• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.79-104
• /
• 2020

Recently, as deep learning has attracted attention, the use of deep learning is being considered as a method for solving problems in various fields. In particular, deep learning is known to have excellent performance when applied to applying unstructured data such as text, sound and images, and many studies have proven its effectiveness. Owing to the remarkable development of text and image deep learning technology, interests in image captioning technology and its application is rapidly increasing. Image captioning is a technique that automatically generates relevant captions for a given image by handling both image comprehension and text generation simultaneously. In spite of the high entry barrier of image captioning that analysts should be able to process both image and text data, image captioning has established itself as one of the key fields in the A.I. research owing to its various applicability. In addition, many researches have been conducted to improve the performance of image captioning in various aspects. Recent researches attempt to create advanced captions that can not only describe an image accurately, but also convey the information contained in the image more sophisticatedly. Despite many recent efforts to improve the performance of image captioning, it is difficult to find any researches to interpret images from the perspective of domain experts in each field not from the perspective of the general public. Even for the same image, the part of interests may differ according to the professional field of the person who has encountered the image. Moreover, the way of interpreting and expressing the image also differs according to the level of expertise. The public tends to recognize the image from a holistic and general perspective, that is, from the perspective of identifying the image's constituent objects and their relationships. On the contrary, the domain experts tend to recognize the image by focusing on some specific elements necessary to interpret the given image based on their expertise. It implies that meaningful parts of an image are mutually different depending on viewers' perspective even for the same image. So, image captioning needs to implement this phenomenon. Therefore, in this study, we propose a method to generate captions specialized in each domain for the image by utilizing the expertise of experts in the corresponding domain. Specifically, after performing pre-training on a large amount of general data, the expertise in the field is transplanted through transfer-learning with a small amount of expertise data. However, simple adaption of transfer learning using expertise data may invoke another type of problems. Simultaneous learning with captions of various characteristics may invoke so-called 'inter-observation interference' problem, which make it difficult to perform pure learning of each characteristic point of view. For learning with vast amount of data, most of this interference is self-purified and has little impact on learning results. On the contrary, in the case of fine-tuning where learning is performed on a small amount of data, the impact of such interference on learning can be relatively large. To solve this problem, therefore, we propose a novel 'Character-Independent Transfer-learning' that performs transfer learning independently for each character. In order to confirm the feasibility of the proposed methodology, we performed experiments utilizing the results of pre-training on MSCOCO dataset which is comprised of 120,000 images and about 600,000 general captions. Additionally, according to the advice of an art therapist, about 300 pairs of 'image / expertise captions' were created, and the data was used for the experiments of expertise transplantation. As a result of the experiment, it was confirmed that the caption generated according to the proposed methodology generates captions from the perspective of implanted expertise whereas the caption generated through learning on general data contains a number of contents irrelevant to expertise interpretation. In this paper, we propose a novel approach of specialized image interpretation. To achieve this goal, we present a method to use transfer learning and generate captions specialized in the specific domain. In the future, by applying the proposed methodology to expertise transplant in various fields, we expected that many researches will be actively conducted to solve the problem of lack of expertise data and to improve performance of image captioning.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.127-148
• /
• 2020

The data center is a physical environment facility for accommodating computer systems and related components, and is an essential foundation technology for next-generation core industries such as big data, smart factories, wearables, and smart homes. In particular, with the growth of cloud computing, the proportional expansion of the data center infrastructure is inevitable. Monitoring the health of these data center facilities is a way to maintain and manage the system and prevent failure. If a failure occurs in some elements of the facility, it may affect not only the relevant equipment but also other connected equipment, and may cause enormous damage. In particular, IT facilities are irregular due to interdependence and it is difficult to know the cause. In the previous study predicting failure in data center, failure was predicted by looking at a single server as a single state without assuming that the devices were mixed. Therefore, in this study, data center failures were classified into failures occurring inside the server (Outage A) and failures occurring outside the server (Outage B), and focused on analyzing complex failures occurring within the server. Server external failures include power, cooling, user errors, etc. Since such failures can be prevented in the early stages of data center facility construction, various solutions are being developed. On the other hand, the cause of the failure occurring in the server is difficult to determine, and adequate prevention has not yet been achieved. In