• Journal of Korea Water Resources Association
• /
• v.57 no.1
• /
• pp.35-44
• /
• 2024

Research in dam inflow prediction has actively explored the utilization of data-driven machine learning and deep learning (ML&DL) tools across diverse domains. Enhancing not just the inherent model performance but also accounting for model characteristics and preprocessing data are crucial elements for precise dam inflow prediction. Particularly, existing rainfall data, derived from snowfall amounts through heating facilities, introduces distortions in the correlation between snow accumulation and rainfall, especially in dam basins influenced by snow accumulation, such as Soyang Dam. This study focuses on the preprocessing of rainfall data essential for the application of ML&DL models in predicting dam inflow in basins affected by snow accumulation. This is vital to address phenomena like reduced outflow during winter due to low snowfall and increased outflow during spring despite minimal or no rain, both of which are physical occurrences. Three machine learning models (SVM, RF, LGBM) and two deep learning models (LSTM, TCN) were built by combining rainfall and inflow series. With optimal hyperparameter tuning, the appropriate model was selected, resulting in a high level of predictive performance with NSE ranging from 0.842 to 0.894. Moreover, to generate rainfall correction data considering snow accumulation, a simulated snow accumulation algorithm was developed. Applying this correction to machine learning and deep learning models yielded NSE values ranging from 0.841 to 0.896, indicating a similarly high level of predictive performance compared to the pre-snow accumulation application. Notably, during the snow accumulation period, adjusting rainfall during the training phase was observed to lead to a more accurate simulation of observed inflow when predicted. This underscores the importance of thoughtful data preprocessing, taking into account physical factors such as snowfall and snowmelt, in constructing data models.

• Journal of Wetlands Research
• /
• v.26 no.2
• /
• pp.147-159
• /
• 2024

Lack of streamflow observations makes model calibration difficult and limits model performance improvement. Satellite-based remote sensing products offer a new alternative as they can be actively utilized to obtain hydrological data. Recently, several studies have shown that artificial intelligence-based solutions are more appropriate than traditional conceptual and physical models. In this study, a data-driven approach combining various recurrent neural networks and decision tree-based algorithms is proposed, and the utilization of satellite remote sensing information for AI training is investigated. The satellite imagery used in this study is from MODIS and SMAP. The proposed approach is validated using publicly available data from 25 watersheds. Inspired by the traditional regionalization approach, a strategy is adopted to learn one data-driven model by integrating data from all basins, and the potential of the proposed approach is evaluated by using a leave-one-out cross-validation regionalization setting to predict streamflow from different basins with one model. The GRU + Light GBM model was found to be a suitable model combination for target basins and showed good streamflow prediction performance in ungauged basins (The average model efficiency coefficient for predicting daily streamflow in 25 ungauged basins is 0.7187) except for the period when streamflow is very small. The influence of satellite remote sensing information was found to be up to 10%, with the additional application of satellite information having a greater impact on streamflow prediction during low or dry seasons than during wet or normal seasons.

• The Journal of Korean Society for Radiation Therapy
• /
• v.26 no.2
• /
• pp.313-320
• /
• 2014

