Search

In manufacturing, it is a fatal problem to stop production due to a failure of facilities. Therefore, it is important to predict defects in facilities in advance and proactively maintain them. With the recent development of machine learning and deep learning technologies, research on predictive maintenance is also becoming active, and the performance of machine learning and deep learning technologies varies depending on the dataset. However, the existing dataset simply exists in normal and defective states, and the state between defects and normal was not simulated. Therefore, in this paper, a dataset was constructed by collecting additional attention data containing signals between normal and defective, and a simple classification experiment was conducted to verify the effectiveness of the data. Using the Prognostic dataset proposed in this paper, it is expected to be able to move toward a true predictive maintenance system that performs actions in advance, not after an accident. [ABSTRACT FROM AUTHOR]

This paper analyzes the characteristics of attribute relationships in SNOMED CT, a representative medical terminology system used for documenting diseases, treatments, and other health-related terms in medical records. In SNOMED CT, concepts that have clinical meanings are specifically defined by connecting to other concepts using attribute relationships, forming a complex ontology. This study analyzes the diverse characteristics of attribute relationships using a machine learning technique called association rule analysis. The study results showed that, first, attributes for each of the 19 concept areas are clearly distinguished; second, key attributes with a support of 0.2 or higher were found in the clinical finding and procedure area, which are representative concept areas; third, 14 attribute pairs with a dependency characteristic showing a confidence of 0.95 or higher are identified. Additionally, we visualized support and confidence between attribute relationships through a network graph to understand the associations between attributes. [ABSTRACT FROM AUTHOR]

Although CNN shows high performance in the image recognition field, it has great disadvantages. They are to take a long time to perform machine learning due to a lack of system resources, and to consume a lot of power due to the great volume of computation. A convolution layer is a key element of convolutional neural network processing. This paper presents a convolution layer hardware using processing elements with a low-power multiplier. The low-power multiplier reduces the switching activity by increasing the exchange rate between the multiplier and multiplicand. The proposed convolution layer hardware was implemented on the Intel DE1-SoC FPGA Board using Verilog-HDL. The performance was verified by comparing the exchange rate with the existing multipliers when performing multiplications to process convolutional operations for the MNIST image database. [ABSTRACT FROM AUTHOR]

In this paper, a study was conducted to predict the defects in MCT processed finished products using a total of eight types of supervised learning models, one of the representative deep learning algorithms for artificial intelligence technologies. As a result of comparing the performance indicators and required execution time among machine learning models, it was confirmed that XGBM, GBM, Light GBM and Stacking Ensemble showed the highest accuracy, precision, recall, F1 score and AUC. Above all, it was confirmed that the Stacking Ensemble model was the most suitable algorithm for fast defect prediction because it showed the shortest learning execution time of 0.53 seconds. Besides, as a result of the analyzation of the relative importance of characteristic variables using the XG Boost model, it was found that the W-axis absolute coordinate and the X-axis absolute coordinate have a crucial effect on the defect. [ABSTRACT FROM AUTHOR]

In this paper, we present a novel design approach for a wideband non-uniformly spaced linear array (NUSLA) to suppress the effect of interference signals. Notably, a uniform linear array (ULA), which is easy to handle, is widely utilized to generate nulls over a wide band; however, its beamforming performance is limited owing to the ULA structure. Although a NUSLA with nonlinear spacing addresses nonlinear problems that are difficult to handle mathematically, it can exhibit performance improvements surpassing those of ULA structures. However, an additional constraint is required to generate nulls at a specific position, and this has not yet been studied for null generation using wideband NUSLA. In this paper, we propose a novel cost function for designing a wideband NUSLA, which generates a null at the desired position, and we utilize the modified reinforcement learning algorithm (MORELA), which is a heuristic optimization algorithm based on reinforcement learning (RL), to minimize the proposed cost function and analyze the optimized antenna array and weights. Further, we compare the performance of the proposed MORELA based on RL with that of existing heuristic optimization algorithms via computer simulations. [ABSTRACT FROM AUTHOR]

Recently, the development of machine learning and cloud computing has led to the rise of Privacy-Preserving Machine Learning (PPML). Homomorphic encryption enables computations over encrypted data without decryption, and it is secure against adversaries that use quantum computers. This study focuses on learning a credit evaluation model to determine loan eligibility using homomorphic encryption. The study outlines a concrete scenario for learning and using the credit evaluation model in a privacy-preserving way and defines the security requirements. To optimize the credit evaluation model, we use a Support Vector Machine (SVM) training algorithm with high classification accuracy. This paper proposes an efficient credit evaluation model learning system using LS(Linear Square)-SVM model, which is recomposed to an HE-friendly computation. It enables learning of a credit evaluation model over encrypted data while protecting user’s sensitive information and increases the accuracy of credit evaluation predictions. [ABSTRACT FROM AUTHOR]

