Research map / Continual learning / Online and replay-based
CIP-Net: Continual Interpretable Prototype-based Network
Federico Di Valerio, Michela Proietti, Alessio Ragno, Roberto Capobianco
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 prototype regularizer LR (ablation) | CIP-Net · CUB TIL (4 tasks) | final average accuracy: 0.48 -> 0.77 | Improved |
| add prototype regularizer LR (ablation) | CIP-Net · CUB CIL (4 tasks) | final average accuracy: 0.27 -> 0.47 | Improved |
| add learnable temperature tau to normalized head (ablation) | CIP-Net · CUB TIL (4 tasks) | final average accuracy: 0.46 -> 0.77 | Improved |
| add learnable temperature tau to normalized head (ablation) | CIP-Net · CUB CIL (4 tasks) | final average accuracy: 0.21 -> 0.47 | Improved |
| add head decorrelation loss LD (ablation) | CIP-Net · CUB CIL (4 tasks) | final average accuracy: 0.43 -> 0.47 | Improved |
| add head decorrelation loss LD (ablation) | CIP-Net · CUB TIL (4 tasks) | final average accuracy: 0.77 -> 0.77 | No clear effect |
| add Hoyer sparsity loss LH (ablation) | CIP-Net · CUB CIL (4 tasks) | final average accuracy: 0.50 -> 0.47 | No clear effect |
| add Hoyer sparsity loss LH (ablation) | CIP-Net · CUB TIL (4 tasks) | final average accuracy: 0.81 -> 0.77 | No clear effect |
| use CIP-Net instead of ICICLE (shared prototype layer) | CIP-Net · CUB TIL (20 tasks) | final average accuracy: 0.497 -> 0.847 | Improved |
| use CIP-Net instead of ICICLE (shared prototype layer) | ICICLE · CUB CIL (4 tasks) | final average accuracy: 0.350 -> 0.469 | Improved |
| use CIP-Net instead of PASS (exemplar-free, non-interpretable) | PASS · CUB CIL (20 tasks) | final average accuracy: 0.139 -> 0.180 | Improved |
| use CIP-Net instead of PASS | PASS · CUB TIL (4 tasks) | final average accuracy: 0.775 -> 0.772 | No clear effect |
Closest papers by meaning
- A continual learning survey: Defying forgetting in classification tasks (TPAMI 2021 · similarity 0.88)
- A Comprehensive Survey of Continual Learning: Theory, Method and Application (TPAMI 2024 · similarity 0.88)
- Efficient Continual Learning with Modular Networks and Task-Driven Priors (ICLR 2021 · similarity 0.88)
- Prototype-Sample Relation Distillation: Towards Replay-Free Continual Learning (ICML 2023 · similarity 0.87)
- Language Guided Concept Bottleneck Models for Interpretable Continual Learning (CVPR 2025 · similarity 0.87)
- Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning (NeurIPS 2025 · similarity 0.87)
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.
