Critical Reservoir Initialization Criterion
Critical Reservoir Initialization Criterion
Auto-generated stub. Edit this file to add more details.
A random-matrix-based criterion for identifying the critical gain parameter that separates ordered from chaotic regimes in recurrent neural network initialization.
Why It Matters
Provides a principled, theoretically-grounded method for weight initialization in recurrent structures, bridging the gap between random matrix theory and empirical reservoir performance.
Evidence
we derive a simple criterion to estimate the critical g_c for a broad class of recurrent architectures and we show that it closely tracks the gain at which a gated-RNN reservoir achieves peak performance on a chaotic forecasting task.
Related Papers
Metadata & Links
- created_at
- 2026-05-14T06:10:53Z