16 citations · 17 across the 4 of their papers we have counts for
6 papers
Target Layer Regularization for Continual Learning Using Cramer-Wold Generator
Marcin Mazur, Łukasz Pustelnik, Szymon Knop +2
We propose an effective regularization strategy (CW-TaLaR) for solving continual learning problems. It uses a penalizing term expressed by the Cramer-Wold distance between two prob…
Generative models with kernel distance in data space
Szymon Knop, Marcin Mazur, Przemysław Spurek +2
Generative models dealing with modeling a~joint data distribution are generally either autoencoder or GAN based. Both have their pros and cons, generating blurry images or being un…
LocoGAN -- Locally Convolutional GAN
Łukasz Struski, Szymon Knop, Jacek Tabor +2
In the paper we construct a fully convolutional GAN model: LocoGAN, which latent space is given by noise-like images of possibly different resolutions. The learning is local, i.e.…
One-element Batch Training by Moving Window
Przemysław Spurek, Szymon Knop, Jacek Tabor +2
Several deep models, esp. the generative, compare the samples from two distributions (e.g. WAE like AutoEncoder models, set-processing deep networks, etc) in their cost functions.…
Sliced generative models
Szymon Knop, Marcin Mazur, Jacek Tabor +2
In this paper we discuss a class of AutoEncoder based generative models based on one dimensional sliced approach. The idea is based on the reduction of the discrimination between s…
Cramer-Wold AutoEncoder
Szymon Knop, Jacek Tabor, Przemysław Spurek +3
We propose a new generative model, Cramer-Wold Autoencoder (CWAE). Following WAE, we directly encourage normality of the latent space. Our paper uses also the recent idea from Slic…