Research map / Architectures and efficiency / RNNs and normalization
Critical Patch-Aware Sparse Prompting with Decoupled Training for Continual Learning on the Edge
Wonseon Lim, Jaesung Lee, Dae-Won Kim
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 |
|---|---|---|---|
| replace random patch dropping (PD) with critical patch sampling (CPS) (ablation) | CODA-Prompt with 0.5 patch reduction · ImageNet-R (reduction ratio 0.5) | accuracy ACC_N: 45.32 -> 47.16 | Improved |
| add DPCT to PD (ablation) | CODA-Prompt w/ PD · ImageNet-R (reduction ratio 0.5) | accuracy ACC_N: 45.32 -> 47.96 | Improved |
| add DPCT to CPS (ablation) | CODA-Prompt w/ CPS · ImageNet-R (reduction ratio 0.5) | accuracy ACC_N: 47.16 -> 49.28 | Improved |
| add DPCT to CPS (ablation) | CODA-Prompt w/ CPS · ImageNet-R (reduction ratio 0.5) | training time per task: 1,389 s/task -> 1,126 s/task | Improved |
| add CPS and DPCT to CODA-Prompt | CODA-Prompt · ImageNet-R (reduction ratio 0.5) | peak memory: 440 MB -> 253 MB | Improved |
| add CPS and DPCT to CODA-Prompt | CODA-Prompt · ImageNet-R (reduction ratio 0.5) | training time per task: 1,788 s/task -> 1,126 s/task | Improved |
| add CPS and DPCT to CODA-Prompt | CODA-Prompt · ImageNet-R (reduction ratio 0.5) | accuracy ACC_N: 50.24 -> 49.28 | Hurt |
| lower sampling temperature in CPS (ablation) | CPS-Prompt · CUB-200 | accuracy | Improved |
| use moderate phase ratio lambda in 0.4-0.6 (ablation) | CPS-Prompt · CUB-200 | accuracy | Improved |
| use stochastic multinomial sampling instead of deterministic top-k (ablation) | CPS-Prompt · CUB-200 | accuracy | Improved |
| compare CPS-Prompt with C-Prompt | CPS-Prompt · CIFAR-100, ImageNet-R, CUB-200 | memory, time, energy: about 4.3x memory, 3.1x time, 3.3x energy | Improved |
| compare CPS-Prompt with CODA-Prompt | CPS-Prompt · CIFAR-100, ImageNet-R, CUB-200 | memory, time, energy: about 1.6x less memory, 1.5x shorter time, 1.6x less energy | Improved |
Closest papers by meaning
- Consistent Prompting for Rehearsal-Free Continual Learning (CVPR 2024 · similarity 0.89)
- Mixture of Experts Meets Prompt-Based Continual Learning (NeurIPS 2024 · similarity 0.88)
- REP: Resource-Efficient Prompting for Rehearsal-Free Continual Learning (NeurIPS 2025 · similarity 0.88)
- Evolving Parameterized Prompt Memory for Continual Learning (AAAI 2024 · similarity 0.88)
- Training Consistent Mixture-of-Experts-Based Prompt Generator for Continual Learning (AAAI 2025 · similarity 0.87)
- Advancing Prompt-Based Methods for Replay-Independent General Continual Learning (ICLR 2025 · similarity 0.86)
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.
