regression_RobustAutoencoder not implemented

[mlix11]

I read your paper and I liked it a lot. So I want to try the things out on my own.

But I have one Issue, where do you get the tflearn.regression_RobustAutoencoder function, because it seams it is not implemented in tflearn. Is it self-implemented or on a branch/fork of tflearn?

For example in section_5.1_anomaly_detection_CIFAR_10_AE.py:

mue = 0.1
net = tflearn.regression_RobustAutoencoder(net,mue,hidden_layer,decoder, optimizer='adam', learning_rate=0.001,
                        loss='rPCA_autoencoderLoss_FobsquareLoss', metric=None,name="vanilla_autoencoder")

model = tflearn.DNN(net, tensorboard_verbose=0)

Thank you!

[maryam089]

Well I have also read this paper and liked the idea so when i tried to implement the codes in python, I found the similar issue that tflearn.regression_RobustAutoencoder function, is not implemented in tflearn. How to deal with this issue ? @raghavchalapathy Are you able to resolve that issue you mentioned above ? @loose11

[raghavchalapathy]

Its is self implemented essentially only the parameters essential for defining loss function is added

def rPCA_autoencoderLoss_FobsquareLoss(y_pred, y_true,mue,hidden_layer,decode_layer): """ Mean Square Loss.

Arguments:
    y_pred: `Tensor` of `float` type. Predicted values.
    y_true: `Tensor` of `float` type. Targets (labels).

"""
with tf.name_scope("MeanSquare"):
    return tf.reduce_mean(tf.square(y_pred - y_true))

[maryam089]

Kindly update it in your code to remove that bug. Cheers thanks :) @raghavchalapathy

[tayden]

@raghavchalapathy Can you please give more code for the loss function used?

[AdamRain]

Same question...... Can you please kindly upload the regression_RobustAutoencoder function?

[raghavchalapathy]

Hello All

I have updated the source code repository and used Keras/ tensorflow You may see the latest repository here ( This includes RCAE: Robust convolution autoencoder) https://github.com/raghavchalapathy/oc-nn/blob/master/README.md from src.models.RCAE import RCAE_AD

with regards Raghav

[jianzhuwang]

Is there a complete code for restaurant dataset? I did not see the detailed code for that dataset, and I failed to know the network structure and other information. How to obtain the foreground? Also, I have the same question as posted above.

Many thanks for your help.