| title | Explanation |
|---|
This section provides conceptual discussions and explanations of the ideas behind DynVision. Here you'll find in-depth coverage of the theoretical foundations, design principles, and biological inspirations that inform the toolbox.
These pages are understanding-oriented: read them to learn why DynVision works the way it does. For practical steps see the How-to Guides; for factual lookups see the Reference.
- Biological Plausibility: How DynVision implements biologically plausible features
- Temporal Dynamics: Understanding temporal processing in vision models
- Engineering vs. Biological Time: The two unrolling conventions and delay‑conversion formulas
- Design Philosophy: The guiding principles behind DynVision's architecture
- Role of Recurrence: Why recurrent connections matter in visual processing
- Comparison to Neural Data: How model dynamics compare to ECoG recordings and human behavioural data
- Why Snakemake?: The reasoning behind DynVision's workflow orchestration
The following conceptual pages are planned but not yet written. They are tracked in the Documentation TODOs and will be added on demand:
- Continuous vs. discrete time, time constants, propagation delays
- Visual-cortex organization & cortical connectivity
- Comparisons with standard CNNs, other RCNNs, and spiking networks
- Trade-offs in balancing biological fidelity and performance