Search

According to the World Health Organization (WHO), stroke is the second leading cause of death globally, accounting for approximately 11% of total deaths. The dataset consists of stroke data obtained from Kaggle, with each row providing relevant information about a patient. It can be used to predict whether a patient is likely to have a stroke based on various input parameters such as gender, age, various diseases, smoking status etc. The relationships between data columns are examined using correlation coefficients and multicollinearity. For classification, various machine learning methods, DNN, and Ensemble techniques were used. In the DNN, five layers were stacked and applied. The Ensemble algorithm combines RNN and CNN. The Adam optimizer was used to update gradients, and Binary Cross Entropy was used as the Loss Function. For evaluation methods, Accuracy, Recall, ROC-AUC were used to assess the classification techniques. [ABSTRACT FROM AUTHOR]

As a subcategory of machine learning, deep learning is one of the artificial neural network methods. Thanks to generation of new kinds of data and significant improvement in the data processing speed, deep learning technologies have attracted considerable attention. Because data in various domains need to be processed and analyzed, the need for deep learning education targeting is also increasing. However, it is difficult to provide them with systematic deep learning education in a short period of time because there have been many algorithms and they are implemented in a programming language. This paper suggests an extendable pratice method which makes use of core sourse code templates that cover from simple neural network (NN) consisting of fully connected layers to convolutional neural network (CNN). Based on real cases of class operation, the lecture content per class session and the source code templates are provided in this paper. [ABSTRACT FROM AUTHOR]

In this study, we propose a methodology for training a deep learning model using historical time series data to predict current values. Using 11,942 past meteorological observations collected over a specific period from the Osan weather station in Gyeonggi-do, we predicted the atmospheric radio refractivity at the current time. As a result of predicting the atmospheric radio refractivity at 00Z on August 4, 2023, and August 5, 2023, at the Osan weather station in Gyeonggi-do, the average MAPE(mean absolute percentage error) and ARIMA(autoregressive integrated moving average) of the proposed model were 2.75 % and 7.61 %, respectively. [ABSTRACT FROM AUTHOR]

We investigated personal face identification systems using deep learning networks and we proposed a deep neural network architecture for improving false positive rate. Most of conventional face identification systems have the same number of output nodes as the number of faces that can be identified, and each node is trained to identify one registered face. In this paper, we added an extra node to identify unregistered faces for improving false positive rate(FPR) and accuracy by reducing the propagation of matching probabilities toward other output nodes identifying the registered faces, when unregistered faces are input. The proposed model has been trained with the VGGFace2 dataset. The performance of the proposed model was analyzed in terms of accuracy, precision, FPR, and false negative rate (FNR), and was compared to that of the existing model. According to the performance analysis results, when FNR is 5%, the FPR of the proposed model is improved by about 83% compared to the existing model. [ABSTRACT FROM AUTHOR]

In this study, we propose a model for predicting atmospheric refractivity using meteorological data and deep learning. The purpose of this study is to compare the prediction accuracy of traditional interpolation methods and the proposed model, verify whether the deep learning model trained on meteorological data can provide values closer to the true values, and thus demonstrate the potential for utilizing deep learning in predicting atmospheric refractivity. [ABSTRACT FROM AUTHOR]

This study proposes a bent wire monopole antenna to achieve impedance matching in the frequency range of 950−1,050 MHz and a directive radiation pattern in the target direction (φ=180°, 40°<θ<50°). A machine learning technique based on a deep neural network is applied to obtain an optimal antenna design that is bent at three points to adjust the impedance and radiation pattern. The machine learning model is trained using a dataset that includes the antenna geometry and cost, evaluated on how well the antenna performance satisfies the target bandwidth and radiation pattern. After verifying that the machine learning model is trained using the figure of merit, the machine learning model is validated by comparing the cost estimated by the model with that estimated by a commercial electro- magnetic simulator. Next, we obtain the optimal antenna design from the result of a grid search and fabricate it to verify its performance. The results show that the manufactured antenna has a matching bandwidth from 983.5 to 1,037.5 MHz and a radiation gain of approximately 4 dBi in the target direction. These results are in good agreement with the simulated performance of the antenna. Thus, we conclude that the antenna design obtained using the machine learning technique is valid for designing the proposed antenna. [ABSTRACT FROM AUTHOR]

