The safety of children has always been an important issue, and several studies have conducted to determine the stress state a child ensure safety. Audio signals biological including heart rate are known be effective for detection. However, collecting those data requires specialized equipment, which is not appropriate constant monitoring children, advanced analysis required accurate In this regard, we propose detection framework utilizes both audio signal collected from wearable devices,...

Abstract Fashion image datasets, in which each fashion has a label indicating its design attributes and styles, have contributed to the achievement of various machine learning techniques industry. Computer vision studies investigated labeling categories (such as items, colors, materials, details, styles) create datasets for supervised learning. Although considerable number been developed, different style classification criteria exist because lack understanding concerning style. Since styles...

Growing competition among manufacturing businesses and the advent of Fourth Industrial Revolution has meant that many countries are conducting various research projects to understand how introduce populate smart factories. Smart factories expected provide a way solving industries’ complex problems, take role in breakthroughs carry on sustainable business. currently introduction stage, so we should follow up majorities check their tendencies. However, smart-factory is an interdisciplinary...

As unmanned aerial vehicles have become popular, the number of accidents caused by an operator’s inattention increased. To prevent such accidents, operator should maintain attention status. However, limited research has been conducted on brain-computer interface (BCI)-based system with alerting module for recovery vehicles. Therefore, we introduce a detection and that prevents vehicle from falling into status using electroencephalogram signal. The proposed consists following three...

With the advance of military technology, number unmanned combat aerial vehicles (UCAVs) has rapidly increased. However, it been reported that accident rate UCAVs is much higher than manned vehicles. One main reasons for high hypovigilance problem which refers to decrease in vigilance levels UCAV operators while maneuvering. In this paper, we propose detection models based on EEG signal minimize occurrences hypovigilance. To enable detection, have applied hidden Markov (HMMs), two are used...

This study aimed to use quantitative methods and deep learning techniques report sportive fashion trends. We collected images from collections of the past decades utilized multi-label graph convolutional network (ML-GCN) model detect explore hybrid styles. Based on literature review, we proposed a theoretical framework investigate The ML-GCN was designed classify five style categories, “street,” “retro,” “sexy,” “modern,” “sporty,” predictive probabilities styles were extracted....

As language editing became an essential process for enhancing the quality of a research manuscript, there are several companies providing manuscript services. In such companies, submitted proofreading is matched with expert through manual process, which costly and often subjective. The major drawback that it almost impossible to consider inherent characteristics as writing style paragraph composition. To this end, we propose recommendation method services based on matrix factorization,...

사용자의 인구통계학적 정보는 추천 시스템과 같은 개인화 서비스 발달에 도움이 되며, 모바일 사용 데이터는 정보 예측에 활용될 수 있다. 특히 텍스트 성별 효과적인 것으로 알려져 있지만, 프라이버시 이슈가 존재하여 그 활용이 제한되고 본 연구에서는 디바이스 내 예측 방법론을 제안하여 데이터를 사용하면서 이슈를 최소화는 동시에 성별을 효과적으로 예측하고자 한다. 우선, 성별에 따른 특징이 반영된 웹문서를 수집하여 각 특징적 단어 집합과 이모티콘 집합을 구성한다. 내에서 데이터와 비교하여 각각 예측하고, 두 결과를 앙상블하여 최종적인 도출한다. 피실험자들의 사용하여 실험을 수행하였으며 제안 방법론의 우수한 성능을 확인하였다.

Mobile camera modules are manufactured by aligning and assembling multiple differently shaped part lenses. Therefore, selecting the lenses to assemble from candidates (called cavities) determining directional angle of each lens for assembly have been important issues maximize production yield. Currently, this process is manually conducted experts at manufacturing site, manual condition optimization carries risk reduced yield increased failure cost as it largely depends on one’s expertise....

Training large language models (LLMs) is known to be challenging because of the huge computational and memory capacity requirements. To address these issues, it common use a cluster GPUs with 3D parallelism, which splits model along data batch, pipeline stage, intra-layer tensor dimensions. However, parallelism produces additional challenge finding optimal number ways on each dimension mapping split onto GPUs. Several previous studies have attempted automatically find configuration, but many...

To train skilled unmanned combat aerial vehicle (UCAV) operators, it is important to establish a real-time training environment where an enemy appropriately responds the action performed by trainee. This can be addressed constructing inference method for behavior of UCAV operator from given simulation log data. Through this method, virtual capable performing actions that are highly likely made actual operator. achieve this, we propose hybrid sequence (HS) kernel-based hierarchical support...

스마트 기기 사용자의 성별 정보는 성공적인 개인화 서비스를 위해 중요하며, 기기로부터 수집된 멀티 모달 로그 데이터는 예측에 중요한 근거가 된다. 하지만 각 데이터의 특성에 따라 다른 방식으로 예측을 수행해야 한다. 따라서 본 연구에서는 발생한 데이터 중 텍스트, 어플리케이션, 가속도 데이터에 기반한 각기 분류기의 예측 결과를 다수결 앙상블하여 최종 성별을 예측하는 기법을 제안한다. 텍스트 데이터를 이용한 분류기는 유출에 의한 사생활 침해 문제를 최소화하기 웹 문서로부터 성별의 특징적 단어 집합을 도출하고 이를 기기로 전송하여 내에서 분류를 수행한다. 어플리케이션 사용자가 실행한 어플리케이션들에 부여하고 높은 비율을 차지하는 성별로 예측한다. 기반 성별에 따른 인스턴스를 학습한 SVM 모델을 사용하여 주어진 분류한다. 자체 제작한 안드로이드 어플리케이션을 통해 실제 제안하는 평가하였으며 그 결과 성능을 보였다. Gender information of a smart device...

With the advent of 4-th industrial revolution, manufacturing companies have increasing interests in realization smart by utilizing their accumulated facilities data. However, most previous research dealt with structured data such as sensor signals, and only a little focused on unstructured text, which actually comprises large portion Therefore, we propose an association rule mining based facility error pattern extraction framework, where text written operators are analyzed. Specifically,...