-
I have a question that I don't quite understand. In the MultiHeadAttentionLayer module, when calculating the distance gate, you need to normalize full_g.ndata['coors'] first and then use DGL's u_sub_v…
-
could you please share how to turn off equivariance, I would like to apply this model for my use case, where I also don't need equivariance
_Originally posted by @Ala1s in https://github.com/ehooge…
Ala1s updated
1 month ago
-
Hi thanks for your sharing.
However, I am curious about your initialization of point embeddings. You used a Conv1D to initialize the point feature, where the input channel is 3, the coordinates of e…
-
The feature maps have dimensions of eccentricity and angle, rather than horizontal and vertical positions. So we do not have translation equivariance. Also, the image statistics may be different in th…
-
Hi! Thank you for this work!
Following paragraph 5.1 of your paper, I am trying to check the equivariance of the convolution layers. I have a question about Figure 3. Is the error in 10^-6 magnitud…
-
I was trying to use ico cnn to do some experiments about SO(3) rotations.But I was wondering how to prove this network is equivariant to the rotations.On one hand I rotated the signals first and sent …
-
Hi @AliaksandrSiarohin, thanks for share the code, but i am confused about the equivariance_value implemented in model.py
## Value loss part
if self.loss_weights['equivaria…
-
Great work! I'm trying to train the model on Voxceleb2 dataset and the generated frames are not as good as those generated by your provided pretrained model. During training the generator GAN loss is …
-
On your project page and in your youtube video, there are still parts of the image that are clearly not equivariant to rotation; visually speaking they swing all the way from black to white.
Have y…
-
## 一言でいうと
3Dオブジェクトの畳み込みで、回転などをしても同じ物体だと認識させることを目的とした研究。手法としては回転の特徴を得るためにフィルタ自体を回転させ畳み込みを行なっている。回転は24が定義されているが、等価な回転が存在するためそれについては計算を省略している。
### 論文リンク
https://arxiv.org/abs/1804.04458
### 著…