• Journal of the Korea Society of Computer and Information
• /
• v.26 no.12
• /
• pp.1-9
• /
• 2021

Disaster alerts are text messages sent by government to people in the area in the event of a disaster. Since the number of disaster alerts has increased, the number of people who block disaster alerts is increasing as many unnecessary disaster alerts are being received. To solve this problem, this study proposes a deep learning model that automatically classifies disaster alerts by disaster type, and allows only necessary disaster alerts to be received according to the recipient. The proposed model embeds disaster alerts via KoBERT and classifies them by disaster type with LSTM. As a result of classifying disaster alerts using 3 combinations of parts of speech: [Noun], [Noun + Adjective + Verb] and [All parts], and 4 classification models: Proposed model, Keyword classification, Word2Vec + 1D-CNN and KoBERT + FFNN, the proposed model achieved the highest performance with 0.988954 accuracy.

• Journal of Korea Multimedia Society
• /
• v.22 no.9
• /
• pp.1000-1010
• /
• 2019

Image semantic segmentation and dehazing are key tasks in the computer vision. In recent years, researches in both tasks have achieved substantial improvements in performance with the development of Convolutional Neural Network (CNN). However, most of the previous works for semantic segmentation assume the images are captured in clear weather and show degraded performance under hazy images with low contrast and faded color. Meanwhile, dehazing aims to recover clear image given observed hazy image, which is an ill-posed problem and can be alleviated with additional information about the image. In this work, we propose a deep multi-task network for simultaneous semantic segmentation and dehazing. The proposed network takes single haze image as input and predicts dense semantic segmentation map and clear image. The visual information getting refined during the dehazing process can help the recognition task of semantic segmentation. On the other hand, semantic features obtained during the semantic segmentation process can provide cues for color priors for objects, which can help dehazing process. Experimental results demonstrate the effectiveness of the proposed multi-task approach, showing improved performance compared to the separate networks.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.49-54
• /
• 2024

Data imbalance is a common issue encountered in classification problems, stemming from a significant disparity in the number of samples between classes within the dataset. Such data imbalance typically leads to problems in classification models, including overfitting, underfitting, and misinterpretation of performance metrics. Methods to address this issue include resampling, augmentation, regularization techniques, and adjustment of loss functions. In this paper, we focus on loss function adjustment, particularly comparing the performance of various configurations of loss functions (Cross Entropy, Balanced Cross Entropy, two settings of Focal Loss: 𝛼 = 1 and 𝛼 = Balanced, Asymmetric Loss) on Multi-Class black-box video data with imbalance issues. The comparison is conducted using the I3D, and R3D_18 models.

• 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 the Korea Institute of Information and Communication Engineering
• /
• v.24 no.7
• /
• pp.935-941
• /
• 2020

People estimation is important to provide IoT services. Most people counting technologies use camera or sensor data. However, the conventional technologies have the disadvantages of invasion of privacy and the need to install extra infrastructure. This paper proposes a method for estimating the number of people using a Wi-Fi AP. We use channel state information of Wi-Fi and analyze that using deep learning technology. It can be achieved by pre-installed Wi-Fi infrastructure that reduce cost for people estimation and privacy infringement. The proposed algorithm uses a k-binding data for pre-processing process and a 1D-CNN learning model. Two APs were installed to analyze the estimation results of six people. The result of the accurate number estimation was 64.8%, but the result of classifying the number of people into classes showed a high result of 84.5%. This algorithm is expected to be applicable to estimate the density of people in a small space.

• International Journal of Computer Science & Network Security
• /
• v.22 no.9
• /
• pp.175-182
• /
• 2022

In our daily life, we come across different types of information, for example in the format of multimedia and text. We all need different types of information for our common routines as watching/reading the news, listening to the radio, and watching different types of videos. However, sometimes we could run into problems when a certain type of information is required. For example, someone is listening to the radio and wants to listen to jazz, and unfortunately, all the radio channels play pop music mixed with advertisements. The listener gets stuck with pop music and gives up searching for jazz. So, the above example can be solved with an automatic audio classification system. Deep Learning (DL) models could make human life easy by using audio classifications, but it is expensive and difficult to deploy such models at edge devices like nano BLE sense raspberry pi, because these models require huge computational power like graphics processing unit (G.P.U), to solve the problem, we proposed DL model. In our proposed work, we had gone for a low complexity model for Audio Event Detection (AED), we extracted Mel-spectrograms of dimension 128×431×1 from audio signals and applied normalization. A total of 3 data augmentation methods were applied as follows: frequency masking, time masking, and mixup. In addition, we designed Convolutional Neural Network (CNN) with spatial dropout, batch normalization, and separable 2D inspired by VGGnet [1]. In addition, we reduced the model size by using model quantization of float16 to the trained model. Experiments were conducted on the updated dataset provided by the Detection and Classification of Acoustic Events and Scenes (DCASE) 2020 challenge. We confirm that our model achieved a val_loss of 0.33 and an accuracy of 90.34% within the 132.50KB model size.

