• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2021년도 하계학술대회
• /
• pp.335-338
• /
• 2021

본 논문에서는 기존 AHDRNet에 channel attention 기법을 적용했을 때 성능에 어떠한 변화가 있는지를 평가하였다. 기존 모델의 병합 망에 존재하는 DRDB(Dilated Residual Dense Block) 사이, 그리고 DRDB 내의 확장된 합성곱 레이어 (dilated convolutional layer) 뒤에 또다른 합성곱 레이어를 추가하는 방식으로 channel attention 기법을 적용하였다. 데이터셋은 Kalantari의 데이터셋을 사용하였으며, PSNR(Peak Signal-to-Noise Ratio)로 비교해본 결과 기존의 AHDRNet의 PSNR은 42.1656이며, 제안된 모델의 PSNR은 42.8135로 더 높아진 것을 확인하였다.

• 한국산업정보학회논문지
• /
• 제29권2호
• /
• pp.35-45
• /
• 2024

본 논문에서는 CNN과 Attention 기법을 통한 깊이 영상의 화면 내 예측 방법을 제안한다. 제안하는 방법을 통해 예측하고자 하는 블록 내 화소마다 참조 화소를 선택할 수 있도록 한다. CNN을 통해 예측 블록의 상단과 좌단에서 각각 수직방향과 수평 방향의 공간적 특징을 검출한다. 두 공간적 특징은 예측블록과 참조 화소들에 대한 특징을 예측하기 위해 각각 특징차원과 공간적 차원으로 병합된다. Attention을 통해 예측 블록과 참조 화소간의 상관성을 입력된 공간적 특징을 통해 예측한다. Attention을 통해 예측된 상관성은 CNN 레이어를 통해 화소 도메인으로 복원되어 블록 내 화소 값이 예측된다. 제안된 방법이 VVC의 인트라 모드에 추가되었을 때 화면 예측 오차가 평균 5.8% 감소하였다.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 추계학술대회
• /
• pp.34-36
• /
• 2022

기존 텍스트 감성 분석 모델에서는 일반적으로 전체 텍스트를 직접 모델링하고, 텍스트 내용 간의 계층적 관계를 덜 고려한다. 그러나 감정분석의 구현에서는 많은 텍스트가 여러 감정으로 뒤섞여 있다. 전체의 의미론적 모델링을 직접 수행하면 감성분석 모델의 판단 난도가 높아져 혼합 감정 문장의 분류에 적용하기 어려울 수 있다. 따라서 본 논문에서는 텍스트 계층을 고려한 감성 분석 모델 BHGCN을 제안한다. 이 모델에서는 BERT의 각 레이어의 숨겨진 상태의 출력이 노드로 사용되며, 상위 레이어와 하위 레이어 사이에 직접 연결이 이루어져 의미 계층이 있는 그래프 네트워크를 구축한다. BHGCN 모델은 계층별 의미론에 주의를 기울일 뿐만 아니라 계층적 관계에도 주의를 기울이기 때문에 혼합 감성 분류 작업을 처리하는 데 적합하다. 본 논문에서는 비교 실험을 통해 제안하는 BHGCN 모델이 명백한 경쟁 우위를 보인다는 것을 입증하였다.

• 대한간호학회지
• /
• 제26권2호
• /
• pp.281-289
• /
• 1996

Neural networks have recently attracted considerable attention in the field of classification and other areas. The purpose of this study was to demonstrate an experiment using back-propagation neural network model applied to nursing diagnosis. The network's structure has three layers ; one input layer for representing signs and symptoms and one output layer for nursing diagnosis as well as one hidden layer. The first prototype of a nursing diagnosis system for patients with stomach cancer was developed with 254 nodes for the input layer and 20 nodes for the output layer of 20 nursing diagnoses, by utilizing learning data set collected from 118 patients with stomach cancer. It showed a hitting ratio of .93 when the model was developed with 20,000 times of learning, 6 nodes of hidden layer, 0.5 of momentum and 0.5 of learning coefficient. The system was primarily designed to be an aid in the clinical reasoning process. It was intended to simplify the use of nursing diagnoses for clinical practitioners. In order to validate the developed model, a set of test data from 20 patients with stomach cancer was applied to the diagnosis system. The data for 17 patients were concurrent with the result produced from the nursing diagnosis system which shows the hitting ratio of 85％. Future research is needed to develop a system with more nursing diagnoses and an evaluation process, and to expand the system to be applicable to other groups of patients.

• 한국통신학회논문지
• /
• 제31권12A호
• /
• pp.1189-1197
• /
• 2006

Ultra-wideband(UWB) is a killer technology for short-range wireless communications. In the past, most of the UWB research focused on physical layer but the unique characteristics of UWB make it different to design the upper layer protocols than conventional narrow band systems. Cross-layer protocols have received high attention for UWB networks. In this paper, we investigate the performance of two physical layer schemes: Time Hopping Binary Pulse Position Modulation(TH-BPPM) and Time Hopping Binary Pulse Amplitude Modulation (TH-BPAM) with proposed private MAC protocol for UWB ad-hoc networks. From pulse level to packet level simulation is done in network simulator ns-2 with realistic network environments for varying traffic load, mobility and network density. Our simulation result shows TH-BPAM outperforms TH-BPPM in high traffic load, mobility and dense network cases but in a low traffic load case identical performance is achieved.

