• Journal of KIISE
• /
• v.45 no.1
• /
• pp.30-35
• /
• 2018

GuessWhat?! is a game in which two machine players, composed of questioner and answerer, ask and answer yes-no-N/A questions about the object hidden for the answerer in the image, and the questioner chooses the correct object. GuessWhat?! has received much attention in the field of deep learning and artificial intelligence as a testbed for cutting-edge research on the interplay of computer vision and dialogue systems. In this study, we discuss the objective function and characteristics of the GuessWhat?! game. In addition, we propose a simple solver for GuessWhat?! using a simple rule-based algorithm. Although a human needs four or five questions on average to solve this problem, the proposed method outperforms state-of-the-art deep learning methods using only two questions, and exceeds human performance using five questions.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.6
• /
• pp.313-330
• /
• 2021

In an autonomous driving system equipped with a camera and lidar, depth completion techniques enable dense depth estimation. In particular, using self-supervised learning it is possible to train the depth completion network even without ground truth. In actual autonomous driving, such depth completion should have very short latency as it is the input of other algorithms. So, rather than complicate the network structure to increase the accuracy like previous studies, this paper focuses on network latency. We design a U-Net type network with RegNet encoders optimized for GPU computation. Instead, this paper presents several techniques that can increase accuracy during the process of self-supervised learning. The proposed techniques increase the robustness to unreliable lidar inputs. Also, they improve the depth quality for edge and sky regions based on the semantic information extracted in advance. Our experiments confirm that our model is very lightweight (2.42 ms at 1280x480) but resistant to noise and has qualities close to the latest studies.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.5
• /
• pp.821-830
• /
• 2023

The advancement of Internet network technology has triggered the emergence of new OTT video content platforms, increasing demand for content and altering consumption patterns. This trend is bringing positive changes to the South Korean animation industry, where diverse and high-quality animation content is becoming increasingly important. As investment in technology grows, video production technology continues to advance. Specifically, 3D animation and VFX production technologies are enabling effects that were previously unthinkable, offering detailed and realistic graphics. The Fourth Industrial Revolution is providing new opportunities for this technological growth. The rise of Artificial Intelligence (AI) is automating repetitive tasks, thereby enhancing production efficiency and enabling innovations that go beyond traditional production methods. Cutting-edge technologies like 3D animation and VFX are being continually researched and are expected to be more actively integrated into the production process. Digital technology is also expanding the creative horizons for artists. The future of AI and advanced technologies holds boundless potential, and there is growing anticipation for how these will elevate the video content industry to new heights.

• Convergence Security Journal
• /
• v.24 no.2
• /
• pp.9-17
• /
• 2024

This paper examines the security threats to enterprise Generative Artificial Intelligence systems and proposes countermeasures. As AI systems handle vast amounts of data to gain a competitive edge, security threats targeting AI systems are rapidly increasing. Since AI security threats have distinct characteristics compared to traditional human-oriented cybersecurity threats, establishing an AI-specific response system is urgent. This study analyzes the importance of AI system security, identifies key threat factors, and suggests technical and managerial countermeasures. Firstly, it proposes strengthening the security of IT infrastructure where AI systems operate and enhancing AI model robustness by utilizing defensive techniques such as adversarial learning and model quantization. Additionally, it presents an AI security system design that detects anomalies in AI query-response processes to identify insider threats. Furthermore, it emphasizes the establishment of change control and audit frameworks to prevent AI model leakage by adopting the cyber kill chain concept. As AI technology evolves rapidly, by focusing on AI model and data security, insider threat detection, and professional workforce development, companies can improve their digital competitiveness through secure and reliable AI utilization.

• Journal of Technology Innovation
• /
• v.15 no.2
• /
• pp.123-152
• /
• 2007

The striking characteristic of the contemporary global security environment is that the nature of threats has become diverse and complex. For example, transnational and non-military threats including terrorism and proliferation of weapon of mass destruction has increased. In this security environment, Advanced countries funnel their investments for defense budgets into the assurance of key force capability and R&D of cutting-edge core technologies, in consideration of future battlefield environments so as to get an edge on not only defense science and technology but also intelligence capabilities. As shown by past practices of the korea's defense acquisition, the ministry of national defense has tried to enhance its force capabilities in the short-term by purchasing foreign weapon systems rather than by investing in domestic R&D. Accordingly, the technological gaps between the korea and advanced countries were widened due to both insufficient investment in development of domestic technologies and avoidance of technological transfer by advanced countries. Thus, for the effective execution of the R&D budget and the successful performance of the projects, the importance of selection, management and evaluation of the R&D projects is emphasized. So, The objective of this study is that the analysis of the proposal-selection evaluation system for the realization of the successful defense core technology R&D projects. This study focused on the improvement of the proposal-selection evaluation model which can be applicable to the national defense core R&D projects. Using the improved proposal-selection evaluation system, we propose a model to enhance the reliability of the national defense core technology R&D project evaluation system.

