• Korean Journal of Remote Sensing
• /
• v.39 no.5_1
• /
• pp.481-493
• /
• 2023

In high-density urban areas, the urban heat island effect increases urban temperatures, leading to negative impacts such as worsened air pollution, increased cooling energy consumption, and increased greenhouse gas emissions. In urban environments where it is difficult to secure additional green spaces, rooftop greening is an efficient greenhouse gas reduction strategy. In this study, we not only analyzed the current status of the urban heat island effect but also utilized high-resolution satellite data and spatial information to estimate the available rooftop greening area within the study area. We evaluated the mitigation effect of the urban heat island phenomenon and carbon sequestration capacity through temperature predictions resulting from rooftop greening. To achieve this, we utilized WorldView-2 satellite data to classify land cover in the urban heat island areas of Busan city. We developed a prediction model for temperature changes before and after rooftop greening using machine learning techniques. To assess the degree of urban heat island mitigation due to changes in rooftop greening areas, we constructed a temperature change prediction model with temperature as the dependent variable using the random forest technique. In this process, we built a multiple regression model to derive high-resolution land surface temperatures for training data using Google Earth Engine, combining Landsat-8 and Sentinel-2 satellite data. Additionally, we evaluated carbon sequestration based on rooftop greening areas using a carbon absorption capacity per plant. The results of this study suggest that the developed satellite-based urban heat island assessment and temperature change prediction technology using Random Forest models can be applied to urban heat island-vulnerable areas with potential for expansion.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_3
• /
• pp.1779-1790
• /
• 2023

Satellite-based fog detection algorithms are being developed to detect fog in real-time over a wide area, with a focus on the Korean Peninsula (KorPen). The GEO-KOMPSAT-2A/Advanced Meteorological Imager (GK2A/AMI, GK2A) satellite offers an excellent temporal resolution (10 min) and a spatial resolution (500 m), while GEO-KOMPSAT-2B/Geostationary Ocean Color Imager-II (GK2B/GOCI-II, GK2B) provides an excellent spatial resolution (250 m) but poor temporal resolution (1 h) with only visible channels. To enhance the fog detection level (10 min, 250 m), we developed a fused GK2AB fog detection algorithm (FDA) of GK2A and GK2B. The GK2AB FDA comprises three main steps. First, the Korea Meteorological Satellite Center's GK2A daytime fog detection algorithm is utilized to detect fog, considering various optical and physical characteristics. In the second step, GK2B data is extrapolated to 10-min intervals by matching GK2A pixels based on the closest time and location when GK2B observes the KorPen. For reflectance, GK2B normalized visible (NVIS) is corrected using GK2A NVIS of the same time, considering the difference in wavelength range and observation geometry. GK2B NVIS is extrapolated at 10-min intervals using the 10-min changes in GK2A NVIS. In the final step, the extrapolated GK2B NVIS, solar zenith angle, and outputs of GK2A FDA are utilized as input data for machine learning (decision tree) to develop the GK2AB FDA, which detects fog at a resolution of 250 m and a 10-min interval based on geographical locations. Six and four cases were used for the training and validation of GK2AB FDA, respectively. Quantitative verification of GK2AB FDA utilized ground observation data on visibility, wind speed, and relative humidity. Compared to GK2A FDA, GK2AB FDA exhibited a fourfold increase in spatial resolution, resulting in more detailed discrimination between fog and non-fog pixels. In general, irrespective of the validation method, the probability of detection (POD) and the Hanssen-Kuiper Skill score (KSS) are high or similar, indicating that it better detects previously undetected fog pixels. However, GK2AB FDA, compared to GK2A FDA, tends to over-detect fog with a higher false alarm ratio and bias.