• 소성∙가공
• /
• 제27권5호
• /
• pp.314-322
• /
• 2018

AlSi12 is a heat-resistant aluminum alloy that is lightweight, corrosion-resistant, machinable and attracting attention as a functional material in aerospace and automotive industries. For that reason, AlSi12 powder has been used for high performance parts through 3D printing technology. The purpose of this study is to observe deposition characteristics of AlSi12 powder in a direct energy deposition (DED) process (one of the metal 3D printing technologies). In this study, deposition characteristics were investigated according to various process parameters such as laser power, powder feed rate, scan speed, and slicing layer thickness. In the single track deposition experiment, an irregular bead shape and balling or humping of molten metal were formed below a laser power of 1,000 W, and the good-shaped bead was obtained at 1.0 g/min powder feed rate. Similar results were observed in multi-layer deposition. Observation of deposited height after multi-layer deposition revealed that over-deposition occurred at all conditions. To prevent over-deposition, slicing layer thickness was experimentally determined at given conditions. From these results, this study presented practical conditions for good surface quality and accurate geometry of deposits.

• 한국공간구조학회논문집
• /
• 제16권4호
• /
• pp.45-52
• /
• 2016

The single layer latticed domes have attracted many designers and researchers's attention all of the world, because these structures as spatial structure are of great advantage in not only mechanical rationality but also function, fabrication, construction and economic aspect. But single layer latticed domes are apt to occur the unstable phenomena that are called "buckling" because of the lack of strength of members, instability of structural shape, etc. In the case of latticed dome, there are several types of buckling mode such as overall buckling, local buckling, and member buckling according to the shape of dome, section type of member, the size of member, junction's condition of member and so on. There are many methods to increase the buckling strength of the single layer latticed dome, that is, with the change of geometrical shape of dome, the reinforcement of buckled member, etc. Therefore, the purpose of this study is to verify the reinforcement effect of buckled member when designers reinforce the buckled member to increase the buckling strength of single layer latticed dome with 3-way grid.

• Composites Research
• /
• 제35권3호
• /
• pp.121-126
• /
• 2022

Owing to advantages of polymeric materials for hydrogen tank liner like light-weight property and high specific strength, polymer based composites have gained much attention. Despite of many benefits, polymeric materials for fuel cell tank cause problems which is critical to applications as low gas barrier property, and poor processability when adding fillers. For these reasons, improving gas barrier property of polymer composites is required to study for expanding application fields. This work presents impermeable polymer nanocomposites by introducing thin barrier coating using layer by layer (LBL) deposition method. Also, bi-layered and quad-layered nanocomposites were fabricated and compared for identifying relationship between deposition step and gas barrier property. Reduction in gas permeability was observed without interrupting mechanical property and processability. It is discussed that proper coating conditions were suggested when different coating materials and deposition steps were applied. We investigated morphology, gas barrier property and mechanical properties of fabricated nanocomposites by FE-SEM, Oxygen permeation analyzer, UTM, respectively. In addition, we revealed the mechanism of barrier performance of LBL coating using materials which have high aspect ratio.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2015년도 제49회 하계 정기학술대회 초록집
• /
• pp.61.2-61.2
• /
• 2015

In the past decades, green energy, such as solar energy, wind power, hydropower, biomass energy, geothermal energy, and so on, has been widely investigated and developed to solve energy shortage. Recently, organic solar cells have attracted much attention, because they have many advantages, including low-cost, flexibility, light weight, and easy fabrication [1-3]. Organic solar cells are as a potential candidate of the next generation solar cells. In this abstract, to improve the power conversion efficiency and the stability, the inverted polymer solar cells with various structures were developed [4-6]. The novel cell structures included the P3HT:PCBM inverted polymer solar cells with AZO nanorods array, with pentacene-doped active layer, and with extra P3HT interfacial layer and PCBM interfacial layer. These three difference structures could respectively improve the performance of the P3HT:PCBM inverted polymer solar cells. For the inverted polymer solar cells with AZO nanorods array as the electronic transportation layer, by using the nanorod structure, the improvement of carrier collection and carrier extraction capabilities could be expected due to an increase in contact area between the nanorod array and the active layer. For the inverted polymer solar cells with pentacene-doped active layer, the hole-electron mobility in the active layer could be balanced by doping pentacene contents. The active layer with the balanced hole-electron mobility could reduce the carrier recombination in the active layers to enhance the photocurrent of the resulting inverted polymer solar cells. For the inverted polymer solar cells with extra P3HT and PCBM interfacial layers, the extra PCBM and P3HT interfacial layers could respectively improve the electron transport and hole transport. The extra PCBM interfacial layer served another function was that led more P3HT moving to the top side of the absorption layer, which reduced the non-continuous pathways of P3HT. It indicated that the recombination centers could be further reduced in the absorption layer. The extra P3HT interfacial layer could let the hole be more easily transported to the MoO3 hole transport layer. The high performance of the novel P3HT:PCBM inverted polymer solar cells with various structures were obtained.

• Journal of Information Processing Systems
• /
• 제17권4호
• /
• pp.754-771
• /
• 2021

In the task of continuous dimension emotion recognition, the parts that highlight the emotional expression are not the same in each mode, and the influences of different modes on the emotional state is also different. Therefore, this paper studies the fusion of the two most important modes in emotional recognition (voice and visual expression), and proposes a two-mode dual-modal emotion recognition method combined with the attention mechanism of the improved AlexNet network. After a simple preprocessing of the audio signal and the video signal, respectively, the first step is to use the prior knowledge to realize the extraction of audio characteristics. Then, facial expression features are extracted by the improved AlexNet network. Finally, the multimodal attention mechanism is used to fuse facial expression features and audio features, and the improved loss function is used to optimize the modal missing problem, so as to improve the robustness of the model and the performance of emotion recognition. The experimental results show that the concordance coefficient of the proposed model in the two dimensions of arousal and valence (concordance correlation coefficient) were 0.729 and 0.718, respectively, which are superior to several comparative algorithms.