search

jchibane / if-net

Implicit Feature Network (IF-Net) - Codebase
305 stars 59 forks source link

Why does there exist a displacement #6

Closed pengsida closed 4 years ago

pengsida commented 4 years ago

I noticed that there are some displacements in the networks. For ShapeNet32Vox, the displacement is 0.035: https://github.com/jchibane/if-net/blob/master/models/local_model.py#L40 For other networks, the displacement is 0.0722: https://github.com/jchibane/if-net/blob/master/models/local_model.py#L124 Could you tell me the reason?

jchibane commented 4 years ago

Hi,

please check the supplementary (paragraph: "Continuous Feature Extraction") for a detailed description: https://virtualhumans.mpi-inf.mpg.de/papers/chibane20ifnet/chibane20ifnet_supp.pdf

and let me know if you have further questions!

Best, Julian

pengsida commented 4 years ago

Ok, I get it. Thanks for your response. Have a good day!

Leerw commented 3 years ago

Hi,

please check the supplementary (paragraph: "Continuous Feature Extraction") for a detailed description: https://virtualhumans.mpi-inf.mpg.de/papers/chibane20ifnet/chibane20ifnet_supp.pdf

and let me know if you have further questions!

Best, Julian

Thanks for your impressive work, I just want to know from what reason you select 0.0722 for d rather than any other number, is there some consideration besides the supp and paper?

Leerw commented 3 years ago

@jchibane

jchibane commented 3 years ago

I used a displacement that looked intuitive for me and measured the distance, it was this value 0.0722 and I took it. Therefore, this is obviously a hyper-parameter that can be optimized. Also, it is a displacement useful for the scale of data used in our experiments. If someone is using other data, for example 10x the size, also the displacement value should be scaled accordingly.

Best, Julian

Leerw commented 3 years ago

I used a displacement that looked intuitive for me and measured the distance, it was this value 0.0722 and I took it. Therefore, this is obviously a hyper-parameter that can be optimized. Also, it is a displacement useful for the scale of data used in our experiments. If someone is using other data, for example 10x the size, also the displacement value should be scaled accordingly.

Best, Julian

Thanks for your answer.