search

zhangyuxin621 / AMSL

Adaptive Memory Networks with Self-supervised Learning for Unsupervised Anomaly Detection (AMSL) -open source
46 stars 2 forks source link

Why is class ```Memory_global``` not called after it is defined in #train.py#? #1

Open KevinStrive88 opened 2 years ago

KevinStrive88 commented 2 years ago

Hi~

  1. I was wondering in https://github.com/zhangyuxin621/AMSL/blob/b72a88618168248ef23760537d9a47a426dabe7d/train.py#L203-L205 class Memory_global should have been called on here after it was defined rather than Memory_local , yes?

  2. Also, the definitions of class Memory_local and Memory_global are almost identical, at the same time, (if Q1 is no exception) APIs of https://github.com/zhangyuxin621/AMSL/blob/b72a88618168248ef23760537d9a47a426dabe7d/train.py#L208-L214 and https://github.com/zhangyuxin621/AMSL/blob/b72a88618168248ef23760537d9a47a426dabe7d/train.py#L220-L226 are also identical, so I don't know that "global memory module to learn the common representations contained in all transformations, and the local memory module to learn augmentation-specific representations for each transformation" in paper.

Thank you for your attention.