• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Journal of Internet Technology
• /
• v.22 no.4
• /
• pp.903-911
• /
• 2021

As cyber-attacks on Cyber-Physical System (CPS) become more diverse and sophisticated, it is important to quickly detect malicious behaviors occurring in CPS. Since CPS can collect sensor data in near real time throughout the process, there have been many attempts to detect anomaly behavior through normal behavior learning from the perspective of data-driven security. However, since the CPS datasets are big data and most of the data are normal data, it has always been a great challenge to analyze the data and implement the anomaly detection model. In this paper, we propose and evaluate the Clustered Deep One-Class Classification (CD-OCC) model that combines the clustering algorithm and deep learning (DL) model using only a normal dataset for anomaly detection. We use auto-encoder to reduce the dimensions of the dataset and the K-means clustering algorithm to classify the normal data into the optimal cluster size. The DL model trains to predict clusters of normal data, and we can obtain logit values as outputs. The derived logit values are datasets that can better represent normal data in terms of knowledge distillation and are used as inputs to the OCC model. As a result of the experiment, the F1 score of the proposed model shows 0.93 and 0.83 in the SWaT and HAI dataset, respectively, and shows a significant performance improvement over other recent detectors such as Com-AE and SVM-RBF.

• Structural Engineering and Mechanics
• /
• v.68 no.1
• /
• pp.39-51
• /
• 2018

In this study, a frequency domain model updating method is presented using power spectral density (PSD) data. It uses the sensitivity of PSD function with respect to the unknown structural parameters through a decomposed form of transfer function. The stiffness parameters are captured with high accuracy through solving the sensitivity equations utilizing the least square approach. Using numerically noise polluted data, the model updating results of a truss model prove robustness of the method against measurement and mass modelling errors. Results prove the capabilities of the method for parameter estimation using highly noise polluted data of low ranges of excitation frequency.

• Journal of Ocean Engineering and Technology
• /
• v.22 no.6
• /
• pp.13-19
• /
• 2008

Abnormally large swells that appeared on the coast of the East Sea in October in 2005 and 2006 were simulated using SWAN model to examine the accuracy of the model for future forecasting Seawind data calculated based on the weather chart ant bottom topography were used for input data, and the model was operated more than 20 days before the observed swells to avoid the problems from the cold start of the model. The comparisons with observed wind and wave data were unsatisfactory and neededmore improvement in terms of swell component in the wave model as well as the quality of seawind data. The satellite wind and wave data can be good candidates for future comparison of the wave model results in the East Sea.

• ETRI Journal
• /
• v.46 no.1
• /
• pp.59-70
• /
• 2024

We introduce a high-performance named entity recognition (NER) model for written and spoken language. To overcome challenges related to labeled data scarcity and domain shifts, we use transfer learning to leverage our previously developed KorBERT as the base model. We also adopt a meta-pseudo-label method using a teacher/student framework with labeled and unlabeled data. Our model presents two modifications. First, the student model is updated with an average loss from both human- and pseudo-labeled data. Second, the influence of noisy pseudo-labeled data is mitigated by considering feedback scores and updating the teacher model only when below a threshold (0.0005). We achieve the target NER performance in the spoken language domain and improve that in the written language domain by proposing a straightforward rollback method that reverts to the best model based on scarce human-labeled data. Further improvement is achieved by adjusting the label vector weights in the named entity dictionary.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.168-170
• /
• 2003

In this paper, we discuss the concept of Spatio-Temporal Data Warehouse and analyze the organization model of spatio-temporal data. Based on the above, we found the framework of Spatio-Temporal Data Warehouse composed of data source, processing tools and application, which covers the whole process from building warehouse to supplying services.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.3
• /
• pp.115-122
• /
• 2019

For Korean phoneme recognition, Hidden Markov-Gaussian Mixture model(HMM-GMM) or hybrid models which combine artificial neural network with HMM have been mainly used. However, current approach has limitations in that such models require force-aligned corpus training data that is manually annotated by experts. Recently, researchers used neural network based phoneme recognition model which combines recurrent neural network(RNN)-based structure with connectionist temporal classification(CTC) algorithm to overcome the problem of obtaining manually annotated training data. Yet, in terms of implementation, these RNN-based models have another difficulty in that the amount of data gets larger as the structure gets more sophisticated. This problem of large data size is particularly problematic in the Korean language, which lacks refined corpora. In this study, we introduce CTC algorithm that does not require force-alignment to create a Korean phoneme recognition model. Specifically, the phoneme recognition model is based on convolutional neural network(CNN) which requires relatively small amount of data and can be trained faster when compared to RNN based models. We present the results from two different experiments and a resulting best performing phoneme recognition model which distinguishes 49 Korean phonemes. The best performing phoneme recognition model combines CNN with 3hop Bidirectional LSTM with the final Phoneme Error Rate(PER) at 3.26. The PER is a considerable improvement compared to existing Korean phoneme recognition models that report PER ranging from 10 to 12.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.2
• /
• pp.587-608
• /
• 2022

The European Union recently established the General Data Protection Regulation (GDPR) for secure data use and personal information protection. Inspired by this, South Korea revised their Personal Information Protection Act, the Act on Promotion of Information and Communications Network Utilization and Information Protection, and the Credit Information Use and Protection Act, collectively known as the "Three Data Bills," which prescribe safe personal information use based on pseudonymous data processing. Based on these bills, the personal data store (PDS) has received attention because it utilizes the MyData service, which actively manages and controls personal information based on the approval of individuals, and it practically ensures their rights to informational self-determination. Various types of PDS models have been developed by several countries (e.g., the US, Europe, and Japan) and global platform firms. The South Korean government has now initiated MyData service projects for personal information use in the financial field, focusing on personal credit information management. There is also a need to verify the efficacy of this service in diverse fields (e.g., medical). However, despite the increased attention, existing MyData models and frameworks do not satisfy security requirements of ensured traceability, transparency, and distributed authentication for personal information use. This study analyzes primary PDS models and compares them to an internationally standardized framework for personal information security with guidelines on MyData so that a proper PDS model can be proposed for South Korea.

• The Korean Journal of Applied Statistics
• /
• v.25 no.6
• /
• pp.925-932
• /
• 2012

The article introduces a method to estimate a cointegrated vector autoregressive model, using mixed-frequency data, in terms of a state-space representation of the vector error correction(VECM) of the model. The method directly estimates the parameters of the model, in a state-space form of its VECM representation, using the available data in its mixed-frequency form. Then it allows one to compute in-sample smoothed estimates and out-of-sample forecasts at their high-frequency intervals using the estimated model. The method is applied to a mixed-frequency data set that consists of the quarterly real gross domestic product and three monthly coincident indicators. The result shows that the method produces accurate smoothed and forecasted estimates in comparison to a method based on single-frequency data.

• Communications for Statistical Applications and Methods
• /
• v.8 no.2
• /
• pp.319-325
• /
• 2001

In this note, we propose a simple fuzzy linear regression model for fuzzy input-output data based on Tanaka's approach. Then an LP-based method to derived the satisfying solution of the decision making is developed.