• International Journal of Contents
• /
• 제15권4호
• /
• pp.59-64
• /
• 2019

In recent years, emotion recognition has been an interesting and challenging topic. Compared to facial expressions and speech modality, gesture-based emotion recognition has not received much attention with only a few efforts using traditional hand-crafted methods. These approaches require major computational costs and do not offer many opportunities for improvement as most of the science community is conducting their research based on the deep learning technique. In this paper, we propose an end-to-end deep learning approach for classifying emotions based on bodily gestures. In particular, the informative keyframes are first extracted from raw videos as input for the 3D-CNN deep network. The 3D-CNN exploits the short-term spatiotemporal information of gesture features from selected keyframes, and the convolutional LSTM networks learn the long-term feature from the features results of 3D-CNN. The experimental results on the FABO dataset exceed most of the traditional methods results and achieve state-of-the-art results for the deep learning-based technique for gesture-based emotion recognition.

• The Korean Journal of Physiology and Pharmacology
• /
• 제24권1호
• /
• pp.81-88
• /
• 2020

Regulator of calcineurin 1 (RCAN1) can be induced by an intracellular calcium increase and oxidative stress, which are characteristic features of temporal lobe epilepsy. Thus, we investigated the spatiotemporal expression and cellular localization of RCAN1 protein and mRNA in the mouse hippocampus after pilocarpine-induced status epilepticus (SE). Male C57BL/6 mice were given pilocarpine hydrochloride (280 mg/kg, i.p.) and allowed to develop 2 h of SE. Then the animals were given diazepam (10 mg/kg, i.p.) to stop the seizures and sacrificed at 1, 3, 7, 14, or 28 day after SE. Cresyl violet staining showed that pilocarpine-induced SE resulted in cell death in the CA1 and CA3 subfields of the hippocampus from 3 day after SE. RCAN1 immunoreactivity showed that RCAN1 was mainly expressed in neurons in the shammanipulated hippocampi. At 1 day after SE, RCAN1 expression became detected in hippocampal neuropils. However, RCAN1 signals were markedly enhanced in cells with stellate morphology at 3 and 7 day after SE, which were confirmed to be reactive astrocytes, but not microglia by double immunofluorescence. In addition, realtime reverse transcriptase-polymerase chain reaction showed a significant upregulation of RCAN1 isoform 4 (RCAN1-4) mRNA in the SE-induced hippocampi. Finally, in situ hybridization with immunohistochemistry revealed astrocytic expression of RCAN1-4 after SE. These results demonstrate astrocytic upregulation of RCAN1 and RCAN1-4 in the mouse hippocampus in the acute and subacute phases of epileptogenesis, providing foundational information for the potential role of RCAN1 in reactive astrocytes during epileptogenesis.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.564-566
• /
• 2003

There are no immutable phenomena in reality. A lot of applications are dealing with data characterized by spatial and temporal and/or uncertain features. Currently, there has no any data model accommodating enough those three elements of spatial objects to directly use in application systems. For such reasons, we introduce a fuzzy spatio -temporal data model (FSTDM) and a method of integrating temporal and fuzzy spatial operators in a unified manner to create fuzzy spatio -temporal (FST) operators. With these operators, complex query expression will become concise. Our research is feasible to apply to the management systems and query processor of natural resource data, weather information, graphic information, and so on.

• Wind and Structures
• /
• 제38권4호
• /
• pp.295-307
• /
• 2024

The funnel-shaped vortex structure of tornadoes results in a spatiotemporally varying wind velocity (speed and direction) field. However, very limited full-scale tornado data along the height and radius positions are available to identify and reliably establish a description of complex vortex structure together with the resulting aerodynamic effects on the high-speed train (HST). In this study, the improved delayed detached eddy simulation (IDDES) for flow structures and aerodynamic pressures around an HST under tornado-like winds are conducted to provide high-fidelity computational fluid dynamics (CFD) results. To demonstrate the accuracy of the numerical method adopted in this study, both field observations and wind-tunnel data are utilized to respectively validate the simulated tornado flow fields and HST aerodynamics. Then, the flow structures and aerodynamic pressures (as well as aerodynamic forces and moments) around the HST at various locations within the tornado-like vortex are comprehensively compared to highlight the importance of considering the complex spatiotemporal wind features in the HST-tornado interactions.