Purpose : The purpose of the study is to evaluate the efficiency of Coaxial MLC VMAT plan (Using $273^{\circ}$ and $350^{\circ}$ collimator angle) That the leaf motion direction aligned with axis of OAR (Organ at risk, It means spinal cord or cauda equine in this study.) compare to Universal MLC VMAT plan (using $30^{\circ}$ and $330^{\circ}$ collimator angle) for spine SBRT. Materials and Methods : The 10 cases of spine SBRT that treated with VMAT planned by Coaxial MLC and Varian TBX were enrolled. Those cases were planned by Eclipse (Ver. 10.0.42, Varian, USA), PRO3 (Progressive Resolution Optimizer 10.0.28) and AAA (Anisotropic Analytic Algorithm Ver. 10.0.28) with coplanar $360^{\circ}$ arcs and 10MV FFF (Flattening filter free). Each arc has $273^{\circ}$ and $350^{\circ}$ collimator angle, respectively. The Universal MLC VMAT plans are based on existing treatment plans. Those plans have the same parameters of existing treatment plans but collimator angle. To minimize the dose difference that shows up randomly on optimizing, all plans were optimized and calculated twice respectively. The calculation grid is 0.2 cm and all plans were normalized to the target V100%=90%. The indexes of evaluation are V10Gy, D0.03cc, Dmean of OAR (Organ at risk, It means spinal cord or cauda equine in this study.), H.I (Homogeneity index) of the target and total MU. All Coaxial VMAT plans were verified by gamma test with Mapcheck2 (Sun Nuclear Co., USA), Mapphan (Sun Nuclear Co., USA) and SNC patient (Sun Nuclear Co., USA Ver 6.1.2.18513). Results : The difference between the coaxial and the universal VMAT plans are follow. The coaxial VMAT plan is better in the V10Gy of OAR, Up to 4.1%, at least 0.4%, the average difference was 1.9% and In the D0.03cc of OAR, Up to 83.6 cGy, at least 2.2 cGy, the average difference was 33.3 cGy. In Dmean, Up to 34.8 cGy, at least -13.0 cGy, the average difference was 9.6 cGy that say the coaxial VMAT plans are better except few cases. H.I difference Up to 0.04, at least 0.01, the average difference was 0.02 and the difference of average total MU is 74.1 MU. The coaxial MLC VMAT plan is average 74.1 MU lesser then another. All IMRT verification gamma test results for the coaxial MLC VMAT plan passed over 90.0% at 1mm / 2%. Conclusion : Coaxial MLC VMAT treatment plan appeared to be favorable in most cases than the Universal MLC VMAT treatment planning. It is efficient in lowering the dose of the OAR V10Gy especially. As a result, the Coaxial MLC VMAT plan could be better than the Universal MLC VMAT plan in same MU.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.1-22
• /
• 2017

Analysis of future business or investment opportunities, such as business feasibility analysis and company or technology valuation, necessitate objective estimation on the relevant market and expected sales. While there are various ways to classify the estimation methods of these new sales or market size, they can be broadly divided into top-down and bottom-up approaches by benchmark references. Both methods, however, require a lot of resources and time. Therefore, we propose a data-based intelligent demand forecasting system to support evaluation of new business. This study focuses on analogical forecasting, one of the traditional quantitative forecasting methods, to develop sales forecasting intelligence systems for new businesses. Instead of simply estimating sales for a few years, we hereby propose a method of estimating the sales of new businesses by using the initial sales and the sales growth rate of similar companies. To demonstrate the appropriateness of this method, it is examined whether the sales performance of recently established companies in the same industry category in Korea can be utilized as a reference variable for the analogical forecasting. In this study, we examined whether the phenomenon of "mean reversion" was observed in the sales of start-up companies in order to identify errors in estimating sales of new businesses based on industry sales growth rate and whether the differences in business environment resulting from the different timing of business launch affects growth rate. We also conducted analyses of variance (ANOVA) and latent growth model (LGM) to identify differences in sales growth rates by industry category. Based on the results, we proposed industry-specific range and linear forecasting models. This study analyzed the sales of only 150,000 start-up companies in Korea in the last 10 years, and identified that the average growth rate of start-ups in Korea is higher than the industry average in the first few years, but it shortly shows the phenomenon of mean-reversion. In addition, although the start-up founding juncture affects the sales growth rate, it is not high significantly and the sales growth rate can be different according to the industry classification. Utilizing both this phenomenon and the performance of start-up companies in relevant industries, we have proposed two models of new business sales based on the sales growth rate. The method proposed in this study makes it possible to objectively and quickly estimate the sales of new business by industry, and it is expected to provide reference information to judge whether sales estimated by other methods (top-down/bottom-up approach) pass the bounds from ordinary cases in relevant industry. In particular, the results of this study can be practically used as useful reference information for business feasibility analysis or technical valuation for entering new business. When using the existing top-down method, it can be used to set the range of market size or market share. As well, when using the bottom-up method, the estimation period may be set in accordance of the mean reverting period information for the growth rate. The two models proposed in this study will enable rapid and objective sales estimation of new businesses, and are expected to improve the efficiency of business feasibility analysis and technology valuation process by developing intelligent information system. In academic perspectives, it is a very important discovery that the phenomenon of 'mean reversion' is found among start-up companies out of general small-and-medium enterprises (SMEs) as well as stable companies such as listed companies. In particular, there exists the significance of this study in that over the large-scale data the mean reverting phenomenon of the start-up firms' sales growth rate is different from that of the listed companies, and that there is a difference in each industry. If a linear model, which is useful for estimating the sales of a specific company, is highly likely to be utilized in practical aspects, it can be explained that the range model, which can be used for the estimation method of the sales of the unspecified firms, is highly likely to be used in political aspects. It implies that when analyzing the business activities and performance of a specific industry group or enterprise group there is political usability in that the range model enables to provide references and compare them by data based start-up sales forecasting system.

