mlsanityData to 7 Oct 2026

Disentangled and Distilled Encoder for Out-of-Distribution Reasoning with Rademacher Guarantees

arXiv 2512.10522 ↗First posted 11 Dec 2025cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Recently, the disentangled latent space of a variational autoencoder (VAE) has been used to reason about multi-label out-of-distribution (OOD) test samples that are derived from different distributions than training samples. Disentangled latent space means having one-to-many maps between latent dimensions and generative factors or important characteristics of an image. This paper proposes a disentangled distilled encoder (DDE) framework to decrease the OOD reasoner size for deployment on resource-constrained devices while preserving disentanglement. DDE formalizes student-teacher distillation for model compression as a constrained optimization problem while preserving disentanglement with disentanglement constraints. Theoretical guarantees for disentanglement during distillation based on Rademacher complexity are established. The approach is evaluated empirically by deploying the compressed model on an NVIDIA

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 23 of its 41 references that are in the map. First posted on arXiv 11 Dec 2025.

Builds on

23 of 41 references.

References · 23

  1. 2021Convex optimizationcited by 2,479
  2. 2017CARLA: An Open Urban Driving Simulatorcited by 2,431
  3. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,362
  4. 2020Knowledge Distillation: A Surveycited by 1,578
  5. 2021High-Performance Large-Scale Image Recognition Without Normalizationcited by 359
  6. 2022Domain Generalization by Mutual-Information Regularization with Pre-trained Modelscited by 133
  7. 2021Model-Based Domain Generalizationcited by 103
  8. 2018The Singular Values of Convolutional Layerscited by 95
  9. 2020Controlling generative models with continuous factors of variationscited by 93
  10. 2021Towards Principled Disentanglement for Domain Generalizationcited by 91
  11. 2021Characterizing signal propagation to close the performance gap in unnormalized ResNetscited by 87
  12. 2019Weakly Supervised Disentanglement with Guaranteescited by 82
11 more
  1. 2022Factorizing Knowledge in Neural Networkscited by 59
  2. 2019Generalization bounds for deep convolutional neural networkscited by 58
  3. 2020Real-time Out-of-distribution Detection in Learning-Enabled Cyber-Physical Systemscited by 46
  4. 2021Constrained Learning with Non-Convex Lossescited by 42
  5. 2020Variational Autoencoders with Riemannian Brownian Motion Priorscited by 34
  6. 2019Hypothesis Set Stability and Generalizationcited by 23
  7. 2018q-Space Novelty Detection with Variational Autoencoderscited by 22
  8. 2021Efficient Out-of-Distribution Detection Using Latent Space of β-VAE for Cyber-Physical Systemscited by 19
  9. 2022Towards Practical Control of Singular Values of Convolutional Layerscited by 9
  10. 2021DisUnknown: Distilling Unknown Factors for Disentanglement Learningcited by 6
  11. 2022Out of Distribution Reasoning by Weakly-Supervised Disentangled Logic Variational Autoencodercited by 2

Built on by

Not cited yet.