Questions tagged [autoencoder]
Autoencoders are a type of neural network that learns a useful encoding for data in an unsupervised manner.
296
questions
0
votes
0answers
12 views
LSTM decoder with 2d's input
I am developing a CNN-LSTM autoencoder in pytorch to predict time sequences.
The CNN input is a RGB image:
RGB image => tensor[Batch size= 4, channel = 3,width= 256, height=256]
and the output is ...
0
votes
0answers
19 views
Conditional variational autoencoder: Feeding labeled MNIST to encoder with Keras
I am looking for a code implementation of a CVAE using MNIST in Keras.
I found this Youtube video:
https://youtu.be/8wrLjnQ7EWQ
that does VAE, but I am not sure how do I convert this and make encoder ...
0
votes
0answers
10 views
How do I prevent infinite variances/standard deviations in my variational autoencoder?
I am working on a project with a variational autoencoder (VAE).
The problem I have is that the encoder part of VAE is producing large log variances, which leads to even larger standard deviations, ...
0
votes
0answers
13 views
single layer autoencoder performing a lot worse than pca
I am trying to use a single layer autoencoder with linear activation function to perform dimensionality reduction on a dataset before clustering.
The data consists of 5000 samples with 2000 features ...
1
vote
0answers
10 views
If the input to the autoencoder is normalized, do we need to use sigmoid on the last layer?
According to: https://stackoverflow.com/questions/65307833/why-is-the-decoder-in-an-autoencoder-uses-a-sigmoid-on-the-last-layer
The last layer activation function contains sigmoid in order to the ...
1
vote
1answer
28 views
Explainability and Autoencoders
suppose I have an autoencoder as a two-stack LSTM that takes in sequences of $n$ features of some length $m$.
Let's say that the dimension of my encoding vector is $k$, so the architecture is of the ...
5
votes
1answer
54 views
Why autoencoders use binary_crossentropy loss and not mean squared error?
Keras autoencoders examples:
(https://blog.keras.io/building-autoencoders-in-keras.html) use binary_crossentropy (BCE) as loss function.
Why they use ...
2
votes
2answers
48 views
How to interpreter Binary Cross Entropy loss function?
I saw some examples of Autoencoders (on images) which use sigmoid as output layer and BinaryCrossentropy as loss function.
The ...
2
votes
0answers
23 views
Convolutional autoencoder - why keras example is asymmetry model?
I'm looking on keras convolutional autoencoder example, and confused with the model structure:
...
0
votes
1answer
21 views
How to walk forward an LSTM autoencoder by n timesteps?
I am able to fit this autoencoder to my sequence in order to reconstruct it.
However, how would I be able to walk this forward 3 timesteps to get [[11.0], [12.0], [13.0]]?
...
1
vote
0answers
49 views
Autoencoder: Size of out_backprop doesn't match computed
This question was asked before and non of the answered worked for, I have the code
...
1
vote
1answer
26 views
Anomaly detection using LSTM AutoEncoder
Having a sequence of 10 days of sensors events, and a true / false label, specifying if the sensor triggered an alert within the 10 days duration:
sensor_id
timestamp
feature_1
feature_2
...
0
votes
0answers
15 views
Anomaly Detection: LSTM Autoencoder Zero Reconstruction Loss on Anomalies
I am using an LSTM Autoencoder model for time series anomaly detection. None of the anomalies get flagged because the reconstruction loss comes out to be zero for all data points on the clearly ...
1
vote
0answers
38 views
How to derive Evidence Lower Bound in the paper “Zero-Shot Text-to-Image Generation”?
Can someone share the derivation of Evidence Lower Bound in this paper ?
Zero-Shot Text-to-Image Generation
The overall procedure can be viewed as maximizing the
evidence lower bound (ELB) (Kingma &...
0
votes
0answers
17 views
Can we use only one autoencoder for removing noise and converting a greyscale image to color image?
I am new to deep learning. out of curiosity, I have doubt about autoencoders. I want to construct a greyscale image by removing noise in it and converting it into a colour image. We can do this by ...
0
votes
0answers
22 views
Autoencoder: good loss values while fitting, but the actual performance is bad
I'm implementing a 1D convolutional autoencoder to reduce the dimensionality of an array. Here is my architecture:
...