• The Journal of Korean Society for Radiation Therapy
• /
• v.27 no.2
• /
• pp.145-156
• /
• 2015

Purpose : The purpose of this study is to evaluate the efficiency of Split VMAT planning(Contouring rectum divided into an upper and a lower for reduce rectum dose) compare to Conventional VMAT planning(Contouring whole rectum) for prostate cancer radiotherapy involving pelvic lymph nodes. Materials and Methods : A total of 9 cases were enrolled. Each case received radiotherapy with Split VMAT planning to the prostate involving pelvic lymph nodes. Treatment was delivered using TrueBeam STX(Varian Medical Systems, USA) and planned on Eclipse(Ver. 10.0.42, Varian, USA), PRO3(Progressive Resolution Optimizer 10.0.28), AAA(Anisotropic Analytic Algorithm Ver. 10.0.28). Lower rectum contour was defined as starting 1cm superior and ending 1cm inferior to the prostate PTV, upper rectum is a part, except lower rectum from the whole rectum. Split VMAT plan parameters consisted of 10MV coplanar $360^{\circ}$ arcs. Each arc had $30^{\circ}$ and $30^{\circ}$ collimator angle, respectively. An SIB(Simultaneous Integrated Boost) treatment prescription was employed delivering 50.4Gy to pelvic lymph nodes and 63~70Gy to the prostate in 28 fractions. $D_{mean}$ of whole rectum on Split VMAT plan was applied for DVC(Dose Volume Constraint) of the whole rectum for Conventional VMAT plan. In addition, all parameters were set to be the same of existing treatment plans. To minimize the dose difference that shows up randomly on optimizing, all plans were optimized and calculated twice respectively using a 0.2cm grid. All plans were normalized to the prostate $PTV_{100%}$ = 90% or 95%. A comparison of $D_{mean}$ of whole rectum, upperr ectum, lower rectum, and bladder, $V_{50%}$ of upper rectum, total MU and H.I.(Homogeneity Index) and C.I.(Conformity Index) of the PTV was used for technique evaluation. All Split VMAT plans were verified by gamma test with portal dosimetry using EPID. Results : Using DVH analysis, a difference between the Conventional and the Split VMAT plans was demonstrated. The Split VMAT plan demonstrated better in the $D_{mean}$ of whole rectum, Up to 134.4 cGy, at least 43.5 cGy, the average difference was 75.6 cGy and in the $D_{mean}$ of upper rectum, Up to 1113.5 cGy, at least 87.2 cGy, the average difference was 550.5 cGy and in the $D_{mean}$ of lower rectum, Up to 100.5 cGy, at least -34.6 cGy, the average difference was 34.3 cGy and in the $D_{mean}$ of bladder, Up to 271 cGy, at least -55.5 cGy, the average difference was 117.8 cGy and in $V_{50%}$ of upper rectum, Up to 63.4%, at least 3.2%, the average difference was 23.2%. There was no significant difference on H.I., and C.I. of the PTV among two plans. The Split VMAT plan is average 77 MU more than another. All IMRT verification gamma test results for the Split VMAT plan passed over 90.0% at 2 mm / 2%. Conclusion : As a result, the Split VMAT plan appeared to be more favorable in most cases than the Conventional VMAT plan for prostate cancer radiotherapy involving pelvic lymph nodes. By using the split VMAT planning technique it was possible to reduce the upper rectum dose, thus reducing whole rectal dose when compared to conventional VMAT planning. Also using the split VMAT planning technique increase the treatment efficiency.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.12 no.4 s.27
• /
• pp.293-300
• /
• 2006