• Journal of Intelligence and Information Systems
• /
• v.15 no.4
• /
• pp.159-182
• /
• 2009

We construct product networks from the retail transaction dataset of an off-line department store. In the product networks, nodes are products, and an edge connecting two products represents the existence of co-purchases by a customer. We measure the quantities frequently used for characterizing network structures, such as the degree centrality, the closeness centrality, the betweenness centrality and the centralization. Using the quantities, gender, age, seasonal, and regional differences of the product networks were analyzed and network characteristics of each product category containing each product node were derived. Lastly, we analyze the correlations among the three centrality quantities and draw a marketing strategy for the cross-selling.

• Journal of Intelligence and Information Systems
• /
• v.15 no.2
• /
• pp.1-13
• /
• 2009

Scene classification and concept-based procedures have been the great interest for image categorization applications for large database. Knowing the category to which scene belongs, we can filter out uninterested images when we try to search a specific scene category such as beach, mountain, forest and field from database. In this paper, we propose an adaptive segmentation method for real-world natural scene classification based on a semantic modeling. Semantic modeling stands for the classification of sub-regions into semantic concepts such as grass, water and sky. Our adaptive segmentation method utilizes the edge detection to split an image into sub-regions. Frequency of occurrences of these semantic concepts represents the information of the image and classifies it to the scene categories. K-Nearest Neighbor (k-NN) algorithm is also applied as a classifier. The empirical results demonstrate that the proposed adaptive segmentation method outperforms the Vogel and Schiele's method in terms of accuracy.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.16 no.11
• /
• pp.7718-7728
• /
• 2015

Semantic similarity between words can be applied in many fields including computational linguistics, artificial intelligence, and information retrieval. In this paper, we present weighted method for measuring a semantic similarity between words in a document. This method uses edge distance and depth of WordNet. The method calculates a semantic similarity between words on the basis of document information. Document information uses word term frequencies(TF) and word concept frequencies(CF). Each word weight value is calculated by TF and CF in the document. The method includes the edge distance between words, the depth of subsumer, and the word weight in the document. We compared out scheme with the other method by experiments. As the result, the proposed method outperforms other similarity measures. In the document, the word weight value is calculated by the proposed method. Other methods which based simple shortest distance or depth had difficult to represent the information or merge informations. This paper considered shortest distance, depth and information of words in the document, and also improved the performance.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.7
• /
• pp.75-82
• /
• 2022

There are various exhibition halls, shopping malls, theme parks around us and analysis of interest in exhibits or contents is mainly done through questionnaires. These questionnaires are mainly depend on the subjective memory of the person being investigated, resulting in incorrect statistical results. Therefore, it is possible to identify an exhibition space with low interest by tracking the movement and counting the number of visitors. Based on this, it can be used as quantitative data for exhibits that need replacement. In this paper, we use deep learning-based artificial intelligence algorithms to recognize visitors, assign IDs to the recognized visitors, and continuously track them to identify the movement path. When visitors pass the counting line, the system is designed to count the number and transmit data to the server for integrated management.

• Journal of Advanced Navigation Technology
• /
• v.27 no.1
• /
• pp.71-76
• /
• 2023

In this paper, a humidity sensor using a microstrip patch antenna(MPA) and polyvinyl alcohol(PVA) is studied. PVA is a polymer material whose permittivity changes with humidity, and a rectangular slot is added to the radiating edge of the MPA, which is sensitive to changes in electric field, in order to increase the sensitivity to changes in relative permittivity. After thinly coating the area around the radiating edge with the rectangular slot of the MPA fabricated on a 0.76 mm-thick RF-35 substrate with PVA, the changes in the resonant frequency and magnitude of the MPA's input reflection coefficient are measured when relative humidity is adjusted from 40% to 80% in 10% increments at a temperature of 25 degrees using a temperature and humidity chamber. Experiment results show that when the relative humidity increases from 40% to 80%, the resonance frequency of the antenna' input reflection coefficient decreases from 2.447 GHz to 2.418 GHz, whereas the magnitude increases from -7.112 dB to -3.428 dB.