Research map / Architectures and efficiency / RNNs and normalization
Selfless Sequential Learning
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 |
|---|---|---|---|
| Add SLNID with randomly initialized fully connected layers | MAS on the 8-task object recognition sequence · 8-task object recognition sequence | average test accuracy: 52.69 -> 54.50 | Helped |
| Add SLNID with pretrained fully connected layers | MAS on the 8-task object recognition sequence · 8-task object recognition sequence | average test accuracy: 52.69 -> 53.77 | Helped |
| Add representation sparsity regularization | MAS on Permuted MNIST · Permuted MNIST, hidden size 128 | average test accuracy: 92.67 -> 95.83 | Helped |
| Add full SLNID, including locality and neuron importance | MAS on CIFAR-100 · CIFAR-100, 10 tasks, hidden size 256 | average test accuracy: 55.06 -> 63.30 | Helped |
| Add neuron importance integration to neural inhibition (ablation) | SNI on CIFAR-100 · CIFAR-100, 10 tasks, hidden size 256 | average test accuracy: 55.30 -> 63.30 | Helped |
| Add locality relaxation to neural inhibition with neuron importance (ablation) | SNID on CIFAR-100 · CIFAR-100, 10 tasks, hidden size 256 | average test accuracy: 61.00 -> 63.30 | Helped |
| Add neuron importance integration (ablation) | SLNI on Permuted MNIST · Permuted MNIST, hidden size 128 | average test accuracy: 95.95 -> 95.83 | Hurt |
| Add locality relaxation (ablation) | SNID on Permuted MNIST · Permuted MNIST, hidden size 128 | average test accuracy: 95.90 -> 95.83 | No clear effect |
| Add MAS importance-weight regularization to SLNID (ablation) | SLNID on CIFAR-100 without hard task boundaries · CIFAR-100 without hard task boundaries, final average accuracy | average accuracy: 70.75% -> 70.94% | Helped |
| Add neuron importance integration to local inhibition (ablation) | SLNI w/o MAS on CIFAR-100 without hard task boundaries · CIFAR-100 without hard task boundaries, final average accuracy | average accuracy: 63.54% -> 70.75% | Helped |
| Add locality relaxation to neural inhibition (ablation) | SNI on Permuted MNIST · Permuted MNIST, hidden size 64 | average test accuracy: 94.89 -> 93.89 | No clear effect |
Closest papers by meaning
- Learning Continually by Spectral Regularization (ICLR 2025 · similarity 0.89)
- Sparse Coding in a Dual Memory System for Lifelong Learning (AAAI 2023 · similarity 0.88)
- Lifelong Neural Predictive Coding: Learning Cumulatively Online without Forgetting (NeurIPS 2022 · similarity 0.87)
- A Statistical Theory of Regularization-Based Continual Learning (ICML 2024 · similarity 0.87)
- Self-Normalized Resets for Plasticity in Continual Learning (ICLR 2025 · similarity 0.86)
- Optimal Rates in Continual Linear Regression via Increasing Regularization (NeurIPS 2025 · similarity 0.85)
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.