SpiNNaker Spatial Learning: SNN Training Data for Pattern Recognition

SpiNNaker Spatial Learning Code And Dataset contains the software and data used in a research article on supervised learning for Spiking Neural Networks (SNNs). The dataset, created by Sergio Davies and colleagues, supports a method using Spike Timing Dependent Plasticity (STDP) and homeostasis to train SNNs to identify spatial patterns. The 5.8 MB resource was last updated on 2026-05-28 and is shared under a CC-BY-4.0 license.

Use Cases

• Training spiking neural networks for static image pattern recognition based on the described spatial pattern method.
• Developing computer network traffic classifiers based on the principle of converting packet features into spatial spike patterns.
• Benchmarking SNN performance on pattern detection tasks with varying degrees of similarity between patterns.
• Studying the effects of homeostasis on a network's ability to detect imperfectly matching patterns.

Strengths

• Includes both the dataset and the software code used in the original research, enabling full reproducibility.
• The described method achieved 100% accuracy in detecting a single trained pattern in the study.
• Dataset is licensed under the permissive CC-BY-4.0 license, facilitating reuse and derivation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.
• The 5.8 MB size suggests a relatively small-scale dataset.

Provenance

Source

Sergio Davies and co-authors of the associated research article.

Collection Method

Likely generated through simulations or experiments on the SpiNNaker digital architecture for the described research.

Freshness

Last updated 2026-05-28 17:54:14; freshness should be verified.