Fetching from the wire…
Research2026-07-18 · source-backed
arXiv 2607.15217 replaces synchronous layer-by-layer processing with delay-mediated asynchronous propagation through a shared neuron pool, co-evolving topology, weights, delays, and connectivity via a genetic algorithm across a 14,602-gene genome. 204 active paths through 266 hidden neurons, 156 shared, one neuron participating in 11 paths. 85.9% on MNIST features, which is bad. The task is easy and the accuracy is modest, so read this as architecture exploration, not a performance claim. The interesting parts are per-sample adaptive compute depth and the 115 KB model size.
Each link below shares sources, entities, or timing with this story.
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (architecture, co-evolving); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (accuracy, adaptive); pushes against this story (but).
Reported by the same outlet (arxiv.org); overlapping topics (architecture, compute); pushes against this story (but).
Same source domain / Shared topic / Downstream implication
Reported by the same outlet (arxiv.org); overlapping topics (architecture, neuron); traces where this leads (implications).
Same source domain / Shared topic
Reported by the same outlet (arxiv.org); overlapping topics (adaptive, architecture).
Reported by the same outlet (arxiv.org); overlapping topics (accuracy, architecture).
Reported by the same outlet (arxiv.org); overlapping topics (architecture, path).
Reported by the same outlet (arxiv.org); overlapping topics (accuracy, architecture).