• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.31-36
• /
• 2024

Recently, mobile apps that utilize AI technologies are increasing. In the personal mobile environment, performance degradation may occur during the training phase of large AI workload due to limitations in memory capacity. In this paper, we extract memory reference traces of AI workloads and analyze their characteristics. From this analysis, we observe that AI workloads can cause frequent storage access due to weak temporal locality and irregular popularity bias during memory write operations, which can degrade the performance of mobile devices. Based on this observation, we discuss ways to efficiently manage memory write operations of AI workloads using persistent memory-based swap devices. Through simulation experiments, we show that the system architecture proposed in this paper can improve the I/O time of mobile systems by more than 80%.

• The Journal of the Korea Contents Association
• /
• v.21 no.4
• /
• pp.205-219
• /
• 2021

The present study identified determinants to user acceptance of artificial intelligence(AI)-based news anchor. Our conceptual model included three constructs of ability, benevolence, and integrity to determine whether these three constructs are predictive of trust perceived from AI news anchor. This work further examined the influences of social presence, anthropomorphism, perceived usefulness, understanding as well as trust as immediate determinants to user acceptance. The conceptual model was validated on survey data collected from 513 respondents. A series of scale refinement process was conducted by the examination of data normality, common method bias, structure of latent variables as well as internal consistency. In addition, a confirmatory factor analysis was performed to assess the extent to which the sample data collected from survey study measures the constructs adequately. The results from the analysis of structural equation model indicated that, (1) two constructs of ability and integrity were found to be significantly predictive of perceived trust, and (2) anthropomorphism, perceived usefulness, and trust emerged as significant and positive predictors of user acceptance of AI-based news anchor.

• Proceedings of the Korean Institute of Surface Engineering Conference
• /
• 2003.05a
• /
• pp.14-14
• /
• 2003

Silicon carbide (SiC)는 높은 power 영역과 높은 온도영역에서도 작동 가능한 우수한 반도체 물질이다. 또한 우수한 열적 화학적, 안정성을 가지고 있어 가흑한 조건에서의 소자로써도 사용 가능하다. 현재 SiC 적용분야로는 우수한 전기적, 기계적 성질을 이용한 미세소자(MEMS)와 GaN 와 거의 유사한 격자상수를 가지는 것을 이용한 GaN epitaxial 성장의 기판으로도 사용되어진다. 그러나 SiC 는 기존의 습식식각 용매에 대해 화학적 안정성을 가지고 있기 때문에 전자소자의 제작에 있어서 플라즈마를 이용한 건식식각의 중요성이 대두되어지고 있다. 소자제작에 있어 이러한 건식식각시 식각 단면의 제어, 이온에 의한 낮은 손상 정도, 매끄러운 식각 표면, 그리고 고속의 식각 속도둥이 요구되어진다. 본 실험에서는 식각 속도의 증가와 수직한 식각 단면둥을 획득하기 위하여 SF6 플라즈마에서 Source power, dc bias voltage, 그리고 외부에서 인가되는 자속의 세기를 변화시쳐가며 식각 속도, 식각 마스크와의 식각 션택비, 식각 단면둥과 같은 SiC 의 식각 특성을 관찰하였다. 식각 후 식각 단면은 주사전자 현미경(SEM)을 통해 관찰하였다. 본 실험에서의 가장 높은 식각 속도는 분당 1850n 로써 이때의 공정조건은 1400W 의 inductive power, -600V 의 dc bias voltage, 20G 의 외부자속 세기이었다. 또한, 높은 inductive power 조건과 낮은 dc bias voltage 조건에서 Cu는 $SF_6$ 플라즈마 내에서 식각부산물의 증착으로 인해 SiC 와 무한대의 식각선택비를 보였다. 이러한 Cu 마스크를 사용한 SiC 의 식각에서는 식각 후 수직한 식각 단변을 관찰할 수 있었다. 것올 알 수 있다. 따라서, 기존의 pve 보다 세라믹 기판의 경우가 수분 흡수율이 높아 더 오랫동안 전류를 흐르게 하여 방식성이 개선된 것으로 판단된다.을 통해 경도가 증가한 시편의 경우 석출상의 크기가 5nm 이하로 매우 작고 대체로 기지와 연속적인 계면을 형성하나, 열처리가 진행될수록 석 출상의 크기가 커지고 임계크기 이상에 이르면 연속적인 계면은 거의 발견되지 않고, 대부 분 불연속적이고 확연한 계면을 형성함을 관찰 할 수 있었다. 알루미나(${\alpha}-Al_2O_3$) 기판 위에 증착한 $(Ti_{1-x}AI_{x})N$ 피막은 마찬가지로 (200) 우선 방위를 나타내었으나, 그 입자의 크기가 수십 nm로 고속도강위에 증착한 피막에 비해 상당히 크게 형성되었다. 또한 열처리 후에 AIN의 석출이 진행됨에도 불구하고 경도 증가는 나타나지 않고, 열처리가 진행됨에 따라 경도가 감소하는 양상만을 나타내었다. 결국 $(Ti_{1-x}AI_{x})N$ 피막이 열처리 전후에 보아는 기계적 특성의 변화 양상은 열역학적으로 안정한 Wurzite-AlN의 석출에 따른 것으로 AlN 석출상의 크기에 의존하며, 또한 이러한 영향은 $(Ti_{1-x}AI_{x})N$ 피막에 존재하는 AI의 함량이 높고, 초기에 증착된 막의 업자 크기가 작을 수록 클 것으로 여겨진다. 그리고 환경의 의미의 차이에 따라 경관의 미학적 평가가 달라진 것으로 나타났다.corner$적 의도에 의한 경관구성의 일면을 확인할수 있지만 엄밀히 생각하여 보면 이러한 예의 경우도 최락의 총체적인 외형은 마찬가지로 $\ulcorner$순응$\lrcorner$의 범위를 벗어나지 않는다. 그렇기 때문에

