Data Science

Questions tagged [autoencoder]

Ask Question

Autoencoders are a type of neural network that learns a useful encoding for data in an unsupervised manner.

196 questions
Newest
Active
Bountied
Unanswered
Filter by
Sorted by
Tagged with
0
votes
0answers
2 views

How to correctly mirror / inverse an encoder architecture for the decoder?

For the purpose of autoencoders, It seems to be common practice to build the decoder architecture in a way so that it mirrors the encoder architecture. However, there seem to be 2 approaches of doing ...
1
vote
1answer
23 views

Dense output from neural network

I would like to create a loss function that encourages the output of the embedding of an autoencoder to be dense. I don't have an explicit condition for how density is defined, but one option would be ...
0
votes
1answer
17 views

What Non-linearities are best in Denoising RNN Autoencoders and where should the go?

I’m employing a denoising RNN autoencoder for a project relating to motion capture data. This is my first time using auto encoder architectures and I was just wondering what non-linearities should be ...
0
votes
0answers
7 views

Finding differences in smooth embedding spaces?

I'm trying to find statistically significant differences between embeddings (by an autoencoder) of different datasets. At first I tried to embed them separately, and cluster them. However, the ...
1
vote
1answer
32 views

How to extract features from the encoded layer of an autoencoder?

I have done some research on autoencoders, and I have come to understand that they can also be used for feature extraction (see this question on this site as an example). Most of the examples out ...
0
votes
0answers
6 views

How do I use Conv2DTranspose to create my decoder?

I am constructing VAE of input 80x60. I have my encoder below but I am troubles making the decoder as it does not conform to 80 x 60. Here is my decoder. Below is the code for constructing the ...
2
votes
0answers
15 views

Are Denoising Variational Autoencoders deterministic?

I have a pretty good understanding of regular autoencoders and, to a certain extent, of variational autoencoders, where the latent representation is forced to follow specific probabilistic ...
1
vote
1answer
31 views

How to use Variational Autoencoder's μ and σ with user-generated z?

My understanding of VAE is that unlike Autoencoders, it does not directly give you a discrete encoding (latent code vectors n-dim) instead, it gives you both mu and sigma (n- dim mean vectors and n-...
1
vote
1answer
59 views

Steps on how to use autoencoders to reduce dimensions

I have a dataset that contains text columns. I have used tf-idf to convert those text columns to numerical columns. I want to reduce the dimension of the dataset since tf-idf creates a multitude of ...
1
vote
0answers
9 views

How to estimate total correlation KL[q(z)||Πjq(zj)] of VAE after training (useful for latents disentanglement evaluation)

FactorVAE and β-TCVAE both use total correlation (TC) batch estimation for their objectives. Where TC is: $$ KL\bigl( q(z)||\prod\nolimits_{j} q(z_{j})\bigr) $$ both estimates are applied to $q(z|x)$...
0
votes
0answers
14 views

Do I need to train a separate DeepFake model for every input person?

I would like to create a deep fake model of a specific person (we will call him Steve). I would then like to be able to upload a video of any random person and swap their face with Steve's. So far I ...
0
votes
0answers
14 views

What's the difference between encoder/decoder and autoencoder? [duplicate]

What's the difference between the encoder-decoder model and autoencoder model? Or are they the same thing? Thanks!
0
votes
0answers
11 views

chatbot encoder/decoder: why do we need to use chatbot answer as the decoder inputs?

I am looking into the chatbot tutorial at: https://medium.com/predict/creating-a-chatbot-from-scratch-using-keras-and-tensorflow-59e8fc76be79 It uses sequence to sequence model with encoder/decoder ...
0
votes
0answers
16 views

Can I use/modify an Autoencoder to handle missing data?

I am about to implement an Autoencoder to detect anomalies. Therefore, e.g., in my test set, there is a situation where the data stream broke for some days. This results in a lack of data and should ...
2
votes
0answers
57 views

Results are too good.. what is wrong? How to predict correctly?

