Feb 12, 2015 · We introduce a simple modification for autoencoder neural networks that yields powerful generative models. Our method masks the autoencoder's parameters to respect autoregressive constraints: each input is reconstructed only from previous inputs in a …

We introduce a simple modification for autoencoder neural networks that yields powerful generative models. Our method masks the autoencoder’s parameters to respect autoregressive constraints: each input is recon-structed only from previous inputs in a given or-dering.

Jul 6, 2015 · We introduce a simple modification for autoencoder neural networks that yields powerful generative models. Our method masks the autoencoder's parameters to respect autoregressive constraints: each input is reconstructed only from previous inputs in a …

We introduce a simple modification for autoencoder neural networks that yields powerful generative models. Our method masks the autoencoder’s parameters to respect autoregressive constraints: each input is reconstructed only from previous inputs in a given ordering.

MADE: Masked Autoencoder for Distribution Estimation. Paper on arXiv and at ICML2015. This repository is for the original Theano implementation. If you are looking for a PyTorch implementation, thanks to Andrej Karpathy, you can fine one here.

Masked autoencoder for distribution estimation (MADE) is a well-structured density estimator, which alters a simple autoencoder by setting a set of masks on its connections to satisfy the autoregressive condition.

The resulting Masked Autoencoder Distribution Estimator (MADE) preserves the efficiency of a single pass through a regular autoencoder. Implementation on a GPU is straightforward, making the method scalable.

Oct 25, 2019 · 2. Distribution estimation. If we bully the autoencoders just a bit more, by also blinding them partially, we can actually make them learn $P(x)$, i.e. the distribution of $x$. Germain, Gregor & Larochelle $^{[2]}$, posted their findings in the paper MADE: Masked Autoencoder for Density Estimation. In my opion, they made a really elegant ...

Jul 28, 2020 · This property is formally referred to as “autoregression” (dependence on itself), and is implemented in MADE by introducing masks for the weights of the neural network that is used to estimate the distribution of the variable’s element.

PyTorch implementation of the Masked Autoencoder for Distribution Estimation (MADE) [1]. The implemented model supports random ordering of the inputs for order-agnostic training.