PixelCNN Models with Auxiliary Variables for Natural Image Modeling

By Alexander Kolesnikov, Christoph H. Lampert. 2016

53 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Generative Models

Citations in the corpus

Generating Diverse High-Fidelity Images with VQ-VAE-2 (2019, background): “Some of these issues are alleviated by introducing inductive biases such as multi-scale [38, 39, 29, 22] or by modeling the dominant bit planes in an image [18, 17].”
Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images (2020, methodology): “This can be viewed as an learned parallel multiscale generation method, unlike the handcrafted approaches of Kolesnikov & Lampert (2017) Additionally, we found that on all datasets we tested, very deep VAEs used roughly 30% fewer parameters than the PixelCNN (Table 2).”
The challenge of realistic music generation: modelling raw audio at scale (2018, background): “Beyond raw audio generation, several generative models that capture large-scale structure have been proposed for images [2, 3, 12, 27], dialogue [40], 3D shapes [30] and symbolic music [37].”
Analyzing Inverse Problems with Invertible Neural Networks (2018, methodology): “Besides variants of the original architectures [43, 2, 44], the most common realizations use auto-regressive flows, where the density is decomposed according to the Bayesian chain rule [23, 17, 12, 31, 29, 25, 38, 45].”
A Survey on Generative Adversarial Networks: Variants, Applications, and Training (2020, background): “-VAE-2 first encodes the input image into two different latent maps, one for local information, and another for global details of the input image. Stage-II of VQ-VAE-2 imposes an Autoregressive model [14] on both levels of latent space to compress the (B b(c)) (B d(Y)) (B efgf#TAB#(h)) Distinguishin”
On GANs and GMMs (2018, methodology): “Recent advances in deep learning, have also enabled unsupervised learning of full sized images using various models: Variational Auto Encoders [21, 17], PixelCNN [40, 39, 23, 38], Normalizing Flow [9, 8] and Flow GAN [14].”
Controllable Semantic Image Inpainting (2018, methodology): “(z) = E p train(x)[q ˚(z jx)]. Exact computation is intractable, so we estimate the negative entropy term using a mini-batch importance sampling estimator proposed by Chen et al. [2], Esmaeili et al. [6]: E q ˚(z) h logq ˚(z) i ˇ 1 M XM m=1 log h 1 N h q (z jx m)+ N 1 M 1 X 06= q ˚(z jx m0) ii ; (5)”
UViM: A Unified Modeling Approach for Vision with Learned Guiding Codes (2022, background): “Recently, there have been signiﬁcant advances in the modeling of complex structured outputs in the context of language generation and (conditional) image generation: autoregressive models [49, 41, 25], GANs [13], VAE [22], VQVAE [51], diffusion models [45, 18].”