COVID-19 occurred in China in December 2019 and spread to the world. Untact culture becoming commonplace due to the spread of COVID-19, is accelerating digital transformation across society. The government is increasing government fund to overcome national crisis situations such as COVID-19. In order to efficiently execute these government funds, the government and policy financial institutions need to come up with two measures. One is to establish a transparent government fund execution process, and the other is to secure the objectivity of the review process. In this paper, to solve this problem, we designed the execution process of government funds from the government to end users with Consortium Blockchain, and designed a machine learning algorithm for internal review of policy financial institutions. The research data proposed in this paper will be helpful in setting the future government policy direction for government fund execution. [ABSTRACT FROM AUTHOR]

In this paper, we explore customer churn in a telecommunication company, measuring the probability of users discontinuing their service. Our investigation employs various machine learning models, including Decision Tree, Random Forest, XGBoost, LightGBM, SVM, Logistic Regression, and Deep Neural Network. To accomplish this, we leverage data collected from a California-based telecommunication company, initially containing 38 feature data. In order to reduce complexity, we apply Lasso regression to select the five most crucial features for determining customer churn: household situation, service satisfaction, loyalty, payment capability, and contract type. Through performance evaluation, we demonstrate that accurate predictions of customer churn can be achieved even with five features, emphasizing the significance of feature selection. [ABSTRACT FROM AUTHOR]

The condition-based maintenance system(CBMS) is a technology to improve the survivability and operational performance of naval ships. It monitors the condition of major equipment (propulsion system, power system, etc.) and performs functions such as displaying trends through performance evaluation, generating alarm conditions, fault diagnosis, and predicting remaining life. This paper studies ways to increase the accuracy and efficiency of machine learning model design for abnormal diagnosis of Naval propulsion system equipment. The Random Forest and Support Vector Regression(SVR) models used in overseas cases and the XGBoost and Light GBM models, which are most commonly used in various fields, were used to calculate the normal prediction values(Baseline) based on actual operational naval ship propulsion system equipment, and performed condition diagnosis through comparison with real-time measurements. Then, the performance of each regression model was compared and analyzed through RMSE, R2 Score, and learning time. [ABSTRACT FROM AUTHOR]

In this paper, we suggest a reinforcement learning algorithm in satellite-aided communication system. Satellite-aided communication are drawing attention because it can increase stability and robustness of wireless communication. To improve the efficiency of satellite-aided communication, beamhopping method is introduced. In beamhopping, satellite beams are divided and operated in time-based system. The actual efficiency of beamhopping can be depend on the hopping pattern. To determine beamhopping pattern, we introduced multiplay Thompson sampling which is a family of reinforcement learning. We mathematically modeled the features of satellite communication and then applied reinforcement learning. By means of simulations, the performances of proposed learning were analyzed in terms of capacity and regret. [ABSTRACT FROM AUTHOR]

This paper proposes a sign language recognition system based on machine learning, which can infer the meaning of body movement in realtime video. It has a feature that motion vectors constructed with keypoint data of moving body are used for machine learning. This enables the proposed system to reduce the training time of an artificial neural network model and to enhance the recognition accuracy. Since those motion vectors have the same size and format regardless of the different expression times of words, the confidence level of an inference could be further increased. To correctly divide a part of video corresponding to each word in a sentence, we propose an iterative evaluation method of confidence levels for inferences with the gradual length adjustment of video fragment. The experimental results showed that our system outperforms a conventional method using the video itself for training and inference, in training time, inference speed, confidence level of inference and accuracy of video fragmentation. [ABSTRACT FROM AUTHOR]

As machine learning technologies are recently applied to various industries, the importance of XAI (eXplainable Artificial Intelligence) is being emphasized to understand and effectively manage the behavior of developed AI models. In time series models, XAI issues are also critical, however if the number of features is not sufficient, it is not easy to apply XAI techniques. To this end, this paper presents a multi-feature extraction method to ensure an explainable model from univariate time series. First, a multivariate time series is extracted from a univariate time series using a time series decomposition and feature engineering library. Next, a prediction model is generated from the multivariate time series using a machine learning algorithm. Finally, the explanation results for the model created using the XAI technique are represented. In this regard, the proposed method is verified through an experiment using load data of distribution lines of Korea Electric Power Corporation. [ABSTRACT FROM AUTHOR]

