• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.18 no.6
• /
• pp.57-70
• /
• 2023

The investigation delved into 457 pieces of scholarly work, encompassing articles, published theses, and dissertations from the National Research Foundation of Korea, spanning the period of the 1997 IMF financial crisis up to 2022. The materials were sourced using terms such as 'angel investment', 'angel investor', and 'angel investment attraction'. The initial phase involved filtering out redundant entries from the preliminary collection of 267 works, leaving aside pieces that didn't pertain directly to angel investment as indicated in their abstracts. The next stage of the analysis involved a more rigorous selection process. Out of 43 papers earmarked in the preceding cut, only 32 were chosen. The criteria for this focused on the exclusion of conference presentations, articles that were either not submitted or inconclusive, and those that duplicated content under different titles. The final selection of 32 papers underwent a thorough systematic literature review. These documents, all pertinent to angel investment in South Korea, were scrutinized under five distinct categories: 1) publication year, 2) themes of research, 3) strategies employed in the studies, 4) participants involved in the research, and 5) methods of research utilized. This meticulous process illuminated the existing landscape of angel investment studies within Korea. Moreover, this study pinpointed gaps in the current body of research, offering guidance on future scholarly directions and proposing social scientific theories to further enrich the field of angel investment studies and analysis also seeks to pinpoint which areas require additional exploration to energize the field of angel investment moving forward. Through a comprehensive review of literature, this research intends to validate the establishment of future research trajectories and pinpoint areas that are currently and relatively underexplored in Korea's angel investment research stream. This study revealed that current research on domestic angel investment is concentrated on several areas: 1) the traits of angel investors, 2) the motivations behind angel investing, 3) startup ventures, 4) relevant institutions and policies, and 5) the various forms of angel investments. It was determined that there is a need to broaden the scope of research to aid in enhancing and stimulating the scale of domestic angel investing. This includes research into performance analysis of angel investments and detailed case studies in the field. Furthermore, the study emphasizes the importance of diversifying research efforts. Instead of solely focusing on specific factors like investment types, startups, accelerators, venture capital, and regulatory frameworks, there is a call for research that explores a variety of associated variables. These include aspects related to crowdfunding and return on investment in the context of angel investing, ensuring a more holistic approach to research in this domain. Specifically, there's a clear need for more detailed studies focusing on the relationships with variables that serve as dependent variables influencing the outcomes of angel investments. Moreover, it's essential to invigorate both qualitative and quantitative research that delves into the theoretical framework from multiple perspectives. This involves analyzing the structure of variables that have an impact on angel investments and the decisions surrounding these investments, thereby enriching the theoretical foundation of this field. Finally, we presented the direction of development for future research by confirming that the effect on the completeness of the business plan is high or low depending on the satisfaction of the entrepreneurs in addition to the components.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.9 no.1
• /
• pp.119-132
• /
• 2014

This study analyzed the failure prediction model of the firms listed on the KOSDAQ by considering whether there are embezzlement, malpractice and the largest shareholder changes or not. This study composed a total of 166 firms by using two-paired sampling method. For sample of failed firm, 83 manufacturing firms which delisted on KOSDAQ market for 4 years from 2009 to 2012 are selected. For sample of normal firm, 83 firms (with same item or same business as failed firm) that are listed on KOSDAQ market and perform normal business activities during the same period (from 2009 to 2012) are selected. This study selected 80 financial ratios for 5 years immediately preceding from delisting of sample firm above and conducted T-test to derive 19 of them which emerged for five consecutive years among significant variables and used forward selection to estimate logistic regression model. While the precedent studies only analyzed the data of three years immediately preceding the delisting, this study analyzes data of five years immediately preceding the delisting. This study is distinct from existing previous studies that it researches which significant financial characteristic influences the insolvency from the initial phase of insolvent firm with time lag and it also empirically analyzes the usefulness of data by building a firm's fail prediction model which considered embezzlement/malpractice and the largest shareholder changes as dummy variable(non-financial characteristics). The accuracy of classification of the prediction model with dummy variable appeared 95.2% in year T-1, 88.0% in year T-2, 81.3% in year T-3, 79.5% in year T-4, and 74.7% in year T-5. It increased as year of delisting approaches and showed generally higher the accuracy of classification than the results of existing previous studies. This study expects to reduce the damage of not only the firm but also investors, financial institutions and other stakeholders by finding the firm with high potential to fail in advance.

