http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Adaptive Attention Annotation Model: Optimizing the Prediction Path through Dependency Fusion
( Fangxin Wang ),( Jie Liu ),( Shuwu Zhang ),( Guixuan Zhang ),( Yang Zheng ),( Xiaoqian Li ),( Wei Liang ),( Yuejun Li ) 한국인터넷정보학회 2019 KSII Transactions on Internet and Information Syst Vol.13 No.9
Previous methods build image annotation model by leveraging three basic dependencies: relations between image and label (image/label), between images (image/image) and between labels (label/label). Even though plenty of researches show that multiple dependencies can work jointly to improve annotation performance, different dependencies actually do not "work jointly" in their diagram, whose performance is largely depending on the result predicted by image/label section. To address this problem, we propose the adaptive attention annotation model (AAAM) to associate these dependencies with the prediction path, which is composed of a series of labels (tags) in the order they are detected. In particular, we optimize the prediction path by detecting the relevant labels from the easy-to-detect to the hard-to-detect, which are found using Binary Cross-Entropy (BCE) and Triplet Margin (TM) losses, respectively. Besides, in order to capture the inforamtion of each label, instead of explicitly extracting regional featutres, we propose the self-attention machanism to implicitly enhance the relevant region and restrain those irrelevant. To validate the effective of the model, we conduct experiments on three well-known public datasets, COCO 2014, IAPR TC-12 and NUSWIDE, and achieve better performance than the state-of-the-art methods.
Wang Jinhuo,Han Yang,Ge Xiaohong,Qi Zhengbing,Zhao Jun,Wang Rongwen,Wu Huawei,Han Taiping,Sun Shaoxun,Wang Hui,Lin Jia,Liu Yuejun,Kong Xiangsong,Chen Qiming,Zeng Xiangxu 한국유변학회 2023 Korea-Australia rheology journal Vol.35 No.2
Optimisation design of composite structures requires an accurate predictive model for forming behaviour. The simulation process contains a number of model parameters which include transverse and longitudinal viscosities of continuous fibrereinforced viscous composites, fundamental to predicting the shear rheology. Shearing the unidirectional composite along the fibre direction gives a measure of the longitudinal viscosity (LV), whilst shearing across or transverse to the fibre direction gives a measure of the transverse viscosity (TV). Numerous experimental work was conducted in the past to measure these two viscosities for various materials. However, conflicting measurements by different test methods were obtained and these apparent discrepancies had not yet been systematically investigated in any single study. This paper reviews previous work on characterisation techniques to further understand the cause of such discrepancy, and hence to improve measurement accuracy, which would benefit future work on theoretical modelling of the composite viscosities and optimisation simulation of composites forming. Some important findings, such as effects of resin-rich areas, contributory factors of elastic effects, non-Newtonian behaviour for composites with Newtonian matrix, aspect ratio and end effects of test samples, geometry effects of fibres and fibre rearrangement during shearing, existence of a mathematical relationship between LV and TV and necessary benchmarking exercise using Newtonian matrix composites, were summarised.