AI Paper Atlas

Research map / Continual learning / Online and replay-based

Lifelong Generative Modelling Using Dynamic Expansion Graph Model

Fei Ye, Adrian G. Bors

AAAI 2022 · Continual learning › Online and replay-based · approach: Gradient projection and plasticity preservation · 14 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use DEGM-IWELBO-50 instead of the strongest listed GR baselineLifelong VAE · Split MNISTlog-likelihood: -93.57 -> -88.04Improved
Increase DEGM IWELBO weighted samples from 5 to 50 (ablation)DEGM · Split MNISTlog-likelihood: -91.44 -> -88.04Improved
Use DEGM-IWELBO-50 instead of the strongest listed GR baselineLifelong VAE · Split Fashionlog-likelihood: -236.66 -> -233.76Improved
Increase DEGM IWELBO weighted samples from 5 to 50 (ablation)DEGM · Split Fashionlog-likelihood: -235.93 -> -233.76Improved
Use DEGM-IWELBO-50 instead of the strongest listed GR baselineLifelong VAE · COFMIlog-likelihood: -172.10 -> -163.27Improved
Increase DEGM IWELBO weighted samples from 5 to 50 (ablation)DEGM · COFMIlog-likelihood: -164.99 -> -163.27Improved
Use dynamic expansion and graph-based transfer rather than one VAE per taskLifelong VAE · CCCOSCZC averagesquare loss: 381.3 -> 393.8Hurt
Use DEGM rather than one VAE per taskLifelong VAE · CCCOSCZCsquare loss: CelebA: 217.0 -> 229.2; Zappos: 275.8 -> 263.4Mixed
Add a second stochastic layer (ablation)GR VAE · Split MNIST and Split Fashionlog-likelihood: Split MNIST: -93.57 -> -91.23; Split Fashion: -236.66 -> -236.90Mixed
Add a second stochastic layer (ablation)GR VAE · COFMIlog-likelihood: -172.10 -> -188.9Hurt
Use IWELBO with 50 samples rather than ELBO (ablation)DEGM · COFMIlog-likelihood: -168.91 -> -163.27Improved
Use IWELBO with 50 samples rather than ELBO (ablation)DEGM · Split MNIST and Split Fashionlog-likelihood: Split MNIST: -93.51 -> -88.04; Split Fashion: -238.54 -> -233.76Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy