mlsanityData to 7 Oct 2026

Soft-IntroVAE: Analyzing and Improving the Introspective Variational Autoencoder

arXiv 2012.13253 ↗First posted 24 Dec 2020cs.LG cs.AI cs.CVcited by 23

Not on the map. Why?

Abstract (arXiv)

The recently introduced introspective variational autoencoder (IntroVAE) exhibits outstanding image generations, and allows for amortized inference using an image encoder. The main idea in IntroVAE is to train a VAE adversarially, using the VAE encoder to discriminate between generated and real data samples. However, the original IntroVAE loss function relied on a particular hinge-loss formulation that is very hard to stabilize in practice, and its theoretical convergence analysis ignored important terms in the loss. In this work, we take a step towards better understanding of the IntroVAE model, its practical implementation, and its applications. We propose the Soft-IntroVAE, a modified IntroVAE that replaces the hinge-loss terms with a smooth exponential loss on generated samples. This change significantly improves training stability, and also enables theoretical analysis of the complete algorithm. Interestingly, we show that the IntroVAE converges to a distribution that minimizes a sum of KL distance from the data distribution and an entropy term. We discuss the implications of this result, and demonstrate that it induces competitive image generation and reconstruction. Finally, we describe two applications of Soft-IntroVAE to unsupervised image translation and out-of-distribution detection, and demonstrate compelling results. Code and additional information is available on the project website -- https://taldatech.github.io/soft-intro-vae-web

Key facts. Cited by 23 papers in the map (data through 7 Oct 2026). First posted on arXiv 24 Dec 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 5; 2023: 6; 2024: 7; 2025: 2; 2026: 3.

23 papers.

Sort:

By area:

  1. 2023AutoSplice: A Text-prompt Manipulated Image Dataset for Media Forensicscited by 46
  2. 2022Unsupervised Image Representation Learning with Deep Latent Particlescited by 19
  3. 2023Reversing the Abnormal: Pseudo-Healthy Generative Networks for Anomaly Detectioncited by 18
  4. 2022DPD-fVAE: Synthetic Data Generation Using Federated Variational Autoencoders With Differentially-Private Decodercited by 14
  5. 2024Generative AI Enables Medical Image Segmentation in Ultra Low-Data Regimescited by 8
  6. 2022What do we learn? Debunking the Myth of Unsupervised Outlier Detectioncited by 6
  7. 2024Towards Universal Unsupervised Anomaly Detection in Medical Imagingcited by 6
  8. 2025Revisiting Likelihood-Based Out-of-Distribution Detection by Modeling Representationscited by 6
  9. 2022Heterogeneous Face Recognition via Face Synthesis with Identity-Attribute Disentanglementcited by 4
  10. 2022AS-IntroVAE: Adversarial Similarity Distance Makes Robust IntroVAEcited by 4
  11. 2023Fine-Tuning Generative Models as an Inference Method for Robotic Taskscited by 4
  12. 2024Towards Controllable Time Series Generationcited by 4
11 more
  1. 2024Improving Deep Learning-based Automatic Cranial Defect Reconstruction by Heavy Data Augmentation: From Image Registration to Latent Diffusion Modelscited by 2
  2. 2024Multimodal ELBO with Diffusion Decoderscited by 2
  3. 2024Interpretable Representation Learning of Cardiac MRI via Attribute Regularizationcited by 1
  4. 2024Selective Test-Time Adaptation for Unsupervised Anomaly Detection using Neural Implicit Representationscited by 1
  5. 2026Quantum Generative Diffusion Model for Real-World Time Seriescited by 1
  6. 2023Soft-IntroVAE for Continuous Latent space Image Super-Resolutioncited by 0
  7. 2023Attribute Regularized Soft Introspective VAE: Towards Cardiac Attribute Regularization Through MRI Domainscited by 0
  8. 2023Attribute Regularized Soft Introspective Variational Autoencoder for Interpretable Cardiac Disease Classificationcited by 0
  9. 2025A DeepONet joint Neural Tangent Kernel Hybrid Framework for Physics-Informed Inverse Source Problems and Robust Image Reconstructioncited by 0
  10. 2026Multimodal normative modeling in Alzheimers Disease with introspective variational autoencoderscited by 0
  11. 2026Unsupervised Anomaly Detection for Image Dataset Quality Assurance in Multi-Center Breast MRIcited by 0