0
votes
0answers
29 views
GAN model with different optimization functions
Building GAN model contains the following steps:
Build generator model, and choose ...
1
vote
1answer
27 views
Autoencoder give wrong results
I have studied Autoencoders and tried to implement a simple one.
I have built a model with one hidden layer.
I ran it with the MNIST digits dataset and plotted the digits before the Autoencoder and ...
1
vote
1answer
39 views
Latent variable graph in Variational Autoencoder
I followed this Keras documentation guide about Auto Encoders.
At the end of the documentation there is the graph of the latent variable z:
But I can not understand and how to interpret the plot, ...
1
vote
1answer
22 views
Autoencoder train and test accuracy shooting to 99% on few epochs
I am trying to train an autoencoder for dimensionality reduction and hopefully for anomaly detection. My data specifications are as follows.
Unlabeled
1 million data points
9 features
I am trying to ...
0
votes
0answers
14 views
Trouble understanding the odd behavior of a trained Autoencoder
I have built an Autoencoder model (aim is to do anomaly detection). Due to large difference in scale of different features, I have transformed the data into [0,1] interval using ...
0
votes
0answers
4 views
VAE will always results in somewhat different latent vectors for same input?
Hey I was wondering if my intuition is correct that for the same input in a VAE we will get a slightly different vector every time we feed it through the network, due to the random sampling operation?
0
votes
0answers
14 views
Fitting input data into Gaussian distribution
I'm currently reading papers on Variational Autoencoders (VAE). According to this article (http://proceedings.mlr.press/v95/guo18a/guo18a.pdf):
By fitting the input data sample x(i) into the Gaussian ...
0
votes
1answer
27 views
Keras: Very high loss for Autoencoder
I am trying to implement an autoencoder for prediction of multiple labels using Keras. This is a snippet:
...
0
votes
0answers
12 views
Why is the posterior chosen to be normal in variational autoencoder?
Is there any reason for choosing the posterior $q(z|x)$ as normal distribution in variational autoencoder? or is it just for convenience?
0
votes
0answers
28 views
Input pipeline with an autoencoder and tf.data
I am using an autoencoder to detect anomalies in dataset of network traffic. The dataset is a csv file, and is loaded and preprocessed with pandas (encoded categorical features with pandas.get_dummies(...
0
votes
0answers
10 views
Unused bottleneck neurons in autoencoder
I'm using an autoencoder to compress categorical data, by a factor of about 20x.
For this part of my data set, I have roughly 3500 variables, so my final bottleneck size is 180. I'm getting pretty ...
0
votes
0answers
8 views
How to minimalize noise from autoencoder output in audio reconstruction
I am training an autoencoder that takes an audio sample and outputs a variation of the input. The network is working as expected by the final output contains noise and it's not very clear. I am ...
0
votes
1answer
31 views
Help needed in interpreting the loss, val_loss vs epoch plots for an autoencoder training?
I am training a variational autoencoder and I am getting a loss-plot as follows:
Rigt after epoch 224, val-loss overtakes train-loss and sort of getting bigger but at an extremely slow pace as you ...
2
votes
1answer
128 views
Improve Convolutional Autoencoder
I just built a Convolutional Autoencoder to try to reconstruct a time series with shape (4000, 10, 30). This is the code, I used a batch size of 32, but I think it ...
1
vote
2answers
111 views
Autenocoder and anomaly detection task
I'm trying to create an autoencoder for the anomaly detection task, but I'm noticing that even if it performs very well on the training set, it starts to stop recreating half of the test set. I tried ...
0
votes
0answers
40 views
How to apply variational autoencoder for oversampling with cross-validation?
Currently, I have an imbalanced data set with proportions 84% and 16%. I wanna use VAE as oversampling method and I want to determine the best proportions of data that results in better metrics. Also, ...
0
votes
1answer
57 views
Understanding time series anomaly detection using Autoencoder
I'm studying how to detect anomalies in the time series using an Autoeconder. In particular, I'm following the guide posted in the Keras website, but I don't understand why they are creating and how ...
0
votes
0answers
22 views
Dimensionality reduction for geometric curves using an autoencoder - what is wrong?
I am trying to play with toy models in order to study autoencoders. In particular, I want to do dimensionality reduction for simple geometric curves in 3D.
First, I take a toroidal helix.
...
0
votes
0answers
17 views
What is the correct terminology for neural network architectures that expand in their internal layers? (The opposite of a “bottleneck”)
I am aware that an auto-encoder contracts in its hidden layers to form a bottleneck.
In contrast to this, is there a good name for the kind of cell, block or architecture that expands in its internal ...
0
votes
0answers
32 views
Convolutional AE always overfitting time series - what’s wrong?
I've build a CAE for anomaly detection in time series, but it is always overfitting. I've tried data augmentation, short/long inputvector, dropout rates... I don't know what I'm doing wrong, may be ...
1
vote
0answers
32 views
How does an autoencoder 'fill in the blanks' in the context of a recommender system?
My understanding is that an autoencoder takes an input, produces a lower dimensional representation of the input, which should explain the original features in the dataset, and then reconstructs the ...
1
vote
0answers
40 views
Is there any problem with the following Python+TF+Keras code for a custom loss function and network?
I am trying to code a custom loss function for variational autoencoder. I am not using mse for reconstruction loss since I am not learning p(x|z) ~ N(mu,I). Instead ...
4
votes
0answers
431 views
1D CNN Variational Autoencoder Conv1D Size
I am trying to create a 1D variational autoencoder to take in a 931x1 vector as input, but I have been having trouble with two things:
Getting the output size of 931, since maxpooling and upsampling ...
0
votes
1answer
358 views
Autoencoder implementation using ImageDataGenerator
I'm using the concept demonstrated in this paper. Their training data consists of "GOOD" images and "BAD" images. They train the AE using "BAD" images (X) to make it ...
0
votes
0answers
30 views
Use convolutional variational autoencoders for time series prediction
I want to use convolutional variational autoencoders for time series prediction. For example, here is the dimension of my data.
...
0
votes
0answers
58 views
How to convert the reconstruction error from a variational autoencoder into a probability score?
I am faced with a dilemma while working with a variational autoencoder model. I am thinking of converting the reconstruction error into a probability score (aim is to use to determine anomalies ...
0
votes
1answer
70 views
Dimensionality reduction convolutional autoencoders
I don't understand how convolutional autoencoders achieve dimensionality reduction. For FFNN based autoencoder, the reduction is easy to understand: the input layer has N neurons, and the hidden ones ...
0
votes
0answers
20 views
Is Dynamic Time Warping a good loss function for a time series auto-encoder?
I've been trying to implement a multivariate time-series auto encoder. I thought DTW could be a good loss function but my implementation is still too slow.
Anyone has some ideas of pros and cons of ...
0
votes
0answers
13 views
Reversing data through a tensorflow feature_column.embedding_column
I am building an variational autoencoder in Tensorflow, and one of my columns has object data. The data is too sparse (on the order of 2^16 possible values) to use one-hot encoding, it's not ordinal ...
0
votes
0answers
29 views
How to get all data from an ImageDataGenerator?
I want to get the result of all the processing that the generators performed on my images. For example:
x,y = generator.next()
but for all data, not just one batch (...
0
votes
0answers
10 views
Can I replace an embedding layer with hot encoding layer?
I'd like to build an autoencoder for a movie recommendation. My data looks like below
user, movie, rating
1, 123, 3
1, 333, 5
2, 34, 3
...
I have seen people using ...
0
votes
0answers
15 views
How to do feature reduction for a log-linear regression model
I'm building a log-linear regression model and I have 18 different variables in my model. 13 out of 18 variables I'm using are hot-encoded variables for holiday, e.g. showing which holiday it is. I ...
0
votes
0answers
18 views
Reconstructing an audio signal from its Mel-scale spectrogram using an autoencoder
I'm looking for some papers/references that attempted to reconstruct audio signals from their Mel-scale spectrograms using an autoencoder or other neural network.
I am thinking of training the ...
0
votes
0answers
14 views
Single real number evaluation metric for VAE for a regression problem
I've set up a VAE for a regression problem.
...
