• Journal of Advanced Navigation Technology
• /
• v.23 no.1
• /
• pp.84-89
• /
• 2019

This is a pilot study of machine learning based structural health monitoring system using flight data of composite aircraft. In this study, the most suitable machine learning algorithm for structural health monitoring was selected and dimensionality reduction method for application on the actual flight data was conducted. For these tasks, impact test on the cantilever beam with added mass, which is the simulation of damage in the aircraft wing structure was conducted and classification model for damage states (damage location and level) was trained. Through vibration test of cantilever beam with fiber bragg grating (FBG) sensor, data of normal and 12 damaged states were acquired, and the most suitable algorithm was selected through comparison between algorithms like tree, discriminant, support vector machine (SVM), kNN, ensemble. Besides, through neighborhood component analysis (NCA) feature selection, dimensionality reduction which is necessary to deal with high dimensional flight data was conducted. As a result, quadratic SVMs performed best with 98.7% for without NCA and 95.9% for with NCA. It is also shown that the application of NCA improved prediction speed, training time, and model memory.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.95-108
• /
• 2017

Recently, AlphaGo which is Bakuk (Go) artificial intelligence program by Google DeepMind, had a huge victory against Lee Sedol. Many people thought that machines would not be able to win a man in Go games because the number of paths to make a one move is more than the number of atoms in the universe unlike chess, but the result was the opposite to what people predicted. After the match, artificial intelligence technology was focused as a core technology of the fourth industrial revolution and attracted attentions from various application domains. Especially, deep learning technique have been attracted as a core artificial intelligence technology used in the AlphaGo algorithm. The deep learning technique is already being applied to many problems. Especially, it shows good performance in image recognition field. In addition, it shows good performance in high dimensional data area such as voice, image and natural language, which was difficult to get good performance using existing machine learning techniques. However, in contrast, it is difficult to find deep leaning researches on traditional business data and structured data analysis. In this study, we tried to find out whether the deep learning techniques have been studied so far can be used not only for the recognition of high dimensional data but also for the binary classification problem of traditional business data analysis such as customer churn analysis, marketing response prediction, and default prediction. And we compare the performance of the deep learning techniques with that of traditional artificial neural network models. The experimental data in the paper is the telemarketing response data of a bank in Portugal. It has input variables such as age, occupation, loan status, and the number of previous telemarketing and has a binary target variable that records whether the customer intends to open an account or not. In this study, to evaluate the possibility of utilization of deep learning algorithms and techniques in binary classification problem, we compared the performance of various models using CNN, LSTM algorithm and dropout, which are widely used algorithms and techniques in deep learning, with that of MLP models which is a traditional artificial neural network model. However, since all the network design alternatives can not be tested due to the nature of the artificial neural network, the experiment was conducted based on restricted settings on the number of hidden layers, the number of neurons in the hidden layer, the number of output data (filters), and the application conditions of the dropout technique. The F1 Score was used to evaluate the performance of models to show how well the models work to classify the interesting class instead of the overall accuracy. The detail methods for applying each deep learning technique in the experiment is as follows. The CNN algorithm is a method that reads adjacent values from a specific value and recognizes the features, but it does not matter how close the distance of each business data field is because each field is usually independent. In this experiment, we set the filter size of the CNN algorithm as the number of fields to learn the whole characteristics of the data at once, and added a hidden layer to make decision based on the additional features. For the model having two LSTM layers, the input direction of the second layer is put in reversed position with first layer in order to reduce the influence from the position of each field. In the case of the dropout technique, we set the neurons to disappear with a probability of 0.5 for each hidden layer. The experimental results show that the predicted model with the highest F1 score was the CNN model using the dropout technique, and the next best model was the MLP model with two hidden layers using the dropout technique. In this study, we were able to get some findings as the experiment had proceeded. First, models using dropout techniques have a slightly more conservative prediction than those without dropout techniques, and it generally shows better performance in classification. Second, CNN models show better classification performance than MLP models. This is interesting because it has shown good performance in binary classification problems which it rarely have been applied to, as well as in the fields where it's effectiveness has been proven. Third, the LSTM algorithm seems to be unsuitable for binary classification problems because the training time is too long compared to the performance improvement. From these results, we can confirm that some of the deep learning algorithms can be applied to solve business binary classification problems.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.16 no.4
• /
• pp.153-163
• /
• 2017

This research is based on the previous research that personally preferred safe distance, rotating angle and speed are differentiated. Thus, we use machine learning model for recognizing maneuvering modes trained per personal or per similar driving pattern groups, and we evaluate automatic driving according to maneuvering modes. By utilizing driving knowledge, we subdivided 8 kinds of longitudinal modes and 4 kinds of lateral modes, and by combining the longitudinal and lateral modes, we build 21 kinds of maneuvering modes. we train the labeled data set per time stamp through RNN, LSTM and Bi-LSTM models by the trips of drivers, which are supervised deep learning models, and evaluate the maneuvering modes of automatic driving for the test data set. The evaluation dataset is aggregated of living trips of 3,000 populations by VTTI in USA for 3 years and we use 1500 trips of 22 people and training, validation and test dataset ratio is 80%, 10% and 10%, respectively. For recognizing longitudinal 8 kinds of maneuvering modes, RNN achieves better accuracy compared to LSTM, Bi-LSTM. However, Bi-LSTM improves the accuracy in recognizing 21 kinds of longitudinal and lateral maneuvering modes in comparison with RNN and LSTM as 1.54% and 0.47%, respectively.

