• Korean Journal of Agricultural and Forest Meteorology
• /
• v.21 no.1
• /
• pp.29-41
• /
• 2019

Terrestrial Gross Primary Production (GPP) is the largest global carbon flux, and forest ecosystems are important because of the ability to store much more significant amounts of carbon than other terrestrial ecosystems. There have been several attempts to estimate GPP using mechanism-based models. However, mechanism-based models including biological, chemical, and physical processes are limited due to a lack of flexibility in predicting non-stationary ecological processes, which are caused by a local and global change. Instead mechanism-free methods are strongly recommended to estimate nonlinear dynamics that occur in nature like GPP. Therefore, we used the mechanism-free machine learning techniques to estimate the daily GPP. In this study, support vector machine (SVM), random forest (RF) and artificial neural network (ANN) were used and compared with the traditional multiple linear regression model (LM). MODIS products and meteorological parameters from eddy covariance data were employed to train the machine learning and LM models from 2006 to 2013. GPP prediction models were compared with daily GPP from eddy covariance measurement in a deciduous forest in South Korea in 2014 and 2015. Statistical analysis including correlation coefficient (R), root mean square error (RMSE) and mean squared error (MSE) were used to evaluate the performance of models. In general, the models from machine-learning algorithms (R = 0.85 - 0.93, MSE = 1.00 - 2.05, p < 0.001) showed better performance than linear regression model (R = 0.82 - 0.92, MSE = 1.24 - 2.45, p < 0.001). These results provide insight into high predictability and the possibility of expansion through the use of the mechanism-free machine-learning models and remote sensing for predicting non-stationary ecological processes such as seasonal GPP.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.1031-1042
• /
• 2023

In this study, a deep learning model was developed to predict the yield of cabbage and radish, one of the five major supply and demand management vegetables, using satellite images of Landsat 8. To predict the yield of cabbage and radish in Gangwon-do from 2015 to 2020, satellite images from June to September, the growing period of cabbage and radish, were used. Normalized difference vegetation index, enhanced vegetation index, lead area index, and land surface temperature were employed in this study as input data for the yield model. Crop yields can be effectively predicted using satellite images because satellites collect continuous spatiotemporal data on the global environment. Based on the model developed previous study, a model designed for input data was proposed in this study. Using time series satellite images, convolutional neural network, a deep learning model, was used to predict crop yield. Landsat 8 provides images every 16 days, but it is difficult to acquire images especially in summer due to the influence of weather such as clouds. As a result, yield prediction was conducted by splitting June to July into one part and August to September into two. Yield prediction was performed using a machine learning approach and reference models , and modeling performance was compared. The model's performance and early predictability were assessed using year-by-year cross-validation and early prediction. The findings of this study could be applied as basic studies to predict the yield of field crops in Korea.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.6
• /
• pp.603-616
• /
• 2023

In this study, machine learning-based clustering and classification of residential noise in apartment buildings was conducted using frequency and temporal characteristics. First, a residential noise source dataset was constructed . The residential noise source dataset was consisted of floor impact, airborne, plumbing and equipment noise, environmental, and construction noise. The clustering of residential noise was performed by K-Means clustering method. For frequency characteristics, L_eq and L_max values were derived for 1/1 and 1/3 octave band for each sound source. For temporal characteristics, Leq values were derived at every 6 ms through sound pressure level analysis for 5 s. The number of k in K-Means clustering method was determined through the silhouette coefficient and elbow method. The clustering of residential noise source by frequency characteristic resulted in three clusters for both L_eq and L_max analysis. Temporal characteristic clustered residential noise source into 9 clusters for L_eq and 11 clusters for L_max. Clustering by frequency characteristic clustered according to the proportion of low frequency band. Then, to utilize the clustering results, the residential noise source was classified using three kinds of machine learning. The results of the residential noise classification showed the highest accuracy and f1-score for data labeled with L_eq values in 1/3 octave bands, and the highest accuracy and f1-score for classifying residential noise sources with an Artificial Neural Network (ANN) model using both frequency and temporal features, with 93 % accuracy and 92 % f1-score.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_2
• /
• pp.1605-1613
• /
• 2023