• Economic and Environmental Geology
• /
• v.51 no.4
• /
• pp.381-392
• /
• 2018

This study examines strategies and implementation plans for commercializing $CO_2$ capture and storage, which is an effective method to achieve the national goal of reducing greenhouse gas. In order to secure cost-efficient business model of $CO_2$ capture and storage, we propose four key strategies, including 1) urgent need to select a large-scale storage site and to estimate realistic storage capacity, 2) minimization of source-to-sink distance, 3) cost-effectiveness through technology innovation, and 4) policy implementation to secure public interest and to encourage private sector participation. Based on these strategies, the implementation plans must be designed for enabling $CO_2$ capture and storage to be commercialized until 2030. It is desirable to make those plans in which large-scale demonstration and subsequent commercial projects share a single storage site. In addition, the plans must be able to deliver step-wised targets and assessment processes to decide if the project will move to the next stage or not. The main target of stage 1 (2019 ~ 2021) is that the large-scale storage site will be selected and post-combustion capture technology will be upgraded and commercialized. The site selection, which is prerequisite to forward to the next stage, will be made through exploratory drilling and investigation for candidate sites. The commercial-scale applicability of the capture technology must be ensured at this stage. Stage 2 (2022 ~ 2025) aims design and construction of facility and infrastructure for successful large-scale demonstration (million tons of $CO_2$ per year), i.e., large-scale $CO_2$ capture, transportation, and storage. Based on the achievement of the demonstration project and the maturity of carbon market at the end of stage 2, it is necessary to decide whether to enter commercialization of $CO_2$ capture and storage. If the commercialization project is decided, it will be possible to capture and storage 4 million tons of $CO_2$ per year by the private sector in stage 3 (2026 ~ 2030). The existing facility, infrastructure, and capture plant will be upgraded and supplemented, which allows the commercialization project to be cost-effective.

• Proceedings of the Optical Society of Korea Conference
• /
• 2003.07a
• /
• pp.60-61
• /
• 2003

A new approach to reduce the computation time of genetic algorithm (GA) for making binary phase holograms is described. Synthesized holograms having diffraction efficiency of 75.8% and uniformity of 5.8% are proven in computer simulation and experimentally demonstrated. Recently, computer-generated holograms (CGHs) having high diffraction efficiency and flexibility of design have been widely developed in many applications such as optical information processing, optical computing, optical interconnection, etc. Among proposed optimization methods, GA has become popular due to its capability of reaching nearly global. However, there exits a drawback to consider when we use the genetic algorithm. It is the large amount of computation time to construct desired holograms. One of the major reasons that the GA' s operation may be time intensive results from the expense of computing the cost function that must Fourier transform the parameters encoded on the hologram into the fitness value. In trying to remedy this drawback, Artificial Neural Network (ANN) has been put forward, allowing CGHs to be created easily and quickly (1), but the quality of reconstructed images is not high enough to use in applications of high preciseness. For that, we are in attempt to find a new approach of combiningthe good properties and performance of both the GA and ANN to make CGHs of high diffraction efficiency in a short time. The optimization of CGH using the genetic algorithm is merely a process of iteration, including selection, crossover, and mutation operators [2]. It is worth noting that the evaluation of the cost function with the aim of selecting better holograms plays an important role in the implementation of the GA. However, this evaluation process wastes much time for Fourier transforming the encoded parameters on the hologram into the value to be solved. Depending on the speed of computer, this process can even last up to ten minutes. It will be more effective if instead of merely generating random holograms in the initial process, a set of approximately desired holograms is employed. By doing so, the initial population will contain less trial holograms equivalent to the reduction of the computation time of GA's. Accordingly, a hybrid algorithm that utilizes a trained neural network to initiate the GA's procedure is proposed. Consequently, the initial population contains less random holograms and is compensated by approximately desired holograms. Figure 1 is the flowchart of the hybrid algorithm in comparison with the classical GA. The procedure of synthesizing a hologram on computer is divided into two steps. First the simulation of holograms based on ANN method [1] to acquire approximately desired holograms is carried. With a teaching data set of 9 characters obtained from the classical GA, the number of layer is 3, the number of hidden node is 100, learning rate is 0.3, and momentum is 0.5, the artificial neural network trained enables us to attain the approximately desired holograms, which are fairly good agreement with what we suggested in the theory. The second step, effect of several parameters on the operation of the hybrid algorithm is investigated. In principle, the operation of the hybrid algorithm and GA are the same except the modification of the initial step. Hence, the verified results in Ref [2] of the parameters such as the probability of crossover and mutation, the tournament size, and the crossover block size are remained unchanged, beside of the reduced population size. The reconstructed image of 76.4% diffraction efficiency and 5.4% uniformity is achieved when the population size is 30, the iteration number is 2000, the probability of crossover is 0.75, and the probability of mutation is 0.001. A comparison between the hybrid algorithm and GA in term of diffraction efficiency and computation time is also evaluated as shown in Fig. 2. With a 66.7% reduction in computation time and a 2% increase in diffraction efficiency compared to the GA method, the hybrid algorithm demonstrates its efficient performance. In the optical experiment, the phase holograms were displayed on a programmable phase modulator (model XGA). Figures 3 are pictures of diffracted patterns of the letter "0" from the holograms generated using the hybrid algorithm. Diffraction efficiency of 75.8% and uniformity of 5.8% are measured. We see that the simulation and experiment results are fairly good agreement with each other. In this paper, Genetic Algorithm and Neural Network have been successfully combined in designing CGHs. This method gives a significant reduction in computation time compared to the GA method while still allowing holograms of high diffraction efficiency and uniformity to be achieved. This work was supported by No.mOl-2001-000-00324-0 (2002)) from the Korea Science & Engineering Foundation.

