Research map / Continual learning / Online and replay-based
Organizing recurrent network dynamics by task-computation to enable continual learning
Lea Duncker, Laura Driscoll, Krishna V. Shenoy, Maneesh Sahani, David Sussillo
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 plain SGD with projected-update continual learning rule in sequential training | RNN on DelayPro/DelayAnti/MemoryPro/MemoryAnti · Sequential multi-task neuroscience tasks | task performance after sequence | Improved |
| use projected-update rule instead of EWC | sequentially trained RNN · Sequential multi-task neuroscience tasks | task performance after sequence | Improved |
| use projected-update rule instead of synaptic intelligence | sequentially trained RNN · Sequential multi-task neuroscience tasks | task performance after sequence | Improved |
| pretrain on a task with the same stimulus-response relationship before second task (ablation) | sequentially trained RNN · Second-task learning speed | test log cost decrease rate | Improved |
| train simultaneously on all tasks instead of sequentially with the rule | RNN · Four multi-task suite | task performance | Improved |
| project out subspace of same-relationship task vs opposite-relationship task (ablation) | trained RNN · Hidden-state perturbation | task performance | Hurt |
Closest papers by meaning
- Dynamical phases of short-term memory mechanisms in RNNs (ICML 2025 · similarity 0.89)
- Learning dynamics in linear recurrent neural networks (ICML 2025 · similarity 0.88)
- Emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks (ICLR 2024 · similarity 0.88)
- Feedback control guides credit assignment in recurrent neural networks (NeurIPS 2024 · similarity 0.88)
- Learning system dynamics without forgetting (ICLR 2025 · similarity 0.87)
- Separating the 'what' and 'how' of compositional computation to enable reuse and 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.