• Korean Journal of Cognitive Science
• /
• v.22 no.1
• /
• pp.39-55
• /
• 2011

People tend to overestimate their emotional reactions to events such as physical handicap and buying a new car in the future. Students overestimate their reactions to a future grade as compared to their reactions after receiving the grade. Impact bias refers to people's tendency to overestimate the intensity and the duration of emotional reactions to a future event. The present study explored whether impact bias occurs to risky objects such as nuclear energy, genetically engineered food, and mobile phone. Participants were asked to predict their emotional reactions at three time points, that is, at the present, a week after, and a year after. They predicted their reactions before and after two debiasing tasks. The present study demonstrated a different pattern of impact bias at three time points: A largest bias was observed a week after the present. A defocalism manipulation has eliminated the impact bias whereas an adaptation manipulation has not. Several points were discussed regarding the difference between the previous and the present work.

• Journal of the Korea Society for Simulation
• /
• v.33 no.2
• /
• pp.27-35
• /
• 2024

We aimed to shed light on the issue of perpetuating societal disparities by analyzing the impact of inherent biases present in datasets used for training artificial intelligence models on the predictions generated by Artificial Intelligence(AI). Therefore, to examine the influence of data bias on AI models, we constructed an original dataset containing biases related to gender wage gaps and subsequently created a de-identified dataset. Additionally, by utilizing the decision tree algorithm, we compared the outputs of AI models trained on both the original and de-identified datasets, aiming to analyze how data de-identification affects the biases in the results produced by artificial intelligence models. Through this, our goal was to highlight the significant role of data de-identification not only in safeguarding individual privacy but also in addressing biases within the data.

