• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.7
• /
• pp.157-168
• /
• 2014

We propose an novel artificial visual attention model that is capable of automatic detection and segmentation of saliency region on natural images in this paper. The proposed model is based on human visual perceptions in biological vision and contains there are main contributions. Firstly, we propose a novel framework of artificial visual attention model based on the opponent process theory using intensity and color features, and an entropy filter is designed to perceive salient regions considering the amount of information from intensity and color feature channels. The entropy filter is able to detect and segment salient regions in high segmentation accuracy and precision. Lastly, we also propose an adaptive combination method to generate a final saliency map. This method estimates scores about intensity and color conspicuous maps from each perception model and combines the conspicuous maps with weight derived from scores. In evaluation of saliency map by ROC analysis, the AUC of proposed model as 0.9256 approximately improved 15% whereas the AUC of previous state-of-the-art models as 0.7824. And in evaluation of salient region segmentation, the F-beta of proposed model as 0.7325 approximately improved 22% whereas the F-beta of previous state-of-the-art models.

• Asia pacific journal of information systems
• /
• v.20 no.1
• /
• pp.33-56
• /
• 2010

The current business environment has been characterized as less munificent, highly uncertain and constantly evolving. In this environment, the company with dynamic capability is reported to be more successful than others in building competitive advantage. Dynamic capability focuses on the link between a dynamically changing environment, strategic agility, architectural reconfiguration, and value creation. Being characterized to be flexible and adaptive to market circumstance changes, an organization with dynamic capability is described to have high resource fluidity, which represents business process, resource allocation, human resource management and incentives that make business transformation faster and easier. Successful redeployment of the resources for dynamic adaptation requires organizational forms and reward systems to be well aligned with firm's technological infrastructures and business process. The alignment is considered to be an executive level commitment. Building dynamic capability is knowledge driven; relying on new knowledge to reconfigure firm's resources. Past studies established the link between the effective execution of a knowledge-focused strategy and relevant setting of architectural elements such as human resources, structure, process and information systems. They do not, however, describe in detail the underlying processes by which architectural elements are adjusted in coordinated manners to build knowledge-driven dynamic capability. In fact, understandings of these processes are one of the top issues in IT management. This study analyzed how a Korean corporation with a knowledge-focused strategy aligned its architectural elements to develop the dynamic capability and thus create value in the dynamically changing markets. When the Korean economy was in crisis, the company implemented a knowledge-focused strategy, restructured the organization's architecture by which human and knowledge resources are identified, structured, integrated and coordinated to identify and seize market opportunity. Specifically, the following architectural elements were reconfigured: human resource, decision rights, reward and evaluation systems, process, and IT infrastructure. As indicated by sales growth, the reconfiguration helped the company create value under an extremely turbulent environment. According to Ancona et al. (2001), depending on the types of lenses the organization uses, different types of architecture will emerge. For example, if an organization uses political lenses focusing on power, influence, and conflict. the architecture that leverage power and negotiate across multiple interest groups would emerge. Similarly, if an organization uses economic lenses focusing on the rational behavior of organizational actors making choices based on the costs and benefits of action, organizational architecture should be designed to motivate and provide incentives for the actors (Smith, 2001). Compared to this view, information processing perspectives consider architecture to be designed to maximize the capacity of information processing by the actors. Using knowledge lenses, the company studied in this research established architectural elements in a manner that allows the firm to effectively structure knowledge resources to form dynamic capability. This study is revelatory single case with a historic perspective. As a result of this study, a set of propositions and a framework are derived, which can be used for architectural alignment.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.326-334
• /
• 2022

In this paper, we propose speech enhancement algorithm as a pre-processing for robust speech recognition in noisy environments. Auxiliary-function-based Independent Vector Analysis (AuxIVA) is performed with weighted covariance matrix using time-varying variances with scaling factor from target masks representing time-frequency contributions of target speech. The mask estimates can be obtained using Neural Network (NN) pre-trained for speech extraction or diffuseness using Coherence-to-Diffuse power Ratio (CDR) to find the direct sounds component of a target speech. In addition, outputs for omni-directional noise are closely chained by sharing the time-varying variances similarly to independent subspace analysis or IVA. The speech extraction method based on AuxIVA is also performed in Independent Low-Rank Matrix Analysis (ILRMA) framework by extending the Non-negative Matrix Factorization (NMF) for noise outputs to Non-negative Tensor Factorization (NTF) to maintain the inter-channel dependency in noise output channels. Experimental results on the CHiME-4 datasets demonstrate the effectiveness of the presented algorithms.

