… based on convolutional neural network has developed rapidly, and various improved
algorithms and models … detection framework based on convolutional neural network is analyzed …

… classification, we introduce the architecture of classic vision Transformers and compare it with
convolutional neural net… , we discuss Transformer-based models and their roles on feature …

… models include convolutional neural networks (CNN), deep … This step is repeated through
multiple neural layers. Then, … layer, where predictions and masks are made, leading to the …

… models in natural language processing in recent years, researchers have explored the
application of Transformer-based models … for the smooth L1 loss used in the Mask R-CNN paper. …

… Automatic modulation classification based on hyperbolic-tangent-… Automatic
modulation classification using deep residual neural network with masked model- …

… U-Net 和Mask RCNN 的卷积神经网络语义分割模型,在 像素级别分割虫害对番茄的影响,其中
Mask RCNN实现了85.67% 的平均精度,U-Net… region of China using deep learning models[J]. …

… 图3(d)的效果, 被称为频率掩蔽(frequency masking).实验证明该方 … diction of power systems
based on deep residual network and data … learning for automatic modulation classification[J]. …

… bias, the masks for the glass potential were automatically chosen to be all glass area at
least 3 fibril diameters away from the fibril masks edge and the fibril potential mask was a stripe …

… of classification. In this paper, we carried out user guidance based deep learning classification
… We select six popular deep learning models on three real datasets for the evaluation. The …

… 遮罩聲學模型 遮罩聲學模型是受到自然語言處理中的遮罩語言模型(Masked Language
Modeling) 所啟發,其概念如圖2.9所示,藉由隨機屏蔽訊號中的某一部份,讓模型學習如 何從其他未受…