• Smart Media Journal
• /
• v.13 no.2
• /
• pp.44-51
• /
• 2024

The construction of a stress measurement detaset plays a curcial role in various modern applications. In particular, for the efficient training of artificial intelligence models for stress measurement, it is essential to compare various biases and construct a quality-controlled dataset. In this paper, we propose the construction of a stress measurement dataset with quality management through the comparison of various biases. To achieve this, we introduce strss definitions and measurement tools, the process of building an artificial intelligence stress dataset, strategies to overcome biases for quality improvement, and considerations for stress data collection. Specifically, to manage dataset quality, we discuss various biases such as selection bias, measurement bias, causal bias, confirmation bias, and artificial intelligence bias that may arise during stress data collection. Through this paper, we aim to systematically understand considerations for stress data collection and various biases that may occur during the construction of a stress dataset, contributing to the construction of a dataset with guaranteed quality by overcoming these biases.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2001.05a
• /
• pp.35-37
• /
• 2001

The Al etching was studied employing negative ions generated in the downstream $Cl_2$ plasma. In order to etch the Al film practically on an insulator covered electrode coupled with RF power, reduction of the negative self bias voltage (Vdc) was examined using a magnetic filter which trapped electrons. Addition of $SF_6$ and $H_2$ to a $Cl_2/BCl_3$ mixture reduced significantly Vdc.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.228-228
• /
• 2016

증발산량은 수자원 부존량 평가, 물수지 분석, 지구의 물순환 및 에너지 순환을 이해하기 위해서 알아야할 수문량이나, 이를 산정하기 위하여 단순한 가정을 하거나 경험식을 사용하는 접근에는 신뢰성에 문제가 생긴다. 본 연구에서는 아시아 지역내의 여러 지점에서 에디공분산 시스템을 활용해 플럭스 자료를 구축해놓은 Asia Fluxnet의 자료를 활용해 보완관계법(Complimentary relationship) 기반으로 제한된 기상자료를 이용해 구한 증발산량을 산정하는 방법론들을 평가하였다. Granger and Gary(GG)는 실제 증발산량은 습윤조건의 증발산량의 2배에 잠재 증발산량간의 차와 같다는 보완관계를 수정하여 일반화하고, 잠재 증발산량을 산정하는 경험식을 제시하였다. 이러한 수정된 보완관계식을 활용한 GG 방법론을 활용하여 산정한 증발산량을 측정된 증발산량과 비교한 정확성을 정량화 하기 위해 Average root mean square error (RMSE), mean absolute bias (BIAS), coefficient of determination ($R^2$)과 같은 통계값을 이용하였다. 최종적으로 각 사이트의 기후를 Aridity Index (AI)를 이용하여 분류하였으며 분류된 기후별로 GG 방법론의 적용성을 검토하였다.

• Atmosphere
• /
• v.29 no.4
• /
• pp.463-480
• /
• 2019

This study classified heavy rain types from K-means clustering for the hourly relationship between rainfall intensity and cloud top height over the Korean peninsula, and then examined their statistical characteristics for the period of June~August 2013~2018. Total rainfall amount of warm-type events was 2.65 times larger than that of the cold-type, whereas the lightning frequency divided by total rainfall for the warm-type was only 46% of the cold-type. Typical cold-type cases exhibited high cloud top height around 16 km, large reflectivity in the upper layer, and frequent lightning flashes under convectively unstable condition. Phenomenally, the cold-type cases corresponded to cloud cluster or multi-cell thunderstorms. However, two warm-type cases related to Changma and typhoon were characterized by heavy rainfall due to long duration, relatively low cloud top height and upper-level reflectivity, and the absence of lightning under the convectively neutral and extremely humid conditions. This study further confirmed that the forecast skill of rainfall could be improved by applying correction factor with the overestimation for cold-type and underestimation for warm-type cases in the Local Data Assimilation and Prediction System (LDAPS) operational model (e.g., BIAS score was improved by 5%).

• Convergence Security Journal
• /
• v.24 no.2
• /
• pp.63-70
• /
• 2024

With the development of artificial intelligence technology, our lives are changing in innovative ways, but at the same time, new ethical issues are emerging. In particular, issues of discrimination due to algorithm and data bias, deep fakes, and personal information leakage issues are judged to be social priorities that must be resolved as artificial intelligence services expand. To this end, this paper examines the concept of artificial intelligence and ethical issues from the perspective of artificial intelligence ethics, and includes each country's ethical guidelines, laws, artificial intelligence impact assessment system, artificial intelligence certification system, and the current status of technologies related to artificial intelligence algorithm transparency to prevent this. We would like to examine and suggest institutional improvement measures to strengthen artificial intelligence ethics.