1321–1340
This preprint reports strong benchmark scores and controlled tests with a Piper robot from one model trained on real-robot, simulated and human-hand data, but direct transfer to an ARX robot reached 35%.
A version 1 arXiv preprint reports an initial Baniwa speech-recognition baseline from 32.4 minutes of manually transcribed audio, mostly isolated words and short utterances.
An arXiv preprint reports a parallel design that matches the best sequential limit for certain query counts in a low-noise model.
A preprint comparison found neural networks suited scalar response maps, while reduced dynamics extended field predictions to unsimulated spacings.
A theoretical construction omits 0, 1 and infinity in each half-plane, yet is not of bounded type there.
Preprint: Branch-dependent calculations connect the leptonic phase, neutrino mass and a tiny parity-breaking parameter.
Preprint: Laboratory tests found the pulse patterns expected by a two-dimensional friction theory under several controlled fault conditions.
Preprint: A group-shared design is reported to lower parameter and deployment burdens in named benchmarks and tested mobile setups.
Preprint: A formal construction extends the theory to circles, matches two formulations and classifies representations in specified bordism settings.
An arