zurutech / anomaly-toolbox Goto Github PK

• Anomaly Toolbox version: Latest
• Python version: 3.8
• Operating System: Linux

Description

Isn't the generator loss supposed to be computed based on the discriminator output of the regenerated image? Therefore shouldn't it be d_gex instead of g_ex as the input to bce_g_loss? Below is from trainers ganomaly.py

Discriminator on the reconstructed real data g_ex
d_gex, _ = self.discriminator(inputs=g_ex, training=True)

Encode the reconstructed real data g_ex
e_gex = self.encoder(g_ex, training=True)

Discriminator Loss
d_loss = self._minmax(d_x_features, d_gex_features)
d_loss = self._minmax(d_x, d_gex)

Generator Loss
adversarial_loss = losses.adversarial_loss_fm(d_f_x, d_f_x_hat)
bce_g_loss = generator_bce(g_ex, from_logits=True)

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

• Anomaly Toolbox version: 0.1.0
• Python version: 3.9
• Operating System: Arch Linux

Description

Bug on the core anomaly-toolbox app. The app does not start because of the copyright description that shadows the #!/usr/bin/env python3 shebang line header.

What I Did

anomaly-box.py --run-all --hps-path path/to/config/hparams.json --dataset MNIST

I found this repository from your study "A Survey on GANs for Anomaly Detection". Wanting to reimplement the GANomaly model, I've been looking at your implementation. Upon comparison to the original study and code, it seems that the implementation of GANomaly is incorrect.

It seems that you have only one instance of the encoder, whereas the original study has two instances. Moreover, they say the following in the original GANomaly paper:

The second sub-network is the encoder network E that compresses the image ˆx that is reconstructed by the network G. With different parametrization, it has the same architectural details as G_E.

This seems like an issue that could compromise the study results.

is it possible to describe the implementation of the code?

i run this
pip install anomaly-toolbox

in colab it show this error

ERROR: Could not find a version that satisfies the requirement anomaly-toolbox (from versions: none)
ERROR: No matching distribution found for anomaly-toolbox

• Anomaly Toolbox version: 0.1.0
• Python version: 3.9
• Operating System: Arch Linux

Description

In the main README section, the examples are misleading. A correction is needed in the --run-all part of the command.