• Korean Chemical Engineering Research
• /
• v.61 no.4
• /
• pp.523-541
• /
• 2023

As accessibility to 3D printers increases, there is a growing frequency of exposure to chemicals associated with 3D printing. However, research on the toxicity and harmfulness of chemicals generated by 3D printing is insufficient, and the performance of toxicity prediction using in silico techniques is limited due to missing molecular structure data. In this study, quantitative structure-activity relationship (QSAR) model based on data-centric AI approach was developed to predict the toxicity of new 3D printing materials by imputing missing values in molecular descriptors. First, MissForest algorithm was utilized to impute missing values in molecular descriptors of hazardous 3D printing materials. Then, based on four different machine learning models (decision tree, random forest, XGBoost, SVM), a machine learning (ML)-based QSAR model was developed to predict the bioconcentration factor (Log BCF), octanol-air partition coefficient (Log Koa), and partition coefficient (Log P). Furthermore, the reliability of the data-centric QSAR model was validated through the Tree-SHAP (SHapley Additive exPlanations) method, which is one of explainable artificial intelligence (XAI) techniques. The proposed imputation method based on the MissForest enlarged approximately 2.5 times more molecular structure data compared to the existing data. Based on the imputed dataset of molecular descriptor, the developed data-centric QSAR model achieved approximately 73%, 76% and 92% of prediction performance for Log BCF, Log Koa, and Log P, respectively. Lastly, Tree-SHAP analysis demonstrated that the data-centric-based QSAR model achieved high prediction performance for toxicity information by identifying key molecular descriptors highly correlated with toxicity indices. Therefore, the proposed QSAR model based on the data-centric XAI approach can be extended to predict the toxicity of potential pollutants in emerging printing chemicals, chemical process, semiconductor or display process.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.707-723
• /
• 2022

Although satellite-based sea surface temperature (SST) is advantageous for monitoring large areas, spatiotemporal data gaps frequently occur due to various environmental or mechanical causes. Thus, it is crucial to fill in the gaps to maximize its usability. In this study, daily SST composite fields with a resolution of 4 km were produced through a two-step machine learning approach using polar-orbiting and geostationary satellite SST data. The first step was SST reconstruction based on Data Interpolate Convolutional AutoEncoder (DINCAE) using multi-satellite-derived SST data. The second step improved the reconstructed SST targeting in situ measurements based on light gradient boosting machine (LGBM) to finally produce daily SST composite fields. The DINCAE model was validated using random masks for 50 days, whereas the LGBM model was evaluated using leave-one-year-out cross-validation (LOYOCV). The SST reconstruction accuracy was high, resulting in R² of 0.98, and a root-mean-square-error (RMSE) of 0.97℃. The accuracy increase by the second step was also high when compared to in situ measurements, resulting in an RMSE decrease of 0.21-0.29℃ and an MAE decrease of 0.17-0.24℃. The SST composite fields generated using all in situ data in this study were comparable with the existing data assimilated SST composite fields. In addition, the LGBM model in the second step greatly reduced the overfitting, which was reported as a limitation in the previous study that used random forest. The spatial distribution of the corrected SST was similar to those of existing high resolution SST composite fields, revealing that spatial details of oceanic phenomena such as fronts, eddies and SST gradients were well simulated. This research demonstrated the potential to produce high resolution seamless SST composite fields using multi-satellite data and artificial intelligence.

• 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.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.40 no.3
• /
• pp.273-283
• /
• 2020

Because of climate change, the occurrence of localized and heavy rainfall is increasing. It is important to predict floods in urban areas that have suffered inundation in the past. For flood prediction, not only numerical analysis models but also machine learning-based models can be applied. The LSTM (Long Short-Term Memory) neural network used in this study is appropriate for sequence data, but it demands a lot of data. However, rainfall that causes flooding does not appear every year in a single urban basin, meaning it is difficult to collect enough data for deep learning. Therefore, in addition to the rainfall observed in the study area, the observed rainfall in another urban basin was applied in the predictive model. The LSTM neural network was used for predicting the total overflow, and the result of the SWMM (Storm Water Management Model) was applied as target data. The prediction of the inundation map was performed by using logistic regression; the independent variable was the total overflow and the dependent variable was the presence or absence of flooding in each grid. The dependent variable of logistic regression was collected through the simulation results of a two-dimensional flood model. The input data of the two-dimensional flood model were the overflow at each manhole calculated by the SWMM. According to the LSTM neural network parameters, the prediction results of total overflow were compared. Four predictive models were used in this study depending on the parameter of the LSTM. The average RMSE (Root Mean Square Error) for verification and testing was 1.4279 ㎥/s, 1.0079 ㎥/s for the four LSTM models. The minimum RMSE of the verification and testing was calculated as 1.1655 ㎥/s and 0.8797 ㎥/s. It was confirmed that the total overflow can be predicted similarly to the SWMM simulation results. The prediction of inundation extent was performed by linking the logistic regression with the results of the LSTM neural network, and the maximum area fitness was 97.33 % when more than 0.5 m depth was considered. The methodology presented in this study would be helpful in improving urban flood response based on deep learning methodology.

