For PSAM ligand conjugation to THP-1 cells, Ac₄ManNAz treated THP-1 cells were incubated with DBCO-PSMA ligand (ADIBO-GUL) at a final concentration with 100 µM for 1 h. To evaluate the effect on cell recognition, PSMA ligand conjugated THP-1 cells(as effectors) were co-cultured with PSMA positive 22RV1 (as target cells) at 3 : 1 a effector-to-target cell (E/T) ratio. The interaction between THP-1 and 22RV1 was monitored by confocal fluorescence microscopy. For preparing the radiolabeled THP-1, the cells were treated at the activity of ~ 740 kBq of [⁸⁹Zr]Zr(oxinate)₄/5 × 10⁶ cells. Radiolabeled cells were analyzed for determination of cell-associated radioactivity by gamma counting and viability using MTS assay. In the cytotoxicity assay, THP-1 cells did not have any cytotoxicity even when the Ac₄ManNAz concentration was 100 µM. In confocal microscopy and flow cytometry, THP-1 cells were efficiently labeled ADIBO-Cy5.5 in a dose-dependent manner, and the dose of 100 µM was the optimal concentration for the following experiments. The clusters of PSMA ligand-conjugated THP-1 cells and 22RV1 cells were identified, indicating cell-cell recognition over the cell surface between two types of cells. Cell radiolabeling efficiency was 54.5 ± 17.8%. THP-1 labeled with 0.09 ± 0.03 Bq/cell showed no significant cytotoxicity compared to unlabeled THP-1 up to 7 days. We successfully demonstrated that Ac₄ManNAz treated cells were efficiently conjugated with ADIBO-GUL for preparing the PSMA-targeting cells, and [⁸⁹Zr]Zr(oxinate)₄ could be used to label cells without toxicity. It suggested that PSMA-ligand conjugated cell therapy could be improved cell targeting and be monitored by PET imaging.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.5
• /
• pp.21-30
• /
• 2023

Accurate segmentation of the kidney tumor is necessary to identify shape, location and safety margin of tumor in abdominal CT images for surgical planning before renal partial nephrectomy. However, kidney tumor segmentation is challenging task due to the various sizes and locations of the tumor for each patient and signal intensity similarity to surrounding organs such as intestine and spleen. In this paper, we propose a semi-supervised learning-based mean teacher network that utilizes both labeled and unlabeled data using a kidney local guided map including kidney local information to segment small-sized kidney tumors occurring at various locations in the kidney, and analyze the performance according to the kidney tumor size. As a result of the study, the proposed method showed an F1-score of 75.24% by considering local information of the kidney using a kidney local guide map to locate the tumor existing around the kidney. In particular, under-segmentation of small-sized tumors which are difficult to segment was improved, and showed a 13.9%p higher F1-score even though it used a smaller amount of labeled data than nnU-Net.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.27 no.5
• /
• pp.113-119
• /
• 2023

Most infrastructure structures were completed during periods of economic growth. The number of infrastructure structures reaching their lifespan is increasing, and the proportion of old structures is gradually increasing. The functions and performance of these structures at the time of design may deteriorate and may even lead to safety accidents. To prevent this repercussion, accurate inspection and appropriate repair are requisite. To this end, demand is increasing for computer vision and deep learning technology to accurately detect even minute cracks. However, deep learning algorithms require a large number of training data. In particular, label images indicating the location of cracks in the image are required. To secure a large number of those label images, a lot of labor and time are consumed. To reduce these costs as well as increase detection accuracy, this study proposed a learning structure based on mean teacher method. This learning structure was trained on a dataset of 900 labeled image dataset and 3000 unlabeled image dataset. The crack detection network model was evaluated on over 300 labeled image dataset, and the detection accuracy recorded a mean intersection over union of 89.23% and an F1 score of 89.12%. Through this experiment, it was confirmed that detection performance was improved compared to supervised learning. It is expected that this proposed method will be used in the future to reduce the cost required to secure label images.

• Journal of Navigation and Port Research
• /
• v.47 no.6
• /
• pp.358-366
• /
• 2023

This study aimed to provide a solution for improving ship collision alert of the 'accident vulnerable ship monitoring service' among the 'intelligent marine traffic information system' services of the Ministry of Oceans and Fisheries. The current ship collision alert uses a supervised learning (SL) model with survey labels based on large ship-oriented data and its operators. Consequently, the small ship data and the operator's opinion are not reflected in the current collision-supervised learning model, and the effect is insufficient because the alarm is provided from a longer distance than the small ship operator feels. In addition, the supervised learning (SL) method requires a large number of labeled data, and the labeling process requires a lot of resources and time. To overcome these limitations, in this paper, the classification model of collision alerts for small ships using unlabeled data with the semi-supervised learning (SSL) algorithms (Label Propagation and TabNet) was studied. Results of real-time experiments on small ship operators using the classification model of collision alerts showed that the satisfaction of operators increased.