Domestic laws such as Korea Marine Pollution Prevention Law (KMPPL) which has been mae and amended according to the conclusions and amendments of various international conventions for the prevention a marine pollution such as MARPOL 73/78 were reviewed and compared with the major contents of the relevant international conventions. Alternative measures for legislating new laws or amending existing laws such as KMPPL for the acceptance of major contents of existing international conventions were proposed. Annex VI of MARPOL 73/78 into which the regulations for the prevention of air pollution from ship have been adopted has been recently accepted in KMPPL which should be applied to ships which are the moving sources of air pollution at sea rather tlnn in Korea Air Environment Conservation Law which should be applied to automobiles and industrial installations in land. The major contents of LC 72/95 have been accepted in KMPPL However, a few of substances requiring special care in Annex II of 72LC, a few of items in characteristics and composition for the matter in relation to criteria governing the issue of permits for the dumping of matter at sea in Annex III of 72LC, and a few of items in wastes or other matter that may be considered for dumping in Annex I of 96 Protocol have not been accepted in KMPPL yet. The major contents of OPRC 90 have been accepted in KMPPL. However, oil pollution emergency plans for sea ports and oil handling facilities, and national contingency plan for preparedness and response have not been accepted in KMPPL yet. The waste oil related articles if Basel Convention, which shall regulate and prohibit transboundary movement of hazardous waste, should be accepted in KMPPL in order to prevent the transfer if scrap-purpose tanker ships containing oil/water mixtures and chemicals remained on beard from advanced countries to developing and/or underdeveloped countries. International Convention for the Control if Harmful Anti-Fouling Systems on the Ships should be accepted in KMPPL rather tlnn in Korea Noxious Chemicals Management Law. International Convention for Ship's Ballast Water/Sediment Management should be accepted in KMPPL or by a new law in order to prevent domestic marine ecosystem and costal environment from the invasion of harmful exotic species through the discharge of ship's ballast water.

• Journal of Intelligence and Information Systems
• /
• v.19 no.4
• /
• pp.39-53
• /
• 2013

The emergence of the internet technology and SNS has increased the information flow and has changed the way people to communicate from one-way to two-way communication. Users not only consume and share the information, they also can create and share it among their friends across the social network service. It also changes the Social Media behavior to become one of the most important communication tools which also includes Social TV. Social TV is a form which people can watch a TV program and at the same share any information or its content with friends through Social media. Social News is getting popular and also known as a Participatory Social Media. It creates influences on user interest through Internet to represent society issues and creates news credibility based on user's reputation. However, the conventional platforms in news services only focus on the news recommendation domain. Recent development in SNS has changed this landscape to allow user to share and disseminate the news. Conventional platform does not provide any special way for news to be share. Currently, Social News Service only allows user to access the entire news. Nonetheless, they cannot access partial of the contents which related to users interest. For example user only have interested to a partial of the news and share the content, it is still hard for them to do so. In worst cases users might understand the news in different context. To solve this, Social News Service must provide a method to provide additional information. For example, Yovisto known as an academic video searching service provided time dependent metadata from the video. User can search and watch partial of video content according to time dependent metadata. They also can share content with a friend in social media. Yovisto applies a method to divide or synchronize a video based whenever the slides presentation is changed to another page. However, we are not able to employs this method on news video since the news video is not incorporating with any power point slides presentation. Segmentation method is required to separate the news video and to creating time dependent metadata. In this work, In this paper, a time dependent metadata-based framework is proposed to segment news contents and to provide time dependent metadata so that user can use context information to communicate with their friends. The transcript of the news is divided by using the proposed story segmentation method. We provide a tag to represent the entire content of the news. And provide the sub tag to indicate the segmented news which includes the starting time of the news. The time dependent metadata helps user to track the news information. It also allows them to leave a comment on each segment of the news. User also may share the news based on time metadata as segmented news or as a whole. Therefore, it helps the user to understand the shared news. To demonstrate the performance, we evaluate the story segmentation accuracy and also the tag generation. For this purpose, we measured accuracy of the story segmentation through semantic similarity and compared to the benchmark algorithm. Experimental results show that the proposed method outperforms benchmark algorithms in terms of the accuracy of story segmentation. It is important to note that sub tag accuracy is the most important as a part of the proposed framework to share the specific news context with others. To extract a more accurate sub tags, we have created stop word list that is not related to the content of the news such as name of the anchor or reporter. And we applied to framework. We have analyzed the accuracy of tags and sub tags which represent the context of news. From the analysis, it seems that proposed framework is helpful to users for sharing their opinions with context information in Social media and Social news.