In this paper, we intend to improve the network lifetime by improving the energy efficiency of sensor nodes in a wireless sensor network by utilizing machine learning using K-means clustering algorithm. A wireless sensor network is a wireless network composed of physical devices including batteries as physical sensors. Due to the characteristics of sensor nodes, all resources must be efficiently used to minimize energy consumption to maximize network lifetime. A cluster based approach is used to manage groups of relatively large numbers of nodes. In the proposed protocol, by improving the existing LEACH algorithm, we propose a clustering algorithm that selects a cluster head using a cluster based approach and a location based approach. The performance results to be improved were measured using Matlab simulation. Through the experimental results, K-means clustering was applied to the energy efficiency part. By utilizing K-means, it is confirmed that energy efficiency is improved and the lifetime of the entire network is extended. [ABSTRACT FROM AUTHOR]

Recently, big data analysis can use various techniques according to the development of machine learning. Big data collected in reality lacks an automated refining technique for the same or similar terms based on semantic analysis of the relationship between words. Since most of the big data is described in general sentences, it is difficult to understand the meaning and terms of the sentences. To solve these problems, it is necessary to understand the morphological analysis and meaning of sentences. Accordingly, NLP, a technique for analyzing natural language, can understand the word's relationship and sentences. Among the NLP techniques, the transformer has been proposed as a way to solve the disadvantages of RNN by using self-attention composed of an encoder-decoder structure of seq2seq. In this paper, transformers are used as a way to form associations between words in order to understand the words and phrases of sentences extracted from big data. [ABSTRACT FROM AUTHOR]

Cyber-attacks such as smishing and hacking mail exploiting COVID-19, political and social issues, have recently been continuous. Machine learning and deep learning technology research are conducted to prevent any damage due to cyber-attacks inducing malicious links to breach personal data. It has been concluded as a lack of basis to judge the attacks to be malicious in previous studies since the features of data set were excessively simple. In this paper, nine main features of three types, “URL Days”, “URL Word”, and “URL Abnormal”, were proposed in addition to lexical features of URL which have been reflected in previous research. F1-Score and accuracy index were measured through four different types of machine learning algorithms. An improvement of 0.9% in a result and the highest value, 98.5%, were examined in F1-Score and accuracy through comparatively analyzing an existing research. These outcomes proved the main features contribute to elevating the values in both accuracy and performance. [ABSTRACT FROM AUTHOR]

Among the data used for the diagnosis of calf disease, feces play an important role in disease diagnosis. In the image of calf feces, the health status can be known by the shape, color, and texture. For the fecal image that can identify the health status, data of 207 normal calves and 158 calves with diarrhea were pre-processed according to fecal status and used. In this paper, images of fecal variables are detected among the collected calf data and images are trained by applying GLCM-CNN, which combines the properties of CNN and GLCM, on a dataset containing disease symptoms using convolutional network technology. There was a significant difference between CNN's 89.9% accuracy and GLCM-CNN, which showed 91.7% accuracy, and GLCM-CNN showed a high accuracy of 1.8%. [ABSTRACT FROM AUTHOR]

In this paper, we study a scheduling problem based on reinforcement learning for overlay device-to-device (D2D) communication networks. Even though various technologies for D2D communication networks using Q-learning, which is one of reinforcement learning models, have been studied, Q-learning causes a tremendous complexity as the number of states and actions increases. In order to solve this problem, D2D communication technologies based on Deep Q Network (DQN) have been studied. In this paper, we thus design a DQN model by considering the characteristics of wireless communication systems, and propose a distributed scheduling scheme based on the DQN model that can reduce feedback and signaling overhead. The proposed model trains all parameters in a centralized manner, and transfers the final trained parameters to all mobiles. All mobiles individually determine their actions by using the transferred parameters. We analyze the performance of the proposed scheme by computer simulation and compare it with optimal scheme, opportunistic selection scheme and full transmission scheme. [ABSTRACT FROM AUTHOR]

Machine learning constructs an objective function from learning data, and predicts the result of the data generated by checking the objective function through test data. In machine learning, input data is subjected to a normalisation process through a preprocessing. In the case of numerical data, normalization is standardized by using the average and standard deviation of the input data. In the case of nominal data, which is non-numerical data, it is converted into a one-hot code form. However, this preprocessing alone cannot solve the problem. For this reason, we propose a method that uses ontology to normalize input data in this paper. The test data for this uses the received signal strength indicator (RSSI) value of the Wi-Fi device collected from the mobile device. These data are solved through ontology because they includes noise and heterogeneous problems [ABSTRACT FROM AUTHOR]

