Adversarial Symmetric Variational Autoencoder

By Yunchen Pu, Weiyao Wang, Ricardo Henao, Liqun Chen, Zhe Gan, Chunyuan Li, Lawrence Carin. 2017

81 citations recorded by the source corpus.

A new form of variational autoencoder (VAE) is developed, in which the joint distribution of data and codes is considered in two (symmetric) forms: ($i$) from observed data fed through the encoder to yield codes, and ($ii$) from latent codes drawn from a simple prior and propagated through the decoder to manifest data. Lower bounds are learned for marginal log-likelihood fits observed data and latent codes. When learning with the variational bound, one seeks to minimize the symmetric Kullback-Leibler divergence of joint density functions from ($i$) and ($ii$), while simultaneously seeking to maximize the two marginal log-likelihoods. To facilitate learning, a new form of adversarial training is developed. An extensive set of experiments is performed, in which we demonstrate state-of-the-art data reconstruction and generation on several image benchmark datasets.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Generative Models, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Generative Models

In Probabilistic & Bayesian Methods

Citations in the corpus

A practical tutorial on autoencoders for nonlinear feature fusion: Taxonomy, models, software and guidelines (2018, methodology): “Other generative models based on similar principles are Variational Recurrent AEs [85], PixelGAN AEs [86] and Adversarial Symmetric Variational AEs [87].”
Learning Representations by Maximizing Mutual Information in Variational Autoencoders (2019, neutral): “For example [16], [17] replace the KL divergence with the Jensen-Shannon divergence which enables the data and latent codes to be treated in a symmetric manner.”
Learning latent representations across multiple data domains using Lifelong VAEGAN (2020, background): “presentations, which is specic to the VAE. Adversarial learning is performed in order to match either the data distribution [36], the latent variables distribution [42], or their joint distributions [10,15,16,37,43,49,61]. These methods perform well only when trained on a single dataset and their pe”
ALICE: Towards Understanding Adversarial Learning for Joint Distribution Matching (2017, background): “Efforts have been made to simultaneously learn an efficient bidirectional model that can produce high-quality samples for both the latent and data spaces [3, 4, 8, 9, 10, 11].”
Distribution Matching in Variational Inference (2018, background): “This leads to the fusion of VAEs and GANs, of which we develop a deeper understanding by examining existing and new VAE-GAN hybrids [25, 30, 32, 38, 41, 44].”
Machine Learning in Drug Discovery: A Review (2021, methodology): “In many situations, the drug produces invalid SMILES syntax, so as to defeat this issue, grammar variational autoencoder was utilized for developing SMILES syntax with more effectiveness (Pu et al. 2017) (Fig.”
Triangle Generative Adversarial Networks (2017, result): “ that uses supervised learning [27] for joint-distribution matching, and variational autoencoders for semi-supervised learning [28, 29]. Lastly, our work is also closely related to the recent work of [30, 31, 32], which treats one of the domains as latent variables. 4 Experiments We present results ”
Lifelong Teacher-Student Network Learning (2021, background): “VAE [30] adopts a simple and fixed prior which can not represent exactly the true posterior p(z|x) [49], leading to vague generation results [50].”
Deep Mixture Generative Autoencoders (2021, methodology): “Importance sampling [31], [32] is also used in VAEs for improving the quality of the inference.”
Generative Adversarial Networks (GANs) (2020, background): “[31] proposed a new class of statistical divergence, Relaxed Wasserstein (RW) divergence for the large-scale computations.”
Symmetric Variational Autoencoder and Connections to Adversarial Learning (2017, methodology): “ore of ALICE, in which the model fails to generate images. Table 2: Unsupervised Inception Score on CIFAR-10 Model IS ALI [Dumoulin et al., 2017] 5.34 .05 DCGAN [Radford et al., 2016] 6.16 .07 ASVAE [Pu et al., 2017] 6.89 .05 WGAN-GP 6.56 .05 WGAN-GP ResNet [Ishaan Gulrajani, 2017] 7.86 .07 sVAE (ou”
A Lagrangian Perspective on Latent Variable Generative Models (2018, methodology): “Latent variable generative models are designed to accomplish a wide variety of tasks in computer vision (Radford et al., 2015; Kuleshov & Ermon, 2017), natural language processing (Yang et al., 2017), reinforcement learning (Li et al., 2017b), compressed sensing Dhar et al. (2018),etc. Prominent exa”
Unsupervised Multi-Target Domain Adaptation: An Information Theoretic Approach (2018, unknown): “p(d) = 1 N ∑N i=1 δ(d − di )) as in the case of variational autoencoders (VAE) [15], [16], p(zs |x), p(zp|x) as deterministic functions of x as p(zs |x) = δ ( zs − Es(x; θs) ) and p(zp|x) = δ ( zp − E p(x; θp) ) , and the variational distributions q(y|zs), q(x|z) and, q(d|z) as q(y|zs) = SoftMax(C(z”
JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets (2018, background): “Repre-sentative work includes the AAE (Makhzani et al., 2015), VAE-GAN (Larsen et al., 2015), AVB (Mescheder et al., 2017), AS-VAE (Pu et al., 2017), SVAE (Chen et al., 2018), etc .”
Towards Better Representations with Deep/Bayesian Learning (2018, unknown): “…the variants of VAEs and GANs (Pu et al., 2015; Stevens et al., 2017; Fan et al., 2016a; Pu et al., 2016a,b; Chen et al., 2018b; Li et al., 2018a; Pu et al., 2017b; Chen et al., 2018a; Xu et al., 2017; Wang et al., 2017b; Pu et al., 2018b; Yuan et al., 2017, 2018; Chapfuwa et al., 2018), sigmoid…”