Research map / Architectures and efficiency / RNNs and normalization
Cortico-cerebellar networks as decoupling neural interfaces
Joseph Oliver Pemberton, Ellen Boven, Richard Apps, Rui Ponte Costa
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 cerebellar predicted feedback (ccRNN vs cRNN) | ccRNN · Line drawing | learning speed / variability | Improved |
| Add cerebellar module | ccRNN · Line drawing | ataxia score | Improved |
| Shorter cortical temporal feedback window | ccRNN · Line drawing | MSE relative to cRNN | Improved |
| Continuous external feedback instead of sparse | ccRNN · Line drawing | benefit over cRNN | Hurt |
| Ablate cerebellar module partway through training (ablation) | ccRNN · Line drawing | training error | Hurt |
| Add cerebellar module | ccRNN · ld-seqMNIST | learning speed and ataxia score | Improved |
| Add cerebellar module | ccRNN · dd-seqMNIST (digit drawing) | learning speed | Mixed |
| Add cerebellar module | ccRNN · Sequential MNIST classification | accuracy: ~10% higher accuracy after about 10 epochs | Improved |
| Add cerebellar module | ccRNN · Caption generation | BLEU: 0.2288 -> 0.2294 | No clear effect |
| Add cerebellar module | ccRNN · Caption generation | CIDEr: 0.6928 -> 0.6940 | No clear effect |
| Add cerebellar module | ccRNN · Caption generation | METEOR: 0.2114 -> 0.2112 | No clear effect |
| Add cerebellar module | ccRNN · Caption generation | learning curve (bits per word) | Improved |
Closest papers by meaning
- Feedback control guides credit assignment in recurrent neural networks (NeurIPS 2024 · similarity 0.87)
- Scaling Laws and Tradeoffs in Recurrent Networks of Expressive Neurons (NeurIPS 2026 · similarity 0.84)
- Emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks (ICLR 2024 · similarity 0.84)
- A Scalable, Causal, and Energy Efficient Framework for Neural Decoding with Spiking Neural Networks (NeurIPS 2025 · similarity 0.84)
- Unconditional stability of a recurrent neural circuit implementing divisive normalization (NeurIPS 2024 · similarity 0.84)
- Few-shot Algorithms for Consistent Neural Decoding (FALCON) Benchmark (NeurIPS 2024 · similarity 0.83)
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.