I am about to evaluate a neural network and want to check whether the predictions make sense. The variables: ...
0
votes
0answers
16 views

normalization/standardization of input/output of autoencoder and Gaussian Process

I have two machine learning algorithms that deal with time series data. My data consist of 1500 time series, each of 500 time components. The first machine learning algorithm is an autoencoder, ...
0
votes
0answers
15 views

Looking for a good text autoencoder model (for text reconstruction)

I am looking for a simple and good model that can learn to encode and reconstruct text sentences to use it in some downstream task. I tried a tensorflow seq2seq model here, but it doesn't do ...
0
votes
0answers
11 views

Removing layers from a convolutional encoding-decoding network

I've been reading this paper on Style-transfer (Universal Style Transfer via Feature Transforms): A crucial part of the algorithm (Section 3.3) uses a pre-trained VGG-19 network as an encoder to ...
0
votes
1answer
21 views

Validation generator in Autoencoder returning NaN

I am trying to build a fairly simple autoencoder using Keras on the OpenImages dataset. Here is the architecture of the ae: ...
0
votes
1answer
37 views

Keras - Autoencoder different from Encoder + Decoder

I build a CNN 1d Autoencoder in Keras, following the advice in this SO question, where Encoder and Decoder are separated. My goal is to re-use the decoder, once the Autoencoder has been trained. The ...
0
votes
0answers
12 views

VAE latent space dimension with Weibull PDFs as input data

I am currently using a Variational Autoencoder to reconstruct PDFs of Weibull distributions with varying parameters (sampled uniformly from a given parameter span). The PDFs are generated by sampling ...
0
votes
1answer
24 views

why do we operate with graphical models in VAE, if there are no probabilites involved?

In the variational autoencoder, I often see graphical models e.g. $P(X|Z)$ for the decoder, but when I looked at code, I don't see any random variables, I see just deterministic network, with ...
0
votes
1answer
11 views

how to design sigma(std) vector to have values above > 0 in VAE autoencoder

I am trying to understand how to design mu and sigma vector in VAE autoencoder, as we know std value can not be larger than 0. If I use linear function, the value can be lower than zero, so I think ...
2
votes
1answer
184 views

Autoencoders for the compression of time series

I am trying to use autoencoder (simple, convolutional, LSTM) to compress time series. Here are the models I tried. Simple autoencoder: ...
0
votes
0answers
68 views

How can we use resnet in autoencoders?

I am creating an unsupervised classifier model, for which i want to use resnet 50 on a custom database and used the top layers of resnet as start point of my autoencoder. How to proceed with this. ...
0
votes
0answers
18 views

How to measure performance of a denoising autoencoder?

Is there a way to theoretically estimate (avoiding brute force) what is the minimum signal-to-noise ratio required for a denoising autoencoder (like a regular filter)? What I want to know specifically ...
0
votes
0answers
19 views

Combining multiple loss functions in the right way

I am trying to build an (variational) Autoencoder to generate fake but representative data from a generic data set with a couple of numeric and categorical columns. So far I have built functions that ...
0
votes
1answer
31 views

Autoencoder or layer-based dimensionality reduction?

I have a few TB of wide data. I want to reduce the number of features in my dataset before feeding my dataset into a classification model... or should I not? Obviously, I will want to try both ...
0
votes
1answer
14 views

How to design n-dimensional feature descriptor similar as the input image?

I am re-writing the H-Net code in Keras for cross-domain image similarity. The network architecture is described in the attached paper. I wrote the encoder and decoder parts but unable to get similar ...
1
vote
1answer
332 views

Autoencoder: using cosine distance as loss function

I'm trying to train an autoencoder (in PyTorch) to reconstruct gene profiles. At the moment I'm using the Mean Squared Error (MSE) loss for training: the model is not overfitting and both the training ...
1
vote
0answers
15 views

Autoencoders linear latent space

According to "Linear interpolation in latent space" in https://hackernoon.com/latent-space-visualization-deep-learning-bits-2-bd09a46920df and others, the latent space representation of an autoencoder ...
1
vote
0answers
37 views

Disadvantages of Mean Squared Error? [closed]

I'm using mean squared error as reconstruction error for my autoencoder. The dataset is ECG (time series) and model is conv1d. I assumed MSE as the best option for reconstruction error, but it's ...
8
votes
1answer
678 views

How does an encoder-decoder network work?