• The Korean Journal of Archival Studies
• /
• no.20
• /
• pp.151-212
• /
• 2009

Museums are producing new value and being redefined as places that reproduce context, as the process of globalization are being reflected in museum activities. The new additional functions and roles to the traditional mission of museums allow artworks to find potential functions of art archive and meseum. At the same time, the public has faced originality and aura of an artwork by viewing the physical subject. However, with the appearance of a new digital object, the initiative of viewing has moved over from the artwork to the hands of the public. Now, the public does not go to the museum to see an artwork, but has started to adopt to an opposite paradigm of bringing the artwork forward to the screen. Therefore, they are not satisfied any longer with just seeing an artwork, but demand more information about the artworks and reproduce it as knowledge. Therefore, this study aimed to find types and characteristics through definition and range selection of art archive at this point where the value of art archive is enhanced and systematic management is required, and to present record management methods according to art archive structure and core execution function. It especially stressed that the basis of overall art archive definition was in an 'approach' paradigm rather than a 'preservation' paradigm, and embodied various application methods of digitalized art records. The digital object of an artwork was recognized as the first materialization of an actual artwork, and the digital original of an artwork was presented as the core record. Art archive managed under physical and intellectual control were organically restructured focusing on digital original copies of artworks, which are the core record in a digital technology environment, and could be provided to users in forms of various services that meet their demands. The beginning of systematic management of such art records will become a first step to enhance historical value, establish art cultural identity, and truly possess art culture.

• Journal of the Korean Applied Science and Technology
• /
• v.40 no.4
• /
• pp.685-698
• /
• 2023

This study aimed to identify binge-eating behavior and food addiction in Korean population and to determine their associations with obesity, eating behaviors, mental health and cognitive characteristics. We collected clinical questionnaire scores related to eating problems (e.g. binge eating, food addiction, food cravings), mental health (e.g. depression), and cognitive functions (e.g. impulsivity, emotion regulation) in 257 Korean adults in the normal and the obese weight ranges. Binge-eating and food addiction were most frequent in obese women (binge-eating: 46.6%, food addiction: 29.3%) when we divided the participants into 4 groups depending on gender and obesity status. The independence test using the data with propensity score matching confirmed that binge-eating and food addiction were more prevalent in obese individuals. Finally, we constructed the logistic regression models using forward selection method to evaluate the influence of various clinical questionnaire scores on binge-eating and food addiction respectively. Binge-eating was significantly associated with the clinical scales of eating disorders, food craving, state anxiety, and emotion regulation (cognitive reappraisal) as well as food addiction. Food addiction demonstrated the significant effect of food craving, binge-eating, the interaction of obesity and age, and years of education. In conclusion, we found that binge-eating and food addiction are much more frequent in females and obese individuals. Both binge-eating and food addiction commonly involved eating problems (e.g. food craving), but there was difference in mental health and cognitive risk factors. Therefore, it is required to distinguish food addiction from binge-eating and investigate intrinsic and environmental risk factors for each pathology.