The physical properties of the ocean interior are determined by temperature and salinity. To observe them, we rely on satellite observations for broad regions of oceans. However, the satellite for salinity measurement, Soil Moisture Active Passive (SMAP), has low temporal and spatial resolutions; thus, more is needed to resolve the fast-changing coastal environment. To overcome these limitations, the algorithm to use the Geostationary Ocean Color Imager-II (GOCI-II) of the Geo-Kompsat-2B (GK-2B) was developed as the inputs for a Multi-layer Perceptron Neural Network (MPNN). The result shows that coefficient of determination (R²), root mean square error (RMSE), and relative root mean square error (RRMSE) between GOCI-II based sea surface salinity (SSS) (GOCI-II SSS) and SMAP was 0.94, 0.58 psu, and 1.87%, respectively. Furthermore, the spatial variation of GOCI-II SSS was also very uniform, with over 0.8 of R² and less than 1 psu of RMSE. In addition, GOCI-II SSS was also compared with SSS of Ieodo Ocean Research Station (I-ORS), suggesting that the result was slightly low, which was further analyzed for the following reasons. We further illustrated the valuable information of high spatial and temporal variation of GOCI-II SSS to analyze SSS variation by the 11th typhoon, Hinnamnor, in 2022. We used the mean and standard deviation (STD) of one day of GOCI-II SSS, revealing the high spatial and temporal changes. Thus, this study will shed light on the research for monitoring the highly changing marine environment.

• Journal of Intelligence and Information Systems
• /
• v.24 no.3
• /
• pp.1-19
• /
• 2018

Large amount of data is now available for research and business sectors to extract knowledge from it. This data can be in the form of unstructured data such as audio, text, and image data and can be analyzed by deep learning methodology. Deep learning is now widely used for various estimation, classification, and prediction problems. Especially, fashion business adopts deep learning techniques for apparel recognition, apparel search and retrieval engine, and automatic product recommendation. The core model of these applications is the image classification using Convolutional Neural Networks (CNN). CNN is made up of neurons which learn parameters such as weights while inputs come through and reach outputs. CNN has layer structure which is best suited for image classification as it is comprised of convolutional layer for generating feature maps, pooling layer for reducing the dimensionality of feature maps, and fully-connected layer for classifying the extracted features. However, most of the classification models have been trained using online product image, which is taken under controlled situation such as apparel image itself or professional model wearing apparel. This image may not be an effective way to train the classification model considering the situation when one might want to classify street fashion image or walking image, which is taken in uncontrolled situation and involves people's movement and unexpected pose. Therefore, we propose to train the model with runway apparel image dataset which captures mobility. This will allow the classification model to be trained with far more variable data and enhance the adaptation with diverse query image. To achieve both convergence and generalization of the model, we apply Transfer Learning on our training network. As Transfer Learning in CNN is composed of pre-training and fine-tuning stages, we divide the training step into two. First, we pre-train our architecture with large-scale dataset, ImageNet dataset, which consists of 1.2 million images with 1000 categories including animals, plants, activities, materials, instrumentations, scenes, and foods. We use GoogLeNet for our main architecture as it has achieved great accuracy with efficiency in ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Second, we fine-tune the network with our own runway image dataset. For the runway image dataset, we could not find any previously and publicly made dataset, so we collect the dataset from Google Image Search attaining 2426 images of 32 major fashion brands including Anna Molinari, Balenciaga, Balmain, Brioni, Burberry, Celine, Chanel, Chloe, Christian Dior, Cividini, Dolce and Gabbana, Emilio Pucci, Ermenegildo, Fendi, Giuliana Teso, Gucci, Issey Miyake, Kenzo, Leonard, Louis Vuitton, Marc Jacobs, Marni, Max Mara, Missoni, Moschino, Ralph Lauren, Roberto Cavalli, Sonia Rykiel, Stella McCartney, Valentino, Versace, and Yve Saint Laurent. We perform 10-folded experiments to consider the random generation of training data, and our proposed model has achieved accuracy of 67.2% on final test. Our research suggests several advantages over previous related studies as to our best knowledge, there haven't been any previous studies which trained the network for apparel image classification based on runway image dataset. We suggest the idea of training model with image capturing all the possible postures, which is denoted as mobility, by using our own runway apparel image dataset. Moreover, by applying Transfer Learning and using checkpoint and parameters provided by Tensorflow Slim, we could save time spent on training the classification model as taking 6 minutes per experiment to train the classifier. This model can be used in many business applications where the query image can be runway image, product image, or street fashion image. To be specific, runway query image can be used for mobile application service during fashion week to facilitate brand search, street style query image can be classified during fashion editorial task to classify and label the brand or style, and website query image can be processed by e-commerce multi-complex service providing item information or recommending similar item.

• Journal of KIISE
• /
• v.43 no.10
• /
• pp.1085-1093
• /
• 2016

Community-based Question Answering system is a system which provides answers for each question from the documents uploaded on web communities. In order to enhance the capacity of question analysis, former methods have developed specific rules suitable for a target region or have applied machine learning to partial processes. However, these methods incur an excessive cost for expanding fields or lead to cases in which system is overfitted for a specific field. This paper proposes a multiple machine learning method which automates the overall process by adapting appropriate machine learning in each procedure for efficient processing of community-based Question Answering system. This system can be divided into question analysis part and answer selection part. The question analysis part consists of the question focus extractor, which analyzes the focused phrases in questions and uses conditional random fields, and the question type classifier, which classifies topics of questions and uses support vector machine. In the answer selection part, the we trains weights that are used by the similarity estimation models through an artificial neural network. Also these are a number of cases in which the results of morphological analysis are not reliable for the data uploaded on web communities. Therefore, we suggest a method that minimizes the impact of morphological analysis by using character features in the stage of question analysis. The proposed system outperforms the former system by showing a Mean Average Precision criteria of 0.765 and R-Precision criteria of 0.872.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.2
• /
• pp.79-88
• /
• 2004