• Journal of Internet Computing and Services
• /
• v.21 no.5
• /
• pp.139-148
• /
• 2020

Infectious diseases have long plagued mankind, and predicting and preventing them has been a big challenge for mankind. For this reasen, various studies have been conducted so far to predict infectious diseases. Most of the early studies relied on epidemiological data from the Centers for Disease Control and Prevention (CDC), and the problem was that the data provided by the CDC was updated only once a week, making it difficult to predict the number of real-time disease outbreaks. However, with the emergence of various Internet media due to the recent development of IT technology, studies have been conducted to predict the occurrence of infectious diseases through web data, and most of the studies we have researched have been using single Web data to predict diseases. However, disease forecasting through a single Web data has the disadvantage of having difficulty collecting large amounts of learning data and making accurate predictions through models for recent outbreaks such as "COVID-19". Thus, we would like to demonstrate through experiments that models that use multiple Web data to predict the occurrence of infectious diseases through LSTM models are more accurate than those that use single Web data and suggest models suitable for predicting infectious diseases. In this experiment, we predicted the occurrence of "Malaria" and "Epidemic-parotitis" using a single web data model and the model we propose. A total of 104 weeks of NEWS, SNS, and search query data were collected, of which 75 weeks were used as learning data and 29 weeks were used as verification data. In the experiment we predicted verification data using our proposed model and single web data, Pearson correlation coefficient for the predicted results of our proposed model showed the highest similarity at 0.94, 0.86, and RMSE was also the lowest at 0.19, 0.07.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.28 no.7
• /
• pp.1216-1221
• /
• 2022

Ships tend to be larger to increase the efficiency of cargo transportation. Larger ships lead to increased travel time for ship workers, increased work intensity, and reduced work efficiency. Problems such as increased work intensity are reducing the influx of young people into labor, along with the phenomenon of avoidance of high intensity labor by the younger generation. In addition, the rapid aging of the population and decrease in the young labor force aggravate the labor shortage problem in the maritime industry. To overcome this, the maritime industry has recently introduced technologies such as an intelligent production design platform and a smart production operation management system, and a smart autonomous logistics system in one of these technologies. The smart autonomous logistics system is a technology that delivers various goods using intelligent mobile robots, and enables the robot to drive itself by using sensors such as lidar and camera. Therefore, in this paper, it was checked whether the mobile robot could autonomously drive to the stop sign by detecting the passage way of the ship deck. The autonomous driving was performed by detecting the passage way of the ship deck through the camera mounted on the mobile robot based on the data learned through Nvidia's End-to-end learning. The mobile robot was stopped by checking the stop sign using SSD MobileNetV2. The experiment was repeated five times in which the mobile robot autonomously drives to the stop sign without deviation from the ship deck passage way at a distance of about 70m. As a result of the experiment, it was confirmed that the mobile robot was driven without deviation from passage way. If the smart autonomous logistics system to which this result is applied is used in the marine industry, it is thought that the stability, reduction of labor force, and work efficiency will be improved when workers work.

• The Journal of the Acoustical Society of Korea
• /
• v.43 no.1
• /
• pp.9-18
• /
• 2024