Recently, due to the popularization of 3D depth cameras, new researches and opportunities have been made in research conducted on RGB images, but estimation of human hand pose is still classified as one of the difficult topics. In this paper, we propose a robust estimation method of human hand pose from various input 3D depth images using a learning algorithm. The proposed approach first generates a skeleton-based hand model and then aligns the generated hand model with three-dimensional point cloud data. Then, using a random forest-based learning algorithm, the hand pose is strongly estimated from the aligned hand model. Experimental results in this paper show that the proposed hierarchical approach makes robust and fast estimation of human hand posture from input depth images captured in various indoor and outdoor environments. [ABSTRACT FROM AUTHOR]

The butterfly species recognition technology based on machine learning using images has the effect of reducing a lot of time and cost of those involved in the related field to understand the diversity, number, and habitat distribution of butterfly species. In order to improve the accuracy and time efficiency of butterfly species classification, various features used as the inputs of machine learning models have been studied. Among them, branch length similarity(BLS) entropy or color intensity entropy methods using the concept of entropy showed higher accuracy and shorter learning time than other features such as Fourier transform or wavelet. This paper proposes a feature extraction algorithm using RGB color intensity entropy for butterfly color images. In addition, we develop butterfly recognition systems that combines the proposed feature extraction method with representative ensemble models and evaluate their performance. [ABSTRACT FROM AUTHOR]

In this paper, a method for nonlinear target recognition via machine learning is presented. The nonlinear radar environment used in this study was a frequency-modulated continuous-wave (FMCW) nonlinear radar with a transmit frequency band of 3.0∼3.2 GHz and received a frequency band of 6∼6.4 GHz corresponding to the second harmonics. Nonlinear radar measurements were performed using four types of electronic devices as nonlinear targets. Statistical parameters were extracted from the measured amplitude spectrum of the received harmonic responses for each target to successfully construct a classification algorithm. The extracted characteristic data were then used to construct and verify a support vector machine (SVM) classifier. The accuracy of the target classification by the trained SVM classifier was confirmed through verification data, and an accuracy of 85 % with 10-fold cross-validation was demonstrated. [ABSTRACT FROM AUTHOR]

Recently, applying artificial intelligence technologies in various fields of production has drawn an upsurge of research interest due to the increase for smart factory and artificial intelligence technologies. A great deal of effort is being made to introduce artificial intelligence algorithms into the defect detection task. Particularly, detection of defects on the surface of metal has a higher level of research interest compared to other materials (wood, plastics, fibers, etc.). In this paper, we compare and analyze the speed and performance of defect classification by combining machine learning techniques (Support Vector Machine, Softmax Regression, Decision Tree) with dimensionality reduction algorithms (Principal Component Analysis, AutoEncoders) and two convolutional neural networks (proposed method, ResNet). To validate and compare the performance and speed of the algorithms, we have adopted two datasets ((i) public dataset, (ii) actual dataset), and on the basis of the results, the most efficient algorithm is determined. [ABSTRACT FROM AUTHOR]

The purpose of this study is to set the rate of change between the market price of the next day and the previous day to be predicted as the predicted value, and the market price for each section is generated by dividing the stock price ranking of the next day to be predicted at regular intervals, which is different from the previous papers that predict the market price. We would like to propose a new time series data prediction method that predicts the market price change rate of the final next day through a model using the rate of change as the predicted value. The change in the performance of the model according to the degree of subdivision of the predicted value and the type of input data was analyzed. [ABSTRACT FROM AUTHOR]

User interactions are being developed in various forms, and in particular, interactions using human gestures are being actively studied. Among them, hand gesture recognition is used as a human interface in the field of realistic media based on the 3D Hand Model. The use of interfaces based on hand gesture recognition helps users access media media more easily and conveniently. User interaction using hand gesture recognition should be able to view images by applying fast and accurate hand gesture recognition technology without restrictions on the computer environment. This paper developed a fast and accurate user hand gesture recognition algorithm using the open source media pipe framework and machine learning's k-NN (K-Nearest Neighbor). In addition, in order to minimize the restriction of the computer environment, a stereoscopic image control system based on user hand gesture recognition was designed and implemented using a web service environment capable of Internet service and a docker container, a virtual environment. [ABSTRACT FROM AUTHOR]

