Research map / Continual learning / Online and replay-based
Measuring Catastrophic Forgetting in Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use PathNet rather than the MLP baseline | Incremental learning on MNIST with data permutations · MNIST, data permutation | Ω_all: 0.702 -> 0.848 | Helped |
| Use EWC rather than the MLP baseline | Incremental learning on CUB-200 with data permutations · CUB-200, data permutation | Ω_all: 0.635 -> 0.762 | Helped |
| Use PathNet rather than the MLP baseline | Incremental learning on AudioSet with data permutations · AudioSet, data permutation | Ω_all: 0.446 -> 0.615 | Helped |
| Use rehearsal with GeppNet rather than the MLP baseline | Incremental class learning on MNIST · MNIST, incremental class learning | Ω_all: 0.181 -> 0.922 | Helped |
| Use rehearsal with GeppNet rather than the MLP baseline | Incremental class learning on CUB-200 · CUB-200, incremental class learning | Ω_all: 0.031 -> 0.585 | Helped |
| Use rehearsal with GeppNet rather than the MLP baseline | Incremental class learning on AudioSet · AudioSet, incremental class learning | Ω_all: 0.044 -> 0.947 | Helped |
| Use EWC rather than the MLP baseline | Multi-modal incremental learning on CUB-200 · CUB-200, multi-modal | Ω_all: 0.610 -> 0.872 | Helped |
| Use EWC rather than the MLP baseline | Multi-modal incremental learning on AudioSet · AudioSet, multi-modal | Ω_all: 0.589 -> 0.984 | Helped |
| Use GeppNet+STM rather than GeppNet | Incremental class learning · CUB-200 and AudioSet | Ω_all: CUB-200: 0.585 -> 0.626; AudioSet: 0.947 -> 0.920 | Mixed |
| Learn with a larger model capacity | FEL · CUB-200 and AudioSet | Ω_all: CUB-200 data permutation: 0.184 -> 0.732; AudioSet data permutation: 0.239 -> 0.444 | Mixed |
Closest papers by meaning
- A comprehensive, application-oriented study of catastrophic forgetting in DNNs (ICLR 2019 · similarity 0.92)
- Effect of scale on catastrophic forgetting in neural networks (ICLR 2022 · similarity 0.91)
- Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgetting (ICML 2019 · similarity 0.91)
- Towards Understanding Catastrophic Forgetting in Two-layer Convolutional Neural Networks (ICML 2025 · similarity 0.89)
- Kolmogorov-Arnold Networks Still Catastrophically Forget but Differently from MLP (AAAI 2025 · similarity 0.89)
- When to Learn and When to Stop: Quitting at the Optimal Time (Student Abstract) (AAAI 2025 · similarity 0.88)
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.