• 한국노년학
• /
• v.39 no.2
• /
• pp.325-345
• /
• 2019

The purpose of this study is a exploratory study for understanding the process of the social isolation of the socially isolated elderly through the approach to their life history. The research was analyzed by one of the methods of qualitative research on life history, the conceptual framework of 'Dimensions, turning, and adaptation' of Mandelbaum(1973). According to the results of this study, the socially isolated elderly people were found to be socially isolated by experiencing complex difficulties such as family disconnection, poverty, poor job and health deterioration. Specifically, in the area of life, there was experience of poor relationship with parent, absence of family, poverty of family and unfavorable relationship with surrounding people in life with original family before isolation. They had bad jobs in the labor market, such as hard labor, delivery, business, and chores. In the area of turning point, we experienced family break due to the separation of the original family and the spouse due to various reasons such as financial crisis, parental divorce and death, spouse affair, economic difficulty. In a transitional stage in the life, many reasons such as the financial crisis, the death of parents, the extramarital affair and economic difficulties led to the disconnection from their original family and their spouses. In an adaptive phase, participants accepted the changed life at each turning point in their lives, carrying out their roles, compromising and trying to adapt properly. He said that their current life, which has entered the social safety net system of the people's basic recipients, has led him to live a more stable life and is adapting to personal hobbies and vicarious satisfaction through networks. This result is somewhat different from previous studies in which isolated elderly people were severely exposed to the risk of depression and loneliness. However, we should also consider the characteristics of this study that interviewed elderly people with relatively low isolation. Based on the results of this research, he presented various practical policy implications.

• Journal of Environmental Policy
• /
• v.13 no.1
• /
• pp.69-93
• /
• 2014

The aim of this research was to develop a climate change vulnerability index at the district level (Si, Gun, Gu) with respect to the health care sector in Korea. The climate change vulnerability index was esimated based on the four major causes of climate-related illnesses : vector, flood, heat waves, and air pollution/allergies. The vulnerability assessment framework consists of six layers, all of which are based on the IPCC vulnerability concepts (exposure, sensitivity, and adaptive capacity) and the pathway of direct and indirect impacts of climate change modulators on health. We collected proxy variables based on the conceptual framework of climate change vulnerability. Data were standardized using the min-max normalization method. We applied the analytic hierarchy process (AHP) weight and aggregated the variables using the non-compensatory multi-criteria approach. To verify the index, sensitivity analysis was conducted by using another aggregation method (geometric transformation method, which was applied to the index of multiple deprivation in the UK) and weight, calculated by the Budget Allocation method. The results showed that it would be possible to identify the vulnerable areas by applying the developed climate change vulnerability assessment index. The climate change vulnerability index could then be used as a valuable tool in setting climate change adaptation policies in the health care sector.

• Asia pacific journal of information systems
• /
• v.19 no.2
• /
• pp.95-116
• /
• 2009

Increasingly the institutional theory has been an important theoretical view of decision making process and IT adoption in many academic researches. This study used the institutional theory as a lens through which we can understand the factors that enable the effective appropriation of advanced information technology. It posits that mimetic, coercive, and normative pressures existing in an institutionalized environment could influence the participation of top managers or decision makers and the involvement of users toward an effective use of IT in their tasks. Since the introduction of IT, organizational members have been using IT in their daily tasks, creating and recreating rules and resources according to their own methods and needs. That is to say, the adaptation process of the IT and outcomes are different among organizations. The previous studies on a diverse use of IT refer to the appropriation of technology from the social technology view. Users appropriate IT through not only technology itself, but also in terms of how they use it or how they make the social practice in their use of it. In this study, the concepts of institutional pressure, appropriation, participation of decision makers, and involvement of users toward the appropriation are explored in the context of the appropriation of the mobile electronic medical record system (M-EMRS) in particularly a hospital setting. Based on the conceptual definition of institutional pressure, participation and involvement, operational measures are reconstructed. Furthermore, the concept of appropriation is measured in the aspect of three sub-constructs-consensus on appropriation, faithful appropriation, and attitude of use. Grounded in the relevant theories to appropriation of IT, we developed a research framework in which the effects of institutional pressure, participation and involvement on the appropriation of IT are analyzed. Within this theoretical framework, we formulated several hypotheses. We developed a second order institutional pressure and appropriation construct. After establishing its validity and reliability, we tested the hypotheses with empirical data from 101 users in 3 hospitals which had adopted and used the M-EMRS. We examined the mediating effect of the participation of decision makers and the involvement of users on the appropriation and empirically validated their relationships. The results show that the mimetic, coercive, and normative institutional pressure has an effect on the participation of decision makers and the involvement of users in the appropriation of IT while the participation of decision makers and the involvement of users have an effect on the appropriation of IT. The results also suggest that the institutional pressure and the participation of decision makers influence the involvement of users toward an appropriation of IT. Our results emphasize the mediating effect of the institutional pressure on the appropriation of IT. Namely, the higher degree of the participation of decision makers and the involvement of users, the more effective appropriation users will represent. These results provide strong support for institutional-based variables as predictors of appropriation. These findings also indicate that organizations should focus on the role of participation of decision makers and the involvement of users for the purpose of effective appropriation, and these are the practical implications of our study. The theoretical contribution of this study is lies in the integrated model of the effect of institutional pressure on the appropriation of IT. The results are consistent with the institutional theory and support previous studies on adaptive structuration theory.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.