In the future, various products are created in various fields using artificial intelligence. In this age, it is a very important problem to know the operation principle of artificial intelligence learning method and to use it correctly. This paper introduces artificial intelligence learning methods that have been known so far. Learning of artificial intelligence is based on the fixed point iteration method of mathematics. The GD(Gradient Descent) method, which adjusts the convergence speed based on the fixed point iteration method, the Momentum method to summate the amount of gradient, and finally, the Adam method that mixed these methods. This paper describes the advantages and disadvantages of each method. In particularly, the Adam method having adaptivity controls learning ability of machine learning. And we analyze how these methods affect digital signals. The changes in the learning process of digital signals are the basis of accurate application and accurate judgment in the future work and research using artificial intelligence. [ABSTRACT FROM AUTHOR]

This paper proposes reinforcement learning to solve the node-disjoint path problem which establishes multipath for reliable data transmission in wireless ad-hoc networks. The node-disjoint path problem is a problem of determining a plurality of paths so that the intermediate nodes do not overlap between the source and the destination. In this paper, we propose an optimization method considering transmission distance in a large-scale wireless ad-hoc network using Q-learning in reinforcement learning, one of machine learning. Especially, in order to solve the node-disjoint path problem in a large-scale wireless ad-hoc network, a large amount of computation is required, but the proposed reinforcement learning efficiently obtains appropriate results by learning the path. The performance of the proposed reinforcement learning is evaluated from the viewpoint of transmission distance to establish two node-disjoint paths. From the evaluation results, it showed better performance in the transmission distance compared with the conventional simulated annealing. [ABSTRACT FROM AUTHOR]

In this paper, we propose a classification method for radar signals depending on the type of threat by applying machine learning to parameter data of radar signals . Currently, the army uses a library of mapping relations between the parameters and the types of threat to recognize threat signals. This approach has certain limitations when classifying signals and recognizing new types of threat or types of threat that do not exist in the current libraries. In this paper, we propose an automatic radar signal classification method depending on the type of threat that uses only parameter data without a library. A convolutional neural network is used as the classifier and machine learning is applied to train the classifier. The proposed method does not use a library, and hence, can classify threat signals that are new or do not exist in the current library. [ABSTRACT FROM AUTHOR]

A malfunction or breakdown of a manufacturing facility leads to product defects and the suspension of production lines, resulting in huge financial losses for manufacturers. Due to the spread of smart factory services, a large amount of data is being collected in factories, and AI-based research is being conducted to predict and diagnose manufacturing facility breakdowns or manufacturing site efficiency. However, because of the characteristics of manufacturing data, such as a severe class imbalance about abnormalities and ambiguous label information that distinguishes abnormalities, developing classification or anomaly detection models is highly difficult. In this paper, we present an deep learning algorithm for anomaly detection of a manufacturing facility using reconstruction loss of CNN-based model and ananlyze its performance. The algorithm detects anomalies by relying solely on normal data from the facility's manufacturing data in the exclusion of abnormal data. [ABSTRACT FROM AUTHOR]

Commercialization of public technology is the transfer of government-led scientific and technological innovation and R&D results to the private sector, and is recognized as a key achievement driving economic growth. Therefore, in order to activate technology transfer, various machine learning methods are being studied to identify success factors or to match public technology with high commercialization potential and demanding companies. However, public technology commercialization data is in the form of a table and has a problem that machine learning performance is not high because it is in an imbalanced state with a large difference in success-failure ratio. In this paper, we present a method of utilizing CTGAN to resolve imbalances in public technology data in tabular form. In addition, to verify the effectiveness of the proposed method, a comparative experiment with SMOTE, a statistical approach, was performed using actual public technology commercialization data. In many experimental cases, it was confirmed that CTGAN reliably predicts public technology commercialization success cases. [ABSTRACT FROM AUTHOR]

In this paper, we propose a classification method of pulsed radar signal attributes by applying machine learning to a convolutional neural network and a long short-term memory(CNN-LSTM) network when the radio frequency and pulse repetition interval of the signal changes with a long period. CNN has a function to extract data features, and LSTM shows good classification performance when sequential data are highly correlated. However, LSTM has a problem that performance is degraded when the length of the input data is long. In the proposed method, CNN is applied to extract compressed features from long-period data to reduce the length of inputs to LSTM. In simulation results, the proposed method shows more than 95% correct classification rate when the data drop rate is 10% for 182,516 total radar signal attribute combinations. [ABSTRACT FROM AUTHOR]

Machine learning has a close relationship with cognitive psychology and brain science and is developing together. This paper analyzes the OASIS-3 dataset using machine learning techniques and proposes a model for predicting dementia. Dimensional reduction through PCA (Principal Component Analysis) is performed on the data quantifying the volume of each area among OASIS-3 data, and only important elements (features) are extracted and then various machine learning including gradient boosting and stacking Apply the models and compare the performance of each. Unlike previous studies, the proposed technique has a great differentiation because it uses not only the brain biometric data, but also basic information data such as the participant's gender and medical information data of the participant. In addition, it was shown that the proposed technique through various performance evaluations is a model that can better predict dementia by finding features that are more related to dementia among various numerical data. [ABSTRACT FROM AUTHOR]

