No abstract

This study proposes the prefixSpan based pattern mining using time sliding weight from streaming data. To discover sequential patterns, it applies a time sliding weight to create a structure of projected DB Tree. For the time sliding weight, a time window is applied to the sequential data to calculate the label and support of the window. When a projected DB Tree is designed, the time weight calculated for each pattern is inserted in a table. At this time, the tree is updated by deleting the node whose time weight is less than the reference value. For this reason, whenever data is updated, the tree is sorted again. The reordering process removes the pattern of less influence by applying time weights. Therefore, it is possible to construct a projected DB Tree that can extract influential patterns. The performance of the proposed method is evaluated in three aspects. Firstly, the conventional PrefixSpan algorithm is compared with the proposed PrefixSpan algorithm based on time sliding weight in terms of the pattern generation time according to size of data. Secondly, the fitness of the proposed algorithm is evaluated through cross-validation. Thirdly, GSP, SPADE, and prefixSpan algorithms are compared with the proposed algorithm through F-measure. As a result of the evaluation, the proposed algorithm improves accuracy 75% more than PrefixSpan algorithm and the sequential pattern algorithms of GSP and SPADE. Regarding the comparison of F-measure based on precision and recall, the proposed one improves its performance about 83%.

Korean people are exposed to stress due to the constant competitive structure caused by rapid industrialization. As a result, there is a need for ways that can effectively manage stress and help improve quality of life. Therefore, this study proposes an activity recommendation model using rank correlation for chronic stress management. Using Spearman’s rank correlation coefficient, the proposed model finds the correlations between users’ Positive Activity for Stress Management (PASM), Negative Activity for Stress Management (NASM), and Perceived Stress Scale (PSS). Spearman’s rank correlation coefficient improves the accuracy of recommendations by putting a basic rank value in a missing value to solve the sparsity problem and cold-start problem. For the performance evaluation of the proposed model, F-measure is applied using the average precision and recall after five times of recommendations for 20 users. As a result, the proposed method has better performance than other models, since it recommends activities with the use of the correlation between PASM and NASM. The proposed activity recommendation model for stress management makes it possible to manage user’s stress effectively by lowering the user’s PSS using correlation.

High-quality, large-capacity data are essential for training a deep learning vision model. However, to construct crop image data, absolute growth time is required for crop growth. In addition, it is characterized by unbalanced data, with fewer abnormal data than normal data. Therefore, building highquality, large-scale datasets is challenging. Many studies have used data augmentation of plant images to solve this problem. However, plants require data augmentation that does not compromise their color, texture, or shape. This study proposes the use of salient target augmentation (STAug) as a data augmentation technique to protect the colors and shapes of plant images. The proposed method pastes one image's salient target into a different image to mix the two images. It uses a salient object detection model to generate a salient object mask of the plant. Using the generated mask, a salient target was identified and cropped in a plant image, and the cropped image data were pasted to different background data for augmentation. Concat mask, a combination of each image's salient object mask, was designed to create the label of the generated image. It is possible to create a rigid classification model by augmenting the data without damaging the plant features. To verify the performance of the proposed STAug, we compared its performance with that of other data-augmentation policies. When STAug and other augmentation techniques were applied in combination, an accuracy of 0.9733 was achieved. We demonstrated a better classification performance than when it was not applied.