• Progress in Medical Physics
• /
• v.23 no.2
• /
• pp.91-98
• /
• 2012

Verification of internal organ motion during treatment and its feedback is essential to accurate dose delivery to the moving target. We developed an offline based internal organ motion verification system (IMVS) using cine EPID images and evaluated its accuracy and availability through phantom study. For verification of organ motion using live cine EPID images, a pattern matching algorithm using an internal surrogate, which is very distinguishable and represents organ motion in the treatment field, like diaphragm, was employed in the self-developed analysis software. For the system performance test, we developed a linear motion phantom, which consists of a human body shaped phantom with a fake tumor in the lung, linear motion cart, and control software. The phantom was operated with a motion of 2 cm at 4 sec per cycle and cine EPID images were obtained at a rate of 3.3 and 6.6 frames per sec (2 MU/frame) with $1,024{\times}768$ pixel counts in a linear accelerator (10 MVX). Organ motion of the target was tracked using self-developed analysis software. Results were compared with planned data of the motion phantom and data from the video image based tracking system (RPM, Varian, USA) using an external surrogate in order to evaluate its accuracy. For quantitative analysis, we analyzed correlation between two data sets in terms of average cycle (peak to peak), amplitude, and pattern (RMS, root mean square) of motion. Averages for the cycle of motion from IMVS and RPM system were $3.98{\pm}0.11$ (IMVS 3.3 fps), $4.005{\pm}0.001$ (IMVS 6.6 fps), and $3.95{\pm}0.02$ (RPM), respectively, and showed good agreement on real value (4 sec/cycle). Average of the amplitude of motion tracked by our system showed $1.85{\pm}0.02$ cm (3.3 fps) and $1.94{\pm}0.02$ cm (6.6 fps) as showed a slightly different value, 0.15 (7.5% error) and 0.06 (3% error) cm, respectively, compared with the actual value (2 cm), due to time resolution for image acquisition. In analysis of pattern of motion, the value of the RMS from the cine EPID image in 3.3 fps (0.1044) grew slightly compared with data from 6.6 fps (0.0480). The organ motion verification system using sequential cine EPID images with an internal surrogate showed good representation of its motion within 3% error in a preliminary phantom study. The system can be implemented for clinical purposes, which include organ motion verification during treatment, compared with 4D treatment planning data, and its feedback for accurate dose delivery to the moving target.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.231-252
• /
• 2021

Artificial intelligence (AI) is a key technology that will change the future the most. It affects the industry as a whole and daily life in various ways. As data availability increases, artificial intelligence finds an optimal solution and infers/predicts through self-learning. Research and investment related to automation that discovers and solves problems on its own are ongoing continuously. Automation of artificial intelligence has benefits such as cost reduction, minimization of human intervention and the difference of human capability. However, there are side effects, such as limiting the artificial intelligence's autonomy and erroneous results due to algorithmic bias. In the labor market, it raises the fear of job replacement. Prior studies on the utilization of artificial intelligence have shown that individuals do not necessarily use the information (or advice) it provides. Algorithm error is more sensitive than human error; so, people avoid algorithms after seeing errors, which is called "algorithm aversion." Recently, artificial intelligence has begun to be understood from the perspective of the augmentation of human intelligence. We have started to be interested in Human-AI collaboration rather than AI alone without human. A study of 1500 companies in various industries found that human-AI collaboration outperformed AI alone. In the medicine area, pathologist-deep learning collaboration dropped the pathologist cancer diagnosis error rate by 85%. Leading AI companies, such as IBM and Microsoft, are starting to adopt the direction of AI as augmented intelligence. Human-AI collaboration is emphasized in the decision-making process, because artificial intelligence is superior in analysis ability based on information. Intuition is a unique human capability so that human-AI collaboration can make optimal decisions. In an environment where change is getting faster and uncertainty increases, the need for artificial intelligence in decision-making will increase. In addition, active discussions are expected on approaches that utilize artificial intelligence for rational decision-making. This study investigates the impact of artificial intelligence on decision-making focuses on human-AI collaboration and the interaction between the decision maker personal traits and advisor type. The advisors were classified into three types: human, artificial intelligence, and human-AI collaboration. We investigated perceived usefulness of advice and the utilization of advice in decision making and whether the decision-maker's personal traits are influencing factors. Three hundred and eleven adult male and female experimenters conducted a task that predicts the age of faces in photos and the results showed that the advisor type does not directly affect the utilization of advice. The decision-maker utilizes it only when they believed advice can improve prediction performance. In the case of human-AI collaboration, decision-makers higher evaluated the perceived usefulness of advice, regardless of the decision maker's personal traits and the advice was more actively utilized. If the type of advisor was artificial intelligence alone, decision-makers who scored high in conscientiousness, high in extroversion, or low in neuroticism, high evaluated the perceived usefulness of the advice so they utilized advice actively. This study has academic significance in that it focuses on human-AI collaboration that the recent growing interest in artificial intelligence roles. It has expanded the relevant research area by considering the role of artificial intelligence as an advisor of decision-making and judgment research, and in aspects of practical significance, suggested views that companies should consider in order to enhance AI capability. To improve the effectiveness of AI-based systems, companies not only must introduce high-performance systems, but also need employees who properly understand digital information presented by AI, and can add non-digital information to make decisions. Moreover, to increase utilization in AI-based systems, task-oriented competencies, such as analytical skills and information technology capabilities, are important. in addition, it is expected that greater performance will be achieved if employee's personal traits are considered.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.27-65
• /
• 2020