• The Korean Journal of Applied Statistics
• /
• v.36 no.2
• /
• pp.129-139
• /
• 2023

Forecasting volatility is essential to avoiding the risk caused by the uncertainties of an financial asset. Complicated financial volatility features such as ambiguity between non-stationarity and stationarity, asymmetry, long-memory, sudden fairly large values like outliers bring great challenges to volatility forecasts. In order to address such complicated features implicity, we consider machine leaning models such as LSTM (1997) and GRU (2014), which are known to be suitable for existing time series forecasting. However, there are the problems of vanishing gradients, of enormous amount of computation, and of a huge memory. To solve these problems, a causal temporal convolutional network (TCN) model, an advanced form of 1D CNN, is also applied. It is confirmed that the overall forecasting power of TCN model is higher than that of the RNN models in forecasting VIX, VXD, and VXN, the daily volatility indices of S&P 500, DJIA, Nasdaq, respectively.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.1
• /
• pp.347-353
• /
• 2023

Currently, soldiers enlisted in the military (Army) are receiving measurements (automatic, manual) of body parts and trying on sample clothing at boot training centers, and then receiving clothing in the desired size. Due to the low accuracy of the measured size during the measurement process, in the military, which uses a relatively more detailed sizing system than civilian casual clothes, the supplied clothes do not fit properly, so the frequency of changing the clothes is very frequent. In addition, there is a problem in that inventory is managed inefficiently by applying the measurement system based on the old generation body shape data collected more than a decade ago without reflecting the western-changed body type change of the MZ generation. That is, military uniforms of the necessary size are insufficient, and many unnecessary-sized military uniforms are in stock. Therefore, in order to reduce the frequency of clothing replacement and improve the efficiency of stock management, deep learning-based automatic measurement of body size, big data analysis, and machine learning-based "Personalized Combat Uniform Automatic Recommendation System for Enlisted Soldiers" is proposed.

• Korean Journal of Agricultural Science
• /
• v.50 no.4
• /
• pp.675-696
• /
• 2023

Terpenoids, also referred to as terpenes, are a large family of naturally occurring chemical compounds present in the essential oils extracted from medicinal plants. In this study, a nondestructive methodology was created by combining ATR-FT-IR (attenuated total reflectance-Fourier transform infrared), and Raman spectroscopy for the terpenoids assessment in medicinal plants essential oils from ten different geographical locations. Partial least squares regression (PLSR) and support vector regression (SVR) were used as machine learning methodologies. However, a deep learning based model called as one-dimensional convolutional neural network (1D CNN) were also developed for models comparison. With a correlation coefficient (R²) of 0.999 and a lowest RMSEP (root mean squared error of prediction) of 0.006% for the prediction datasets, the SVR model created for FT-IR spectral data outperformed both the PLSR and 1 D CNN models. On the other hand, for the classification of essential oils derived from plants collected from various geographical regions, the created SVM (support vector machine) classification model for Raman spectroscopic data obtained an overall classification accuracy of 0.997% which was superior than the FT-IR (0.986%) data. Based on the results we propose that FT-IR spectroscopy, when coupled with the SVR model, has a significant potential for the non-destructive identification of terpenoids in essential oils compared with destructive chemical analysis methods.

• Journal of Platform Technology
• /
• v.6 no.1
• /
• pp.3-8
• /
• 2018

This paper proposes a technique for recognizing online handwritten cursive data obtained by tracing a motion trajectory while a user is in the 3D space based on a convolution neural network (CNN) algorithm. There is a difficulty in recognizing the virtual character input by the user in the 3D space because it includes both the character stroke and the movement stroke. In this paper, we divide syllable into consonant and vowel units by using labeling technique in addition to the result of localizing letter stroke and movement stroke in the previous study. The coordinate information of the separated consonants and vowels are converted into image data, and Korean handwriting recognition was performed using a convolutional neural network. After learning the neural network using 1,680 syllables written by five hand writers, the accuracy is calculated by using the new hand writers who did not participate in the writing of training data. The accuracy of phoneme-based recognition is 98.9% based on convolutional neural network. The proposed method has the advantage of drastically reducing learning data compared to syllable-based learning.