The adaptation of smartfarm technology that converges ICT is increasing productivity and competitiveness in the agriculture. Technologies have been developed that enable environmental monitoring through various sensors and automatic control of the cultivation environment, and researches are underway to advance smartfarm technology using data generated from smartfarms. In this paper, an environmental control method to reduce the energy consumption of a smartfarm by using the environment and control data of the smartfarm is proposed. It was confirmed that energy consumption could be reduced compared to an independent environmental control method by creating an environmental prediction model using accumulated environmental data and selecting a control method to minimize energy consumption in a given situation by considering multiple environmental factors. In the future, research is needed to obtain higher energy efficiency through the advancement of the predictive model and the improvement of the complex control algorithms. [ABSTRACT FROM AUTHOR]

Character recognition is a technology required in various platforms, such as smart parking and text to speech, and many studies are being conducted to improve its performance through new attempts. However, with low-quality image used for character recognition, a difference in resolution of the training image and test image for character recognition occurs, resulting in poor accuracy. To solve this problem, this paper designed an end-to-end learning neural network that combines image super-resolution and character recognition so that the character recognition model performance is robust against various quality data, and implemented an alternative whole learning algorithm to learn the whole neural network. An alternative end-to-end learning and recognition performance test was conducted using the license plate image among various text images, and the effectiveness of the proposed algorithm was verified with the performance test. [ABSTRACT FROM AUTHOR]

In this paper, a machine learning-based automatic infographic recommendation system is proposed to improve the existing infographic production method. This system consists of a part that machine learning multiple infographic images and a part that automatically recommends infographics with artificial intelligence only by inputting basic data from the user. The recommended infographics are provided in the form of a library, and additional data can be input by drag & drop method. In addition, the infographic image is designed to be dynamically adjusted according to the size of the input data. As a result of analyzing the machine learning-based automatic infographic recommendation process, the matching success rate for layout and keyword was very high, and the matching success rate for type was rather low. In the future, a study to improve the matching success rate for the image type for each part of the infographic will be needed. [ABSTRACT FROM AUTHOR]

Nowadays, fake news from newspapers and social media is a serious issue in news credibility. Some of machine learning methods (such as LSTM, logistic regression, and Transformer) has been applied for fact checking. In this paper, we present Transformer-based fact checking model which improves computational efficiency. Locality Sensitive Hashing (LSH) is employed to efficiently compute attention value so that it can reduce the computation time. With LSH, model can group semantically similar words, and compute attention value within the group. The performance of proposed model is 75% for accuracy, 42.9% and 75% for Fl micro score and F1 macro score, respectively. [ABSTRACT FROM AUTHOR]

In this paper, we propose a method for predicting in advance whether pedestrians will enter the hazardous area after the current time using the pedestrian trajectory prediction method and an efficient simplification method of the trajectory prediction network. In addition, we propose a method to apply KD(Knowledge Distillation) to a small network for real-time operation in an embedded environment. Using the correlation between predicted future paths and hazard zones, we determined whether to enter or not, and applied efficient KD when learning small networks to minimize performance degradation. Experimentally, it was confirmed that the model applied with the simplification method proposed improved the speed by 37.49% compared to the existing model, but led to a slight decrease in accuracy. As a result of learning a small network with an initial accuracy of 91.43% using KD, It was confirmed that it has improved accuracy of 94.76%. [ABSTRACT FROM AUTHOR]

There have been recently an increasing number of people working out at home. However, many of them do not have face-to-face guidance from experts, so they cannot effectively correct their wrong pose. This may lead to strain and injury to those doing home training. To tackle this problem, this paper proposes a video data-based pose classification and correction system for home training. The proposed system classifies poses using the multi-layer perceptron and pose estimation model, and corrects poses based on joint angels estimated. A voting algorithm that considers the results of successive frames is applied to improve the performance of the pose classification model. Multi-layer perceptron model for post classification shows the highest accuracy with 0.9. In addition, it is shown that the proposed voting algorithm improves the accuracy to 0.93. [ABSTRACT FROM AUTHOR]