The importance of active sonar systems is emerging due to the quietness of underwater targets and the increase in ambient noise due to the increase in maritime traffic. However, the low signal-to-noise ratio of the echo signal due to multipath propagation of the signal, various clutter, ambient noise and reverberation makes it difficult to identify underwater targets using active sonar. Attempts have been made to apply data-based methods such as machine learning or deep learning to improve the performance of underwater target recognition systems, but it is difficult to collect enough data for training due to the nature of sonar datasets. Methods based on mathematical modeling have been mainly used to compensate for insufficient active sonar data. However, methodologies based on mathematical modeling have limitations in accurately simulating complex underwater phenomena. Therefore, in this paper, we propose a sonar signal synthesis method based on a deep neural network. In order to apply the neural network model to the field of sonar signal synthesis, the proposed method appropriately corrects the attention-based encoder and decoder to the sonar signal, which is the main module of the Tacotron model mainly used in the field of speech synthesis. It is possible to synthesize a signal more similar to the actual signal by training the proposed model using the dataset collected by arranging a simulated target in an actual marine environment. In order to verify the performance of the proposed method, Perceptual evaluation of audio quality test was conducted and within score difference -2.3 was shown compared to actual signal in a total of four different environments. These results prove that the active sonar signal generated by the proposed method approximates the actual signal.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.57-73
• /
• 2021

Maintenance and prevention of failure through anomaly detection of ICT infrastructure is becoming important. System monitoring data is multidimensional time series data. When we deal with multidimensional time series data, we have difficulty in considering both characteristics of multidimensional data and characteristics of time series data. When dealing with multidimensional data, correlation between variables should be considered. Existing methods such as probability and linear base, distance base, etc. are degraded due to limitations called the curse of dimensions. In addition, time series data is preprocessed by applying sliding window technique and time series decomposition for self-correlation analysis. These techniques are the cause of increasing the dimension of data, so it is necessary to supplement them. The anomaly detection field is an old research field, and statistical methods and regression analysis were used in the early days. Currently, there are active studies to apply machine learning and artificial neural network technology to this field. Statistically based methods are difficult to apply when data is non-homogeneous, and do not detect local outliers well. The regression analysis method compares the predictive value and the actual value after learning the regression formula based on the parametric statistics and it detects abnormality. Anomaly detection using regression analysis has the disadvantage that the performance is lowered when the model is not solid and the noise or outliers of the data are included. There is a restriction that learning data with noise or outliers should be used. The autoencoder using artificial neural networks is learned to output as similar as possible to input data. It has many advantages compared to existing probability and linear model, cluster analysis, and map learning. It can be applied to data that does not satisfy probability distribution or linear assumption. In addition, it is possible to learn non-mapping without label data for teaching. However, there is a limitation of local outlier identification of multidimensional data in anomaly detection, and there is a problem that the dimension of data is greatly increased due to the characteristics of time series data. In this study, we propose a CMAE (Conditional Multimodal Autoencoder) that enhances the performance of anomaly detection by considering local outliers and time series characteristics. First, we applied Multimodal Autoencoder (MAE) to improve the limitations of local outlier identification of multidimensional data. Multimodals are commonly used to learn different types of inputs, such as voice and image. The different modal shares the bottleneck effect of Autoencoder and it learns correlation. In addition, CAE (Conditional Autoencoder) was used to learn the characteristics of time series data effectively without increasing the dimension of data. In general, conditional input mainly uses category variables, but in this study, time was used as a condition to learn periodicity. The CMAE model proposed in this paper was verified by comparing with the Unimodal Autoencoder (UAE) and Multi-modal Autoencoder (MAE). The restoration performance of Autoencoder for 41 variables was confirmed in the proposed model and the comparison model. The restoration performance is different by variables, and the restoration is normally well operated because the loss value is small for Memory, Disk, and Network modals in all three Autoencoder models. The process modal did not show a significant difference in all three models, and the CPU modal showed excellent performance in CMAE. ROC curve was prepared for the evaluation of anomaly detection performance in the proposed model and the comparison model, and AUC, accuracy, precision, recall, and F1-score were compared. In all indicators, the performance was shown in the order of CMAE, MAE, and AE. Especially, the reproduction rate was 0.9828 for CMAE, which can be confirmed to detect almost most of the abnormalities. The accuracy of the model was also improved and 87.12%, and the F1-score was 0.8883, which is considered to be suitable for anomaly detection. In practical aspect, the proposed model has an additional advantage in addition to performance improvement. The use of techniques such as time series decomposition and sliding windows has the disadvantage of managing unnecessary procedures; and their dimensional increase can cause a decrease in the computational speed in inference.The proposed model has characteristics that are easy to apply to practical tasks such as inference speed and model management.