particular, this is the reason why server failures do not occur singularly, cause other server failures, or receive something that causes failures from other servers. In other words, while the existing studies assumed that it was a single server that did not affect the servers and analyzed the failure, in this study, the failure occurred on the assumption that it had an effect between servers. In order to define the complex failure situation in the data center, failure history data for each equipment existing in the data center was used. There are four major failures considered in this study: Network Node Down, Server Down, Windows Activation Services Down, and Database Management System Service Down. The failures that occur for each device are sorted in chronological order, and when a failure occurs in a specific equipment, if a failure occurs in a specific equipment within 5 minutes from the time of occurrence, it is defined that the failure occurs simultaneously. After configuring the sequence for the devices that have failed at the same time, 5 devices that frequently occur simultaneously within the configured sequence were selected, and the case where the selected devices failed at the same time was confirmed through visualization. Since the server resource information collected for failure analysis is in units of time series and has flow, we used Long Short-term Memory (LSTM), a deep learning algorithm that can predict the next state through the previous state. In addition, unlike a single server, the Hierarchical Attention Network deep learning model structure was used in consideration of the fact that the level of multiple failures for each server is different. This algorithm is a method of increasing the prediction accuracy by giving weight to the server as the impact on the failure increases. The study began with defining the type of failure and selecting the analysis target. In the first experiment, the same collected data was assumed as a single server state and a multiple server state, and compared and analyzed. The second experiment improved the prediction accuracy in the case of a complex server by optimizing each server threshold. In the first experiment, which assumed each of a single server and multiple servers, in the case of a single server, it was predicted that three of the five servers did not have a failure even though the actual failure occurred. However, assuming multiple servers, all five servers were predicted to have failed. As a result of the experiment, the hypothesis that there is an effect between servers is proven. As a result of this study, it was confirmed that the prediction performance was superior when the multiple servers were assumed than when the single server was assumed. In particular, applying the Hierarchical Attention Network algorithm, assuming that the effects of each server will be different, played a role in improving the analysis effect. In addition, by applying a different threshold for each server, the prediction accuracy could be improved. This study showed that failures that are difficult to determine the cause can be predicted through historical data, and a model that can predict failures occurring in servers in data centers is presented. It is expected that the occurrence of disability can be prevented in advance using the results of this study.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.10
• /
• pp.1-43
• /
• 1971

Experiment I: The objective of this study was to know variation in some selected agronomic characters of sweet sorghum when planted in several growing seasons. The 17 different sweet sorghum varieties having various maturities, and plant, syrup and sugar types were used in this study which had been carried out for the period of two years from 1968 to 1969 at Industrial Crops Division of Crop Experiment Station in Suwon. These varieties were planted at an interval of 20 days from April 5 to August 25 both in 1968 and 1969. The experimental results could be summarized as follows: 1. As planting was made early, the number of days from sowing to germination was getting prolonged while germination took place early when planted at the later date of which air temperature was relatively higher. However, such a tendency was not observed beyond the planting on August 25. In general, a significant negative correlation was found between the number of days from sowing to germination and the average daily temperature but a positive correlation was found between the former and the total accumulated average temperature during the growth period. 2. The period from sowing to heading was generally shortened as planting was getting delayed. The average varietal difference in number of days from sowing to heading was as much as 30.2 days. All the varieties were grouped into early-, medium and late-maturing groups based upon a difference of 10 days in heading. The average number of days from sowing to heading was 78.5$\pm$4.5 days in the early-maturing varieties, 88.5$\pm$4.5 days in the medium varieties and 98.5$\pm$4.5 days in the late-maturing varieties, respectively. The early-maturing varieties had the shortest period to heading when planted from July 15 to August 5, the medium varieties did when planted before July 15 and the late-maturing varieties did when planted before June 5. 3. The relationship between the sowing date (x) and number of days from sowing to heading could be expressed in an equation of y=a+bx. A highly positive correlation was found between the coefficient of the equation(shortening rate in heading time) and the average number of days from sowing to heading. 4. The number of days from sowing to heading was shortened as the daily average temperature during the growth period was getting higher. Early-maturing varieties had the shortest period to heading at a temperature of 24.2$^{\circ}C$, medium varieties at 23.8$^{\circ}C$ and late-maturing varieties at 22.9$^{\circ}C$, respectively. In other words, the number of days from sowing to heading was shortened rapidly in case that the average temperature for 30 days before heading was 22$^{\circ}C$ to $25^{\circ}C$. It prolonged relatively when the temperature was lower than 21$^{\circ}C$. 