• Analytical Science and Technology
• /
• v.7 no.2
• /
• pp.213-224
• /
• 1994

Several isotopomers of cyclooctanone were prepared by selective deuterium substitution. Intrinsic isotope effects on $^{13}C$ NMR chemical shifts of these isotopomers were investigated systematically at low temperature. These istope effects were discussed in relation to the preferred boat-chair conformation of cyclooctanone. Deuterium isotope effects on NMR chemical shifts have been known for a long time. Especially in a conformationally mobile molecule, isotope perturbation could affect NMR signals through a combination of isotope effects on equilibria and intrinsic effects. The distinction between intrinsic and nonintrinsic effects is quite difficult at ambient temperature due to involvement of both equilibrium and intrinsic isotope effects. However if equilibria between possible conformers of cyclooctanone are slowed down enough on the NMR time scale by lowering temperature, it should be possible to measure intrinsic isotope shifts from the separated signals at low temperature. $^{13}C$ NMR has been successfully utilized in the study on molecular conformation in solution when one deals with stable conformers or molecules were rapid interconversion occurs at ambient temperature. The study of dynamic processes in general requires analysis of spectra at several temperature. Anet et al. did $^1H$ NMR study of cyclooctanone at low temperature to freeze out a stable conformation, but were not able initially to deduce which conformation was stable because of the complexity of alkyl region in the $^1H$ NMR spectrum. They also reported the $^1H$ and $^{13}C$ NMR spectra of the $C_9-C_{16}$ cycloalkanones with changing temperature from $-80^{\circ}C$ to $-170^{\circ}C$, but they did not report a variable temperature $^{13}C$ NMR study of cyclooctanone. For the analysis of the intrinsic isotope effect with relation to cylooctanone conformation, $^{13}C$ NMR spectra are obtained in the present work at low temperatures (up to $-150^{\circ}C$) in order to find the chemical shifts at the temperature at which the dynamic process can be "frozen-out" on the NMR time scale and cyclooctanone can be observed as a stable conformation. Both the ring inversion and pseudorotational processes must be "frozen-out" in order to see separate resonances for all eight carbons in cyclooctanone. In contrast to $^1H$ spectra, slowing down just the ring inversion process has no apparent effects on the $^{13}C$ spectra because exchange of environments within the pairs of methylene carbons can still occur by the pseudorotational process. Several isotopomers of cyclooctanone were prepared by selective deuterium substitution (fig. 1) : complete deuterium labeling at C-2 and C-8 positions gave cyclooctanone-2, 2, 8, $8-D_4$ : complete labeling at C-2 and C-7 positions afforded the 2, 2, 7, $7-D_4$ isotopomer : di-deuteration at C-3 gave the 3, $3-D_2$ isotopomer : mono-deuteration provided cyclooctanone-2-D, 4-D and 5-D isotopomers : and partial deuteration on the C-2 and C-8 position, with a chiral and difunctional case catalyst, gave the trans-2, $8-D_2$ isotopomer. These isotopomer were investigated systematically in relation with cyclooctanone conformation and intrinsic isotope effects on $^{13}C$ NMR chemical shifts at low temperature. The determination of the intrinsic effects could help in the analysis of the more complex effects at higher temperature. For quantitative analysis of intrinsic isotope effects, the $^{13}C$ NMR spectrum has been obtained for a mixture of the labeled and unlabeled compounds because the signal separations are very small.

• Journal of Intelligence and Information Systems
• /
• v.24 no.3
• /
• pp.21-44
• /
• 2018

In recent years, the rapid development of internet technology and the popularization of smart devices have resulted in massive amounts of text data. Those text data were produced and distributed through various media platforms such as World Wide Web, Internet news feeds, microblog, and social media. However, this enormous amount of easily obtained information is lack of organization. Therefore, this problem has raised the interest of many researchers in order to manage this huge amount of information. Further, this problem also required professionals that are capable of classifying relevant information and hence text classification is introduced. Text classification is a challenging task in modern data analysis, which it needs to assign a text document into one or more predefined categories or classes. In text classification field, there are different kinds of techniques available such as K-Nearest Neighbor, Naïve Bayes Algorithm, Support Vector Machine, Decision Tree, and Artificial Neural Network. However, while dealing with huge amount of text data, model performance and accuracy becomes a challenge. According to the type of words used in the corpus and type of features created for classification, the performance of a text classification model can be varied. Most of the attempts are been made based on proposing a new algorithm or modifying an existing algorithm. This kind of research can be said already reached their certain limitations for further improvements. In this study, aside from proposing a new algorithm or modifying the algorithm, we focus on searching a way to modify the use of data. It is widely known that classifier performance is influenced by the quality of training data upon which this classifier is built. The real world datasets in most of the time contain noise, or in other words noisy data, these can actually affect the decision made by the classifiers built from these data. In this study, we consider that the data from different domains, which is heterogeneous data might have the characteristics of noise which can be utilized in the classification process. In order to build the classifier, machine learning algorithm is performed based on the assumption that the characteristics of training data and target data are the same or very similar to each other. However, in the case of unstructured data such as text, the features are determined according to the vocabularies included in the document. If the viewpoints of the learning data and target data are different, the features may be appearing different between these two data. In this study, we attempt to improve the classification accuracy by strengthening the robustness of the document classifier through artificially injecting the noise into the process of constructing the document classifier. With data coming from various kind of sources, these data are likely formatted differently. These cause difficulties for traditional machine learning algorithms because they are not developed to recognize different type of data representation at one time and to put them together in same generalization. Therefore, in order to utilize heterogeneous data in the learning process of document classifier, we apply semi-supervised learning in our study. However, unlabeled data might have the possibility to degrade the performance of the document classifier. Therefore, we further proposed a method called Rule Selection-Based Ensemble Semi-Supervised Learning Algorithm (RSESLA) to select only the documents that contributing to the accuracy improvement of the classifier. RSESLA creates multiple views by manipulating the features using different types of classification models and different types of heterogeneous data. The most confident classification rules will be selected and applied for the final decision making. In this paper, three different types of real-world data sources were used, which are news, twitter and blogs.