• Journal of Korean Ophthalmic Optics Society
• /
• v.20 no.4
• /
• pp.491-499
• /
• 2015

Purpose: This study was to examine how decline of visual function affects visual perception by assessing visual perception after improving visual function through visual training, and observing the change in the cognitive ability of visual perception. Methods: This study analyzes the visual perceptual evaluation (TVPS_R) of 23 children below age 13($8.75{\pm}1.66$) who have visual abnormalities, and improves visual function after conducting vision training (vision therapy) of the children. Results: Convergence increased from average $3.39{\pm}2.52{\Delta}$ (prism) to $13.87{\pm}6.04{\Delta}$ in the measurement of long-distance disparate points, and from average $5.48{\pm}3.42{\Delta}$ to $18.43{\pm}7.58{\Delta}$ in the measurement of short-distance disparate points. Short-distance diplopia points increased from $25.87{\pm}7.33cm$ to $7.48{\pm}2.87cm$, and as for accommodative insufficiency, short-distance blur points increased from $19.57{\pm}7.16cm$ to $7.09{\pm}1.88cm$. In the visual perceptual evaluation performed before and after improving visual function, 6 items except visual memory showed statistically significant improvement. By order of significant improvement, response gap was highest with $17.74{\pm}16.94$(p=0.000) in visual closure, followed by $15.65{\pm}17.11$(p=0.000) in visual sequential-memory, $13.65{\pm}16.63$(p=0.001) in visual figure-ground, $12.74{\pm}18.41$(p=0.003) in visual form-constancy, $6.48{\pm}10.07$ (p=0.005) in visual discrimination, and $4.17{\pm}9.33$(p=0.043) in visual spatial-relationship. In the visual perception quotient that added up these scores, the response gap was $15.22{\pm}8.66$(p=0.000), showing a more significant result. Conclusions: Vision training enables efficient visual processing and improves visual perceptual ability. It was confirmed that improvement of visual function through visual training not only improves abnormal visual function but also affects visual perception of children such as learning, perception and recognition.

• KNOM Review
• /
• v.23 no.2
• /
• pp.18-28
• /
• 2020

With the development of the virtual community, the benefits that IT technology provides to people in fields such as healthcare, industry, communication, and culture are increasing, and the quality of life is also improving. Accordingly, there are various malicious attacks targeting the developed network environment. Firewalls and intrusion detection systems exist to detect these attacks in advance, but there is a limit to detecting malicious attacks that are evolving day by day. In order to solve this problem, intrusion detection research using machine learning is being actively conducted, but false positives and false negatives are occurring due to imbalance of the learning dataset. In this paper, a Random Oversampling method is used to solve the unbalance problem of the UNSW-NB15 dataset used for network intrusion detection. And through experiments, we compared and analyzed the accuracy, precision, recall, F1-score, training and prediction time, and hardware resource consumption of the models. Based on this study using the Random Oversampling method, we develop a more efficient network intrusion detection model study using other methods and high-performance models that can solve the unbalanced data problem.

• 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.

• The KIPS Transactions:PartA
• /
• v.11A no.4
• /
• pp.243-250
• /
• 2004

Monitoring is used to see if a real-time system provides a service on time. Generally, monitoring for real-time focuses on investigating the current status of a real-time system. To support a stable performance of a real-time system, it should have not only a function to see the current status of real-time process but also a function to predict executions of real-time processes, however. The legacy prediction model has some limitation to apply it to a real-time monitoring. First, it performs a static prediction after a real-time process finished. Second, it needs a statistical pre-analysis before a prediction. Third, transition probability and data about clustering is not based on the current data. We propose the execution prediction model based on learning algorithm to solve these problems and apply it to real-time monitoring. This model gets rid of unnecessary pre-processing and supports a precise prediction based on current data. In addition, this supports multi-level prediction by a trend analysis of past execution data. Most of all, We designed the model to support dynamic prediction which is performed within a real-time process' execution. The results from some experiments show that the judgment accuracy is greater than 80% if the size of a training set is set to over 10, and, in the case of the multi-level prediction, that the prediction difference of the multi-level prediction is minimized if the number of execution is bigger than the size of a training set. The execution prediction model proposed in this model has some limitation that the model used the most simplest learning algorithm and that it didn't consider the multi-regional space model managing CPU, memory and I/O data. The execution prediction model based on a learning algorithm proposed in this paper is used in some areas related to real-time monitoring and control.

