Timeseries anomaly detection using an Autoencoder - Using Conv1DTranspose about keras-io HOT 7 CLOSED

Sure, note that I adjusted the layers a little but the general notion should be clear:

model = keras.Sequential(
    [
        layers.Input(shape=(x_train.shape[1], x_train.shape[2])),
        layers.Conv1D(
            filters=32, kernel_size=8, padding="same", strides=3, activation="relu"
        ),
        layers.Dropout(rate=0.2),
        layers.Conv1D(
            filters=16, kernel_size=8, padding="same", strides=3, activation="relu"
        ),
        layers.Dropout(rate=0.2),
        layers.Conv1D(
            filters=8, kernel_size=8, padding="same", strides=3, activation="relu"
        ),
        layers.Conv1D(filters=8, kernel_size=8, padding="same", activation="relu"),
        layers.UpSampling1D(size=3),
        layers.Dropout(rate=0.2),        
        layers.Conv1D(filters=16, kernel_size=8, padding="same", activation="relu"),
        layers.UpSampling1D(size=3),
        layers.Dropout(rate=0.2),
        layers.Conv1D(filters=32, kernel_size=8, padding="same", activation="relu"),
        layers.UpSampling1D(size=3),
        layers.Conv1D(filters=1, kernel_size=8, padding="same"),
    ]
)

from keras-io.

Hi,

I had the same problem, turns out that a Conv1DTranspose() layer, if I understood it correctly, is very similar to a
Conv1D() layer followed by a UpSampling1D() layer.

Hope this helps.

from keras-io.

Hi,

I had the same problem, turns out that a Conv1DTranspose() layer, if I understood it correctly, is very similar to a
Conv1D() layer followed by a UpSampling1D() layer.

Hope this helps.

Would you be able to show the code on the workaround?

from keras-io.

You can also define a Conv1DTranspose from scratch like @Guitaricet writes here: tensorflow/tensorflow#30309 (comment)

from keras-io.

Hi @CanK93 ,

We are in the process of going through backlogs. I found this is not an issue now. I have tried executing the Tutorial with TF2.12v and executes fine with Conv1DTranspose layer also.Please refer the attached gist.

Please cross check and feel free to close the issue. Thanks!

from keras-io.

This issue is stale because it has been open for 14 days with no activity. It will be closed if no further activity occurs. Thank you.

from keras-io.

This issue was closed because it has been inactive for 28 days. Please reopen if you'd like to work on this further.

from keras-io.