Gaze detection is to locate the position on a monitor screen where a user is looking by computer vision. Gaze detection systems have numerous fields of application. They are applicable to the man-machine interface for helping the handicapped to use computers and the view control in three dimensional simulation programs. In our work, we implement it with a computer vision system setting a IR-LED based single camera. To detect the gaze position, we locate facial features, which is effectively performed with IR-LED based camera and SVM(Support Vector Machine). When a user gazes at a position of monitor, we can compute the 3D positions of those features based on 3D rotation and translation estimation and affine transform. Finally, the gaze position by the facial movements is computed from the normal vector of the plane determined by those computed 3D positions of features. In addition, we use a trained neural network to detect the gaze position by eye's movement. As experimental results, we can obtain the facial and eye gaze position on a monitor and the gaze position accuracy between the computed positions and the real ones is about 4.8 cm of RMS error.

• Journal of Korea Water Resources Association
• /
• v.51 no.12
• /
• pp.1207-1216
• /
• 2018

Discharge or water level predictions at tidally affected river reaches are currently still a great challenge in hydrological practices. This research aims to predict water level of the tide dominated site, Jamsu bridge in the Han River downstream. Physics-based hydrodynamic approaches are sometimes not applicable for water level prediction in such a tidal river due to uncertainty sources like rainfall forecasting data. In this study, TensorFlow deep learning framework was used to build a deep neural network based LSTM model and its applications. The LSTM model was trained based on 3 data sets having 10-min temporal resolution: Paldang dam release, Jamsu bridge water level, predicted tidal level for 6 years (2011~2016) and then predict the water level time series given the six lead times: 1, 3, 6, 9, 12, 24 hours. The optimal hyper-parameters of LSTM model were set up as follows: 6 hidden layers number, 0.01 learning rate, 3000 iterations. In addition, we changed the key parameter of LSTM model, sequence length, ranging from 1 to 6 hours to test its affect to prediction results. The LSTM model with the 1 hr sequence length led to the best performing prediction results for the all cases. In particular, it resulted in very accurate prediction: RMSE (0.065 cm) and NSE (0.99) for the 1 hr lead time prediction case. However, as the lead time became longer, the RMSE increased from 0.08 m (1 hr lead time) to 0.28 m (24 hrs lead time) and the NSE decreased from 0.99 (1 hr lead time) to 0.74 (24 hrs lead time), respectively.

• Korean Chemical Engineering Research
• /
• v.59 no.2
• /
• pp.209-218
• /
• 2021

First principle-based modeling studies have been performed to improve the heat exchange efficiency of ORV and optimize operation, but the heat transfer coefficient of ORV is an irregular system according to time and location, and it undergoes a complex modeling process. In this study, FNN, LSTM, and AutoML-based modeling were performed to confirm the effectiveness of data-based modeling for complex systems. The prediction accuracy indicated high performance in the order of LSTM > AutoML > FNN in MSE. The performance of AutoML, an automatic design method for machine learning models, was superior to developed FNN, and the total time required for model development was 1/15 compared to LSTM, showing the possibility of using AutoML. The prediction of NG and seawater discharged temperatures using LSTM and AutoML showed an error of less than 0.5K. Using the predictive model, real-time optimization of the amount of LNG vaporized that can be processed using ORV in winter is performed, confirming that up to 23.5% of LNG can be additionally processed, and an ORV optimal operation guideline based on the developed dynamic prediction model was presented.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.14 no.5
• /
• pp.31-39
• /
• 2018

Recently, there has been active studies to provide a user-friendly interface in a virtual reality environment by recognizing user hand gestures based on deep learning. However, most studies use separate sensors to obtain hand information or go through pre-process for efficient learning. It also fails to take into account changes in the external environment, such as changes in lighting or some of its hands being obscured. This paper proposes a hand gesture recognition method based on deep learning that is strong in external environments without the need for pre-process of RGB images obtained from general webcam. In this paper we improve the VGGNet and the GoogLeNet structures and compared the performance of each structure. The VGGNet and the GoogLeNet structures presented in this paper showed a recognition rate of 93.88% and 93.75%, respectively, based on data containing dim, partially obscured, or partially out-of-sight hand images. In terms of memory and speed, the GoogLeNet used about 3 times less memory than the VGGNet, and its processing speed was 10 times better. The results of this paper can be processed in real-time and used as a hand gesture interface in various areas such as games, education, and medical services in a virtual reality environment.