Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning latent representations across multiple data domains using Lifelong VAEGAN
Fei Ye, Adrian G. Bors
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add inference model with VAE and GAN hybrid (L-VAEGAN) instead of VAE-based LGM | lifelong generative replay · MNIST (M-F) | reconstruction MSE: 7.18 -> 4.75 | Improved |
| use L-VAEGAN instead of LGM | lifelong generative replay · MNIST (M-F) | classification accuracy: 91.26 -> 92.53 | Improved |
| use L-VAEGAN instead of VAEGAN with GRM | lifelong generative replay · Fashion (M-F) | reconstruction MSE: 17.03 -> 17.44 | Hurt |
| use L-VAEGAN instead of LGM | lifelong generative replay · Fashion (F-M) | reconstruction MSE: 18.83 -> 13.16 | Improved |
| use L-VAEGAN instead of VAEGAN | lifelong generative replay · Fashion (F-M) | classification accuracy: 64.98 -> 66.97 | Improved |
| add generative replay samples (GRM) during lifelong training (ablation) | semi-supervised L-VAEGAN · MNIST to Fashion accuracy | accuracy | Improved |
| use L-VAEGAN instead of EWC | lifelong supervised · MNIST (M-F) | accuracy: 37.7 -> 98.76 | Improved |
| use L-VAEGAN instead of EWC | lifelong supervised · MNIST (F-M) | accuracy: 99.12 -> 98.77 | Hurt |
| use L-VAEGAN instead of Transfer | lifelong supervised · SVHN (M-S) | accuracy: 89.90 -> 81.97 | Hurt |
| use L-VAEGAN instead of LGAN | lifelong supervised · SVHN (S-M) | accuracy: 76.76 -> 80.99 | Improved |
| use L-VAEGAN instead of LGAN | semi-supervised lifelong · MNIST error | classification error: 5.46 -> 4.34 | Improved |
| use lifelong L-VAEGAN instead of non-lifelong M1+M2 | semi-supervised · MNIST error | classification error: 2.40 -> 4.34 | Hurt |
Closest papers by meaning
- Lifelong Generative Modelling Using Dynamic Expansion Graph Model (AAAI 2022 · similarity 0.88)
- Lifelong GAN: Continual Learning for Conditional Image Generation (ICCV 2019 · similarity 0.88)
- Task-Free Continual Generation and Representation Learning via Dynamic Expansionable Memory Cluster (AAAI 2024 · similarity 0.86)
- Layerwise Proximal Replay: A Proximal Point Method for Online Continual Learning (ICML 2024 · similarity 0.86)
- Recall-Oriented Continual Learning with Generative Adversarial Meta-Model (AAAI 2024 · similarity 0.86)
- Continual Learning in Modern Hopfield Networks with an Application to Diffusion Models (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