With the recent development of deep learning technology, autonomous things technology is attracting attention, and DNNs are widely used in embedded systems such as drones and autonomous vehicles. Embedded systems that can perform large-scale operations and process multiple DNNs for high recognition accuracy without relying on the cloud are being released. DNNs with various levels of priority exist within these systems. DNNs related to the safety-critical applications of autonomous vehicles have the highest priority, and they must be handled first. In this paper, we propose a priority-based scheduling framework for DNNs when multiple DNNs are executed simultaneously. Even if a low-priority DNN is being executed first, a high-priority DNN can preempt it, guaranteeing the fast response characteristics of safety-critical applications of autonomous vehicles. As a result of checking through extensive experiments, the performance improved by up to 76.6% in the actual commercial board. [ABSTRACT FROM AUTHOR]

DNN is expanding its use in embedded systems such as robots and autonomous vehicles. For high recognition accuracy, computational complexity is greatly increased, and multiple DNNs are running aperiodically. Therefore, the ability processing multiple DNNs in embedded environments is a crucial issue. Accordingly, multicore based platforms are being released. However, most DNN models are operated in a batch process, and when multiple DNNs are operated in multicore together, the execution time deviation between each DNN may be large and the end-to-end execution time of the whole DNNs could be long depending on how they are allocated to the cores. In this paper, we solve these problems by providing a framework that decompose each DNN into individual layers and then distribute to multicores through a global queue. As a result of the experiment, the total DNN execution time was reduced by 31%, and when operating multiple identical DNNs, the deviation in execution time was reduced by up to 95.1%. [ABSTRACT FROM AUTHOR]

The human impact of particulate matter are revealed and demand for improved forecast accuracy is increasing. Recently, efforts is made to improve the accuracy of PM10 predictions by using machine learning, but prediction performance is decreasing due to the particulate matter data with a large rate of low concentration occurrence. In this paper, separation prediction model by concentration is proposed to improve the accuracy of PM10 particulate matter forecast. The low and high concentration prediction model was designed using the weather and air pollution factors in Cheonan, and the performance comparison with the prediction models was performed. As a result of experiments with RMSE, MAPE, correlation coefficient, and AQI accuracy, it was confirmed that the predictive performance was improved, and that 20.62% of the AQI high-concentration prediction performance was improved. [ABSTRACT FROM AUTHOR]

The weather has a lot of influence on the cultivation of crops. Weather information on agricultural crop cultivation areas is indispensable for efficient cultivation and management of agricultural crops. Despite the high demand for agricultural weather, research on this is in short supply. In this research, we deal with the production method of agricultural weather in Jeollanam-do, which is the main production area of onions through GloSea5 and deep learning. A deep neural network model using the sliding window method was used and utilized to train daily weather prediction for predicting the agricultural weather. RMSE and MAE are used for evaluating the accuracy of the model. The accuracy improves as the learning period increases, so we compare the prediction performance according to the learning period and the prediction period. As a result of the analysis, although the learning period and the prediction period are similar, there was a limit to reflect the trend according to the seasonal change. a modified deep layer neural network model was presented, that applying the difference between the predicted value and the observed value to the next day predicted value. [ABSTRACT FROM AUTHOR]

The effectiveness of deep neural networks (DNNs) for detection and classification of micro-Doppler signals generated by human walking and background noise sources is investigated. Previous research included a complex process for extracting meaningful features that directly affect classifier performance, and this feature extraction is based on experiences and statistical analysis. However, because a DNN gradually reconstructs and generates features through a process of passing layers in a network, the preprocess for feature extraction is not required. Therefore, binary classifiers and multiclass classifiers were designed and analyzed in which multilayer perceptrons (MLPs) and DNNs were applied, and the effectiveness of DNNs for recognizing micro-Doppler signals was demonstrated. Experimental results showed that, in the case of MLPs, the classification accuracies of the binary classifier and the multiclass classifier were 90.3% and 86.1%, respectively, for the test dataset. In the case of DNNs, the classification accuracies of the binary classifier and the multiclass classifier were 97.3% and 96.1%, respectively, for the test dataset. [ABSTRACT FROM AUTHOR]