Research map / Continual learning / Class-incremental
Towards Realistic Evaluation of Industrial Continual Learning Scenarios with an Emphasis on Energy Consumption and Computational Footprint
Vivek Chavan, Paul Koch, Marian Schlüter, Clemens Briese
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 |
|---|---|---|---|
| introduce periodic joint training update (JTupdate) to incremental learning (ablation) | DER · InVar-100 six-month study | AUCacc: 0.72 -> 0.80 | Improved |
| introduce JTupdate (ablation) | DER · InVar-100 six-month study | PFLOPs: 1.61E15 -> 1.22E15 | Improved |
| introduce JTupdate (ablation) | POD-AANet · InVar-100 six-month study | AUCacc: 0.64 -> 0.78 | Improved |
| introduce JTupdate (ablation) | PODNet · InVar-100 six-month study | AUCacc: 0.58 -> 0.67 | Improved |
| use incremental learning rather than cumulative joint training | CIL methods · ImageNet-Subset | energy (kWh): 33.29 -> 6.12 (FOSTER) | Improved |
| use incremental learning rather than cumulative joint training | CIL methods · ImageNet-Subset | average accuracy (%): 83.7 -> 72.4 (DER best) | Hurt |
| increase rehearsal memory to 2000 per subcategory (ablation) | CIL methods · InVar-100 Experiment 3 | accuracy and energy: average accuracy up to 90.45; energy up to 23.2 kWh | Mixed |
| increase clutter in image context (cluttered vs white background) | CIL methods · InVar-100 Experiment 2 | average incremental accuracy: DER 90.3% -> 58.6% | Hurt |
| compare accuracy-only reporting versus adding energy and FLOPs | CIL evaluation · ImageNet-Subset | energy (kWh): DER 72.4% at 15.08 kWh vs FOSTER 69.8% at 6.12 kWh | Mixed |
Closest papers by meaning
- A Comprehensive Survey of Continual Learning: Theory, Method and Application (TPAMI 2024 · similarity 0.86)
- General Incremental Learning With Domain-Aware Categorical Representations (CVPR 2022 · similarity 0.86)
- Learnability and Algorithm for Continual Learning (ICML 2023 · similarity 0.86)
- Continual Learning Using a Kernel-Based Method Over Foundation Models (AAAI 2025 · similarity 0.85)
- eTag: Class-Incremental Learning via Embedding Distillation and Task-Oriented Generation (AAAI 2024 · similarity 0.85)
- AnaCP: Toward Upper-Bound Continual Learning via Analytic Contrastive Projection (NeurIPS 2025 · similarity 0.84)
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.