Many companies on information and communication technology make public their own developed AI technology, for example, Google's TensorFlow, Facebook's PyTorch, Microsoft's CNTK. By releasing deep learning open source software to the public, the relationship with the developer community and the artificial intelligence (AI) ecosystem can be strengthened, and users can perform experiment, implementation and improvement of it. Accordingly, the field of machine learning is growing rapidly, and developers are using and reproducing various learning algorithms in each field. Although various analysis of open source software has been made, there is a lack of studies to help develop or use deep learning open source software in the industry. This study thus attempts to derive a strategy for adopting the framework through case studies of a deep learning open source framework. Based on the technology-organization-environment (TOE) framework and literature review related to the adoption of open source software, we employed the case study framework that includes technological factors as perceived relative advantage, perceived compatibility, perceived complexity, and perceived trialability, organizational factors as management support and knowledge & expertise, and environmental factors as availability of technology skills and services, and platform long term viability. We conducted a case study analysis of three companies' adoption cases (two cases of success and one case of failure) and revealed that seven out of eight TOE factors and several factors regarding company, team and resource are significant for the adoption of deep learning open source framework. By organizing the case study analysis results, we provided five important success factors for adopting deep learning framework: the knowledge and expertise of developers in the team, hardware (GPU) environment, data enterprise cooperation system, deep learning framework platform, deep learning framework work tool service. In order for an organization to successfully adopt a deep learning open source framework, at the stage of using the framework, first, the hardware (GPU) environment for AI R&D group must support the knowledge and expertise of the developers in the team. Second, it is necessary to support the use of deep learning frameworks by research developers through collecting and managing data inside and outside the company with a data enterprise cooperation system. Third, deep learning research expertise must be supplemented through cooperation with researchers from academic institutions such as universities and research institutes. Satisfying three procedures in the stage of using the deep learning framework, companies will increase the number of deep learning research developers, the ability to use the deep learning framework, and the support of GPU resource. In the proliferation stage of the deep learning framework, fourth, a company makes the deep learning framework platform that improves the research efficiency and effectiveness of the developers, for example, the optimization of the hardware (GPU) environment automatically. Fifth, the deep learning framework tool service team complements the developers' expertise through sharing the information of the external deep learning open source framework community to the in-house community and activating developer retraining and seminars. To implement the identified five success factors, a step-by-step enterprise procedure for adoption of the deep learning framework was proposed: defining the project problem, confirming whether the deep learning methodology is the right method, confirming whether the deep learning framework is the right tool, using the deep learning framework by the enterprise, spreading the framework of the enterprise. The first three steps (i.e. defining the project problem, confirming whether the deep learning methodology is the right method, and confirming whether the deep learning framework is the right tool) are pre-considerations to adopt a deep learning open source framework. After the three pre-considerations steps are clear, next two steps (i.e. using the deep learning framework by the enterprise and spreading the framework of the enterprise) can be processed. In the fourth step, the knowledge and expertise of developers in the team are important in addition to hardware (GPU) environment and data enterprise cooperation system. In final step, five important factors are realized for a successful adoption of the deep learning open source framework. This study provides strategic implications for companies adopting or using deep learning framework according to the needs of each industry and business.