Let's say I trained an encoder-decoder network on a cat dataset using reconstruction error as loss function. The network is fully trained and the decoder is able to reconstruct good cat images. Now ...
0
votes
1answer
165 views

Fully Convolutional Variational Autoencoder

I would like to make a neural network which uses black and white images as input and outputs a colored version of it. The important thing in that process is that the size of the images must stay the ...
0
votes
0answers
24 views

What is the meaning of constraint two neural networks against each other?

I have two identical autoencoders with the same number of layers and parameters. I would like to find the similarity between two images from different domains such as an image captured from the camera ...
1
vote
1answer
69 views

Which algorithm can be used to reduce dimension of multiple time series?

In my dataset, a data point is essentially a Time series of 6 feature over a year per month so in all, it results in 6*12=72 features. I need to find class outliers so I perform dimensionality ...
0
votes
1answer
26 views

Comparison of performance of autoencoder with PCA

I am running PCA and autoencoder (2 hidden layer with relu) on a data. Both PCA and autoencoder give similar accuracy of the order 50%. I have tried different variations of autoencoder: changing ...
1
vote
0answers
8 views

How to calculate a location prior in CT's based on atlas segmentation?

I'm working myself through the paper "Anatomical Priors in Convolutional Networks for Unsupervised Biomedical Segmentation" (https://arxiv.org/abs/1903.03148) I lack to understand how the atlas ...
1
vote
0answers
16 views

How can one be assured that generative models are not memorizing dataset, and that they will generate an unique image outside of dataset?

If all GAN can do is capture the probability distribution of the dataset, then shouldn't they be similar to handing out images from the dataset? How can we verify that the images that they generate ...
0
votes
0answers
8 views

Post Training Sampling from a unit Gaussian?

During the training of a VAE, we sample Z from N(mu, sigma) ( = Q(Z|X) ) and then feed it to the decoder for reconstruction. Why do we, then, feed Z ~ N(0,1) to the decoder after the training is ...
0
votes
0answers
88 views

variational autoencoder doesn't fit with KL loss Keras

Hi I am learning the variational autoencoders and Idesigned my VAE as follow: ...
0
votes
0answers
24 views

Developing an encoder/decoder for image modification

On the project I am currently working on, my goal is to train a neural network to convert images of circles to ellipses in a way that models convolution/blurring in real imaging processes. What ...
1
vote
0answers
76 views

Optimizing parameters for CNN autoencoder based on training and validation loss

I have designed an autoencoder with a encoder and decoder consiting of 2D convolutational layers (the input are 40'000 2D images). I train the autoencoder using adam optimizer. The autoencoders has ...
0
votes
0answers
105 views

How do we overfit a CNN AutoEncoder for anomaly detection?

I have been working on an anomaly detection problem in which I need to treat the images of "street" as an anomaly. The images of "glaciers" will be treated as not-anomaly. The autoencoder which ...
0
votes
0answers
26 views

Keras / TensorFlow 2.0: are UpSampling2D() layers the inverse of Max-Pooling?

I am trying to build a Variational Autoencoder for image data. As I employ MaxPool2D() in the encoding part, I need the reverse it in the decoder. Are ...
0
votes
0answers
13 views

Is my Autoencoder architecture correct?

I am dipping my toe in deep learning, especially autoencoders. I am trying to reconstruct my dataset. The dimension of my dataset is 13000x3. Can the following architecture work? ...
0
votes
0answers
42 views

Using smote for data augmentation of a data set which has no dependent variable

I am trying to use the reconstruction error obtained using an auto encoder to do novelty detection. My data set is of size (4500,55)(Note: this data doesn't have any abnormalities.When an auto-encoder ...
1
vote
1answer
228 views

How do I get similarity with autoencoders

I have build an autoencoder to extract from a very high dimensional (200 dimensions) space a smaller but significant representation (16 dimensions). Now that I have these "encoded" vectors, I would ...
0
votes
0answers
57 views

Replication of Andrew Ng's Sparse Autoencoder

for the past three days I have been trying to replicate the results presented in Andrew Ng's sparse autoencoding lecture (https://web.stanford.edu/class/cs294a/sparseAutoencoder.pdf) however I have ...
0
votes
1answer
59 views

Autoencoder feature extraction without validation set?

I plan to use autoencoder for feature extraction, then use the latent vector for clustering. My autoencoder performs very very well on my training set (loss small and reconstructed image look very ...

1 2 3 4