5. There was a little difference in plant height among varieties. In case of early planting, no noticeable difference in the height was observed. The plant height shortened generally as planting season was delayed. Elongation of plant height was remarkably accelerated as planting was delayed. This tendency was more pronounced in case of early-maturing varieties rather than late-maturing varieties. As a result, the difference in plant height between the maximum and the minimum was greater in late-maturing varieties than in early-maturing varieties. 6. Diameter of the stalk was getting thicker as planted earlier in late-maturing varieties. On the other hand, medium or early-maturing varieties had he thickest diameter when they were planted on April 25. 7. In general, a higher stalk yield was obtained when planted from April 25 to May 15. However, the planting time for the maximum stalk yield varied from one variety to another depending upon maturity of variety. Ear]y-maturing varieties produced the maximum yield when planted about April 25, medium varieties from April 25 to May 15 and late-maturing varieties did when planted from April 5 to May 15 respectively. The yield decreased linearly when they were planted later than the above dates. 8. A varietal difference in Brix % was also observed. The Brix % decreased linearly when the varieties were planted later than May 15. Therefore, a highly negative relationship between planting date(x) and Brix %(y) was detected. 9. The Brix % during 40 to 45 days after leading was the highest at the 1st to the 3rd internodes from the top while it decreased gradually from the 4th internode. It increased again somewhat at the 2nd internode from the ground level. However, it showed a reverse relationship between the Brix % and position of internode before heading. 10. Sugar content in stalk decreased gradually as planting was getting delayed though one variety differed from another. It seemed that sweet sorghum which planted later than June had no value as a sugar crop at all. 11. The Brix % and sugar content in stalk increased from heading and reached the maximum 40 to 45 days after heading. The percentage of purity showed the same tendency as the mentioned characters. Accordingly, a highly positive correlation was observed between. percentage of purity and Brix % or sugar content in stalk. 12. The highest refinable sugar yield was obtained from the planting on April 25 in late-maturing varieties and from that on May 15 in early-maturing varieties. The yield rapidly decreased when planted later than those dates. Such a negative correlation between planting date(x) and refinable sugar yield(y) was highly significant at 1% level. 13. Negative correlations or linear regressions between delayed planting and the number of days from sowing to germination. accumulated temperature during germination period, number of days to heading, accumulated temperature to heading, plant height, stem diameter, stalk weight, Brix %. sugar content, refinable sugar yield or Purity % were obtained. On the other hand, highly positive correlations between the number of days from sowing to heading(x) and Brix %, sugar content, purity %, refinable sugar yield, plant height or stalk yield, between Brix %(x) and purity %, refinable sugar yield or stalk yield, between sugar content(x) and purity% or refinable sugar yield(y), between purity %(x) and refinable sugar yield and between daylength at heading(x) and Brix %. number of days from sowing to heading, sugar content, purity % or refinable sugar yield (y), were found, respectively. Experiment II: The 11 varieties were selected out of the varieties used in Experiment I from ecological and genetic viewpoints. Complete diallel cross were made among them and the heading date, stalk length, stalk yield, Brix %, syrup yield, combining ability and genetic behavior of F$_1$ plants and their parental varieties were investigated. The results could be summarized as follows: 1. In general, number of days to heading showed a partial dominance over earliness or late maturity or had a mid-value, though there were some specific combinations showing a complete dominance or transgressive segregation in maturity. Some combinations showed relatively high general or specific combining abilities in maturity. Therefore, a 50 to 50 segregation ratio in heading date could be estimated in this study and it might be positive to have a selection in early generation since heritability of the character was relatively high. 2. A vigorous hybrid vigor was observed in stalk length. A complete or partial dominant effect of long stalk was obtained. The general combining ability and specific combining ability of stalk length were generally high. Long and short stalks segregated in a ratio of 50:50 and its heritability was relatively low. 3. Except for several specific combinations, high stalk yield seemed to be partial dominant over the low yield. Some varieties demonstrated relatively high general as well as specific combining abilities. It was assumed that several recessive genes were involved in expression of this character. The interaction among regulating recessive genes was also obtained. Accordingly, the heritability of stalk yield seemed to be rather low. 4. The Brix % of hybrid plants located around mid-parental value though some of them showed much higher or lower percentage. It could be explained by the fact that such behavior might be due to partial dominance of Brix %. The varieties with, relatively higher Brix % were high both in general. and specific combining abilities. Therefore, it could be recommended to use the varieties having higher sugar content in order to develop higher-sugar varieties. 5. The syrup yield seemed to be transgressively segregated or completely dominant over low yield. Hybrid vigor of syrup yield was relatively high. No-consistent relationship between general combining ability and specific combining ability was observed. However, some cases demonstrated that the varieties with relatively higher general combining ability had relatively lower specific combining ability. It was assumed that the frequencies of dominant and recessive alleles were almost same.