Although the IoT is showing explosive growth due to the development of technology and the spread of IoT devices and activation of services, serious security risks and financial damage are occurring due to the activities of various botnets. Therefore, it is important to accurately and quickly detect the activities of these botnets. As security in the IoT environment has characteristics that require operation with minimum processing performance and memory, in this paper, the minimum characteristics for detection are selected, and KNN (K-Nearest Neighbor), Naïve Bayes, Decision Tree, Random A comparative study was conducted on the performance of machine learning algorithms such as Forest to detect botnet activity. Experimental results using the Bot-IoT dataset showed that KNN can detect DDoS, DoS, and Reconnaissance attacks most effectively and efficiently among the applied machine learning algorithms. [ABSTRACT FROM AUTHOR]

Recently, the spread of smart TV based Android iOS Set Top box has become common. This paper propose a new approach to control the TV using gestures away from the era of controlling the TV using remote control. In this paper, the AdaBoost algorithm is applied to gesture recognition by using a mono camera. First, we use Camshift-based Body tracking and estimation algorithm based on Gaussian background removal for body coordinate extraction. Using global and local feature vectors, we recognized gestures with speed change. By tracking the time interval trajectories of hand and wrist, the AdaBoost algorithm with CART algorithm is used to train and classify gestures. The principal component feature vector with high classification success rate is searched using CART algorithm. As a result, 24 optimal feature vectors were found, which showed lower error rate (3.73%) and higher accuracy rate (95.17%) than the existing algorithm. [ABSTRACT FROM AUTHOR]

In the field of music, various AI composition methods using machine learning have recently been attempted. However, most of this research has been centered on Western music, and little research has been done on Korean traditional music. Therefore, in this paper, we will create a data set of Korean traditional music, create a melody using three algorithms based on the data set, and compare the results. Three models were selected based on the similarity between language and music, LSTM, Music Transformer and Self Attention. Using each of the three models, a melody generator was modeled and trained to generate melodies. As a result of user evaluation, the Self Attention method showed higher preference than the other methods. Data set is very important in AI composition. For this, a Korean traditional music data set was created, and AI composition was attempted with various algorithms, and this is expected to be helpful in future research on AI composition for Korean traditional music. [ABSTRACT FROM AUTHOR]

Caenorhabditis elegans whose DNA sequence was completely identified is a representative species used in various research fields such as gene functional analysis and animal behavioral research. In the mean time, many researches on the bio-monitoring system to determine whether water is contaminated or not by using the swimming activities of nematodes. In this paper, we show the possibility of using the swimming activities of C. elegans in the development of a machine learning based bio-monitoring system which identifies chemicals that cause water pollution. To characterize swimming activities of nematode, BLS entropy is computed for the nematode in a frame. And, BLS entropy profile, an assembly of entropies, are classified into several patterns using clustering algorithms. Finally these patterns are used to construct data sets. We recorded images of swimming behavior of nematodes in the arenas in which formaldehyde, benzene and toluene were added at a concentration of 0.1 ppm, respectively, and evaluate the performance of the developed HMM [ABSTRACT FROM AUTHOR]

This paper presents an implementation result and performance evaluation of pavilion management services that does not only provide real-time status of the pavilion in the forest but also prediction services through machine learning. The developed hardware prototype detects whether the pavilion is occupied using a motion detection sensor and then sends it to a cloud database along with location information, date and time, temperature, and humidity data. The real-time usage status of the collected data is provided to the user's mobile application. The performance evaluation confirms that the average response time from the hardware module to the applications was 1.9 seconds. The accuracy was 99%. In addition, we implemented a pavilion availability prediction service that applied a machine learning-based SVM (Support Vector Model) model to collected data and provided it through mobile and web applications. [ABSTRACT FROM AUTHOR]

In this paper, an algorithm for determining the coagulant input rate in the drug-injection tank during the process of the water purification plant was derived through big data analysis and prediction based on artificial intelligence. In addition, analysis of big data technology and AI algorithm application methods and existing academic and technical data were reviewed to analyze and review application cases in similar fields. Through this, the goal was to develop an algorithm for determining the coagulant input rate and to present the optimal input rate through autonomous driving simulator and pilot operation of the coagulant input process. Through this study, the coagulant injection rate, which is an output variable, is determined based on various input variables, and it is developed to simulate the relationship pattern between the input variable and the output variable and apply the learned pattern to the decision-making pattern of water plant operating workers. [ABSTRACT FROM AUTHOR]

Recent advance in low cost and light-weight drones has extended their application areas in both military and private sectors. Accordingly surveillance program against unfriendly drones has become an important issue. Drone detection and classification technique has long been emphasized in order to prevent attacks or accidents by commercial drones in urban areas. Most commercial drones have small sizes and low reflection and hence typical sensors that use acoustic, infrared, or radar signals exhibit limited performances. Recently, artificial intelligence algorithm has been actively exploited to enhance radar image identification performance. In this paper, we adopt machined learning algorithm for high resolution radar imaging in drone detection and classification applications. For this purpose, simulation is carried out against commercial drone models and compared with experimental data obtained through high resolution radar field test. [ABSTRACT FROM AUTHOR]