• Phonetics and Speech Sciences
• /
• v.13 no.3
• /
• pp.71-78
• /
• 2021

Most neural-network-based speech synthesis models utilize neural vocoders to convert mel-scaled spectrograms into high-quality, human-like voices. However, neural vocoders combined with mel-scaled spectrogram prediction models demand considerable computer memory and time during the training phase and are subject to slow inference speeds in an environment where GPU is not used. This problem does not arise in linear spectrogram prediction models, as they do not use neural vocoders, but these models suffer from low voice quality. As a solution, this paper proposes a Tacotron 2 and Transformer-based linear spectrogram prediction model that produces high-quality speech and does not use neural vocoders. Experiments suggest that this model can serve as the foundation of a high-quality text-to-speech model with fast inference speed.

• Journal of Korea Water Resources Association
• /
• v.54 no.10
• /
• pp.795-805
• /
• 2021

In this study, after developing an LSTM-based deep learning model for estimating daily runoff in the Soyang River Dam basin, the accuracy of the model for various combinations of model structure and input data was investigated. A model was built based on the database consisting of average daily precipitation, average daily temperature, average daily wind speed (input up to here), and daily average flow rate (output) during the first 12 years (1997.1.1-2008.12.31). The Nash-Sutcliffe Model Efficiency Coefficient (NSE) and RMSE were examined for validation using the flow discharge data of the later 12 years (2009.1.1-2020.12.31). The combination that showed the highest accuracy was the case in which all possible input data (12 years of daily precipitation, weather temperature, wind speed) were used on the LSTM model structure with 64 hidden units. The NSE and RMSE of the verification period were 0.862 and 76.8 m³/s, respectively. When the number of hidden units of LSTM exceeds 500, the performance degradation of the model due to overfitting begins to appear, and when the number of hidden units exceeds 1000, the overfitting problem becomes prominent. A model with very high performance (NSE=0.8~0.84) could be obtained when only 12 years of daily precipitation was used for model training. A model with reasonably high performance (NSE=0.63-0.85) when only one year of input data was used for model training. In particular, an accurate model (NSE=0.85) could be obtained if the one year of training data contains a wide magnitude of flow events such as extreme flow and droughts as well as normal events. If the training data includes both the normal and extreme flow rates, input data that is longer than 5 years did not significantly improve the model performance.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.4
• /
• pp.2060-2077
• /
• 2019

Recently, mobile healthcare services have attracted significant attention because of the emerging development and supply of diverse wearable devices. Smartwatches and health bands are the most common type of mobile-based wearable devices and their market size is increasing considerably. However, simple value comparisons based on accumulated data have revealed certain problems, such as the standardized nature of health management and the lack of personalized health management service models. The convergence of information technology (IT) and biotechnology (BT) has shifted the medical paradigm from continuous health management and disease prevention to the development of a system that can be used to provide ground-based medical services regardless of the user's location. Moreover, the IT-BT convergence has necessitated the development of lifestyle improvement models and services that utilize big data analysis and machine learning to provide mobile healthcare-based personal health management and disease prevention information. Users' health data, which are specific as they change over time, are collected by different means according to the users' lifestyle and surrounding circumstances. In this paper, we propose a prediction model of user physical activity that uses data characteristics-based long short-term memory (DC-LSTM) recurrent neural networks (RNNs). To provide personalized services, the characteristics and surrounding circumstances of data collectable from mobile host devices were considered in the selection of variables for the model. The data characteristics considered were ease of collection, which represents whether or not variables are collectable, and frequency of occurrence, which represents whether or not changes made to input values constitute significant variables in terms of activity. The variables selected for providing personalized services were activity, weather, temperature, mean daily temperature, humidity, UV, fine dust, asthma and lung disease probability index, skin disease probability index, cadence, travel distance, mean heart rate, and sleep hours. The selected variables were classified according to the data characteristics. To predict activity, an LSTM RNN was built that uses the classified variables as input data and learns the dynamic characteristics of time series data. LSTM RNNs resolve the vanishing gradient problem that occurs in existing RNNs. They are classified into three different types according to data characteristics and constructed through connections among the LSTMs. The constructed neural network learns training data and predicts user activity. To evaluate the proposed model, the root mean square error (RMSE) was used in the performance evaluation of the user physical activity prediction method for which an autoregressive integrated moving average (ARIMA) model, a convolutional neural network (CNN), and an RNN were used. The results show that the proposed DC-LSTM RNN method yields an excellent mean RMSE value of 0.616. The proposed method is used for predicting significant activity considering the surrounding circumstances and user status utilizing the existing standardized activity prediction services. It can also be used to predict user physical activity and provide personalized healthcare based on the data collectable from mobile host devices.