• Korean Chemical Engineering Research
• /
• 제62권3호
• /
• pp.274-280
• /
• 2024

기계학습법의 신경망 기술을 이용한 자료분석은 질병 유전자 탐색 및 진단, 신약 개발, 약인성 간 손상 예측 등과 같은 다양한 분야에서 활용되고 있다. 질병 특징 발견을 위한 자료분석은 DNA 정보를 기반으로 이루어질 수 있다. 본 연구에서는 DNA의 분자 정보 중 DNA의 길이와 용액 내 DNA의 길이별 종 개수를 예측하는 신경망을 개발하였다. 겔 전기영동을 통한 기존 방법론의 시간 소요 한계점을 해결하고자, 미세유체역학적 농축 장치의 동역학 자료를 분석 대상으로 하여 실험 분석 과정 중의 시간 소요 문제점을 해결하였다. 동역학 자료를 공간시간 지도로 재구성하여 학습 및 예측에 필요한 계산용량을 낮추었으며, 공간시간 지도에 대한 분석 정확도를 높이기 위해 합성곱 신경망을 활용하였다. 그 결과, 단일 변수 회귀로써의 단일 DNA 길이 예측과 복합 변수 회귀로써의 다종 DNA 길이의 동시 예측 및 이진 분류로써의 DNA 혼합 종 개수 예측을 성공적으로 수행하였다. 추가적으로, 예측 과정 중 발생할 수 있는 예측 편향을 학습 자료 구성 방식을 통한 해결책을 제시하였다. 본 연구를 활용한다면, 광학 측정 자료를 이용하는 액체생검 기반의 세포유리 DNA 분석 및 암 진단 등의 의학 자료 분석을 효과적으로 수행할 수 있을 것이다.

• 한국인터넷방송통신학회논문지
• /
• 제23권3호
• /
• pp.145-151
• /
• 2023

3D-CNN은 시계열 데이터 학습을 위한 딥 러닝 기법 중 하나이다. 이러한 3차원 학습은 많은 매개변수를 생성할 수 있으므로 고성능 기계학습이 필요하거나 학습 속도에 커다란 영향을 미칠 수 있다. 본 연구에서는 손의 동적인 제스처 동작을 시공간적으로 학습할 때, 3D-CNN 모델의 구조적 변화 없이 입력 영상 데이터의 시공간적 변화에 따른 학습 정확성을 분석함으로써, 3D-CNN을 이용한 동적 제스처 학습의 효율성을 높이기 위한 입력 영상 데이터의 최적 조건을 찾고자 한다. 첫 번째로 동적 손 제스처 영상 데이터에서 동적 이미지 프레임의 학습구간을 설정함으로써 제스처 동작간 시간 비율을 조정한다. 둘째로는 클래스간 2차원 교차 상관 분석을 통해 영상 데이터의 이미지 프레임간 유사도를 측정하여 정규화 함으로써 프레임간 평균값을 얻고 학습 정확성을 분석한다. 이러한 분석을 통하여, 동적 손 제스처의 3D-CNN 딥 러닝을 위한 입력 영상 데이터를 효과적으로 선택하는 두 가지 방법을 제안한다. 실험 결과는 영상 데이터 프레임의 학습구간과 클래스간 이미지 프레임간 유사도가 학습 모델의 정확성에 영향을 미칠 수 있음을 보여준다.

• 한국인지과학회:학술대회논문집
• /
• 한국인지과학회 2005년도 춘계학술대회
• /
• pp.23-27
• /
• 2005

언어 특히 의미에 대한 연구와 마음의 구조를 연구하는 인지과학이 어떻게 중요한 관계를 가지고 발전하는가를 살펴보기로 한다. 언어의 구조는 마음의 구조의 일부라고 본 촘스키의 입장에 동조하면서도 의미의 구조에 대해서 소극적인 입장에 서는 그의 입장에서 자유롭게 벗어나 발전하고 있는 의미에 대한 연구를 조명하고 전망하기로 한다. 언어의 의미는 내면적(internal)인 것인가 외부적(external)인 것인가 의미 내용(content)과 맥락(context)의 관계는 어떠한가, 왜 정보구조가 중요한가 등을 점검한다.