In this paper, we present and analyze a method of applying a machine learning to ultrasonic test signals for constant monitoring of the welding faults in a tubular steel tower. For the machine learning, feature selection based on genetic algorithm and fault signal classification using a support vector machine have been used. In the feature selection, the peak value, histogram lower bound, and normal negative log-likelihood from 30 features are selected. Those features clearly indicate the difference of signals according to the depth of faults. In addition, as a result of applying the selected features to the support vector machine, it has been possible to perfectly distinguish between the regions with and without faults. Hence, it is expected that the results of this study will be useful in the development of an early detection system for fault growth based on ultrasonic signals and in the energy transmission related industries in the future. [ABSTRACT FROM AUTHOR]

Information search in video streaming services such as YouTube is replacing traditional information search services. To find desired detailed information in such a video, users should repeatedly navigate several points in the video, resulting in a waste of time and network traffic. In this paper, we propose a method to assist users in searching for information in a video by using DBSCAN clustering and LSTM. Our LSTM model is trained with a dataset that consists of user search sequences and their final target points categorized by DBSCAN clustering algorithm. Then, our proposed method utilizes the trained model to suggest an expected category for the user's desired target point based on a partial search sequence that can be collected at the beginning of the search. Our experiment results show that the proposed method successfully finds user destination points with 98% accuracy and 7s of the time difference by average. [ABSTRACT FROM AUTHOR]

In this paper, a 2D indoor direction and location finding system using a gradient descent machine learning algorithm is presented. The proposed antenna consists of four folded dipole antennas and an SP4T PIN diode switch. The SP4T switch was controlled by four digital output pins of the Arduino microcontroller. The antenna was connected to the USRP N200 SDR and controlled by the LabView platform. LabView transfers the magnitude of the received RF power to MATLAB to calculate the incident angle and distance of an electromagnetic wave source. The measured accuracy of the proposed 2D indoor direction and location finding system was approximately 97 %. [ABSTRACT FROM AUTHOR]

Multi-path system is a system in which utilizes various networks simultaneously. It is expected that multi-path system can enhance communication speed, reliability, security of network. In this paper, we focus on path selection in multi-path system. To select optimal path, we propose deep reinforcement learning algorithm which is rewarded by the round-trip-time (RTT) of each networks. Unlike multi-armed bandit model, deep Q learning is applied to consider rapidly changing situations. Due to the delay of RTT data, we also suggest compensation algorithm of the delayed reward. Moreover, we implement testbed learning server to evaluate the performance of proposed algorithm. The learning server contains distributed database and tens or flow module to efficiently operate deep learning algorithm. By means of simulation, we showed that the proposed algorithm has better performance than lowest RTT about 20%. [ABSTRACT FROM AUTHOR]

High forecast accuracy is required as social issues on particulate matter increase. Therefore, many attempts are being made using machine learning to increase the accuracy of particulate matter prediction. However, due to problems with the distribution of imbalance in the concentration and various characteristics of particulate matter, the learning of prediction models is not well done. In this paper, to solve these problems, a binary classification model was proposed to predict the concentration of particulate matter needed for prediction by dividing it into two classes based on the value of 80μg/m³. Four classification algorithms were utilized for the binary classification of PM10. Classification algorithms used logistic regression, decision tree, SVM, and MLP. As a result of performance evaluation through confusion matrix, the MLP model showed the highest binary classification performance with 89.98% accuracy among the four models. [ABSTRACT FROM AUTHOR]

This paper presents a machine learning (ML) algorithm to detect orthogonal frequency-division multiplexing (OFDM) signals. Spectrum sensing is a key technology in cognitive radio communication, which enhances spectral efficiency. Recently, signals transmitted and received by wireless communication systems have been based on OFDM. These signals contain a pilot signal for channel calibration, which can be detected using a spectral correlation function (SCF). In this process, the FAM algorithm is applied for efficient SCF operation herein. The existence of OFDM signals is determined through a convolutional neural network-based ML algorithm using the SCF as input data. The learning data for ML use the SCF values for OFDM signals with signal-to-noise ratios (SNRs) of −20 to 0 dB. Consequently, on evaluation of the post-learning optimized neural network performance, signals were detected with a probability of 0.9 at the condition of 0.08 false alarm probability for reception signals with SNRs of −12 dB. [ABSTRACT FROM AUTHOR]