The AI knowledge 99.9% of people never discover
A new arXiv preprint introduces Mycelium, a shared workspace connecting researchers and AI agents, routing hypotheses to whoever can act on them.
Uniform mixing matches Transformer spatial attention across six traffic benchmarks at 0.14% MAE gap, while cutting complexity from O(N²) to O(N).
MusicMark, posted to arXiv July 14, embeds watermarks during AI music generation rather than after, surviving attacks that defeat older methods rights
New preprint proposes breaking RL reward signals into named climate-control components that greenhouse growers can audit across simulators and logged data.
New arXiv research proposes a 3GPP-aligned framework detecting GNSS spoofing in mobile networks with over 95% accuracy, filling a gap operators now face.
LoRA cascaded fusion preprint proposes adding AI input modalities one at a time for nurse training action recognition, without full retraining.
arXiv research from Georgia Tech and Stanford shows language models flip individual answers when fed meaningless text, as overall scores hold steady.
HORCRUX adds post-quantum cryptography to small RISC-V chips with up to 129x faster hashing, targeting IoT devices that can't run heavy crypto.
A new simulator from Applied Intuition trains driving policies with zero human demonstrations, topping the InterPlan benchmark on a single GPU.
