Aiming at this problem, this paper proposed an unsupervised learning algorithm named Memory-augmented skip-connected deep autoencoder (Mem-SkipAE) for anomaly detection of rocket engines with multi-source data
The first and third Conv2D blocks are connected via a skip link. The skip connection keeps early layer features, improving the model’s learning capabilities, by adding the output of the first and third convolutional blocks. This operation is performed in the adding layer, then the output of ...
Generally, the architecture of a convolutional encoder is regarded as an integration of a feature learning layer, a nonlinear transform layer, a normalization layer, and a feature pooling layer. At the feature learning layer, the input of each unit is connected to the output of the previous ...
Densely connected convolutional networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 21–26 July 2017; pp. 4700–4708. [Google Scholar] Xie, S.; Girshick, R.; Dollár, P.; Tu, Z.; He, K. Aggregated residual transformations for ...
52,PointConv: Deep Convolutional Networks on 3D Point Clouds,https://github.com/DylanWusee/pointconv,,,Thursday,Poster 3.1,99,Wenxuan Wu,"Wenxuan Wu, Zhongang Qi, Li Fuxin" 163,Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoders,https://github.com/edgarschnf...
SCAE-MT [51] designs a stacked convolutional self-coding network model to extract deep features of hyperspectral remote sensing images. Some studies, such as SAE [52], DNR [53], and SDNE [54], use autoencoder to embed the network representation. However, these works do not resolve our ...
SCAE-MT [51] designs a stacked convolutional self-coding network model to extract deep features of hyperspectral remote sensing images. Some studies, such as SAE [52], DNR [53], and SDNE [54], use autoencoder to embed the network representation. However, these works do not resolve our ...
SCAE-MT [51] designs a stacked convolutional self-coding network model to extract deep features of hyperspectral remote sensing images. Some studies, such as SAE [52], DNR [53], and SDNE [54], use autoencoder to embed the network representation. However, these works do not resolve our ...