• 복식
• /
• 제57권8호
• /
• pp.114-124
• /
• 2007

Contemporary fashion design has been represented as intermedia transcended value and spatiotemporal notion and also had a tendency to concentrate on serial process that materials are transfigured through time rather than existence. These forms related to interaction with time, space and performance as well as compositive genres, hybrid culture, compound gender define as 'serial expression' in this study. The serial expression ran be characterized that system, process, series, enumeration of sequences, depiction of performance, repetition of action in fashion collections and exhibitions of designers. The concept and circumstances made by author as a creator of fashion broaden perceptions of audiences and arouse spectators to participate in the situation as needing immediate attention. The forms of fashion and Conceptual Art in serial expression are analogous and even identical situations represent in fashion collection. Therefore analysis serial forms of art derives formative features: Narrative process, Imitation and Appropriation, Virtual reality and High technology, Hybridism and Convergence. This study suggest a framework to analyze conceptual fashion that give salience to megatrend in contemporary fashion culture on artistic point of view.

• 한국환경농학회지
• /
• 제42권3호
• /
• pp.169-176
• /
• 2023

The emergence and spread of antibiotic-resistant bacteria have been increasing with anthropogenic contamination. Understanding the prevalence and distribution of these resistant bacteria in environments is crucial for effectively managing anthropogenic pollutants. Lake Gyeongpo in the Gangwon Province of South Korea is known for its diverse ecological features and human interactions. The lake is exposed to pollutants from nonpoint sources, including urban areas, agricultural practices, and recreational activities, which can introduce antibiotics and foster antibiotic resistance in bacteria. The present study investigates Lake Gyeongpo as a potential reservoir for antibiotic-resistant bacteria in a natural ecosystem. A total of 203 bacterial isolates were collected from six sampling locations in Lake Gyeongpo during May, July, and November 2022. Most isolates were taxonomically identified as Pseudoalteromonas, Bacillus, Shewanella, and Vibrio spp.; their abundance showed a spatiotemporal distribution. An antibiotic susceptibility test was conducted on 75 isolates using the disk diffusion method with six drugs according to the CLSI guideline; 42 isolates were resistant to one or more antibiotics. Among these, 15 isolates were identified as multidrug resistant bacteria. This finding suggests the potential anthropogenic impact on Lake Gyeongpo and provides valuable insights into the dissemination of antibiotic resistance caused by anthropogenic pollutants.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권1호
• /
• pp.216-238
• /
• 2023

In intelligent transportation systems, traffic management is an important task. The accurate forecasting of traffic characteristics like flow, congestion, and density is still active research because of the non-linear nature and uncertainty of the spatiotemporal data. Inclement weather, such as rain and snow, and other special events such as holidays, accidents, and road closures have a significant impact on driving and the average speed of vehicles on the road, which lowers traffic capacity and causes congestion in a widespread manner. This work designs a model for multivariate short-term traffic congestion prediction using SLSTM_AE-BiLSTM. The proposed design consists of a Bidirectional Long Short Term Memory(BiLSTM) network to predict traffic flow value and a Convolutional Neural network (CNN) model for detecting the congestion status. This model uses spatial static temporal dynamic data. The stacked Long Short Term Memory Autoencoder (SLSTM AE) is used to encode the weather features into a reduced and more informative feature space. BiLSTM model is used to capture the features from the past and present traffic data simultaneously and also to identify the long-term dependencies. It uses the traffic data and encoded weather data to perform the traffic flow prediction. The CNN model is used to predict the recurring congestion status based on the predicted traffic flow value at a particular urban traffic network. In this work, a publicly available Caltrans PEMS dataset with traffic parameters is used. The proposed model generates the congestion prediction with an accuracy rate of 92.74% which is slightly better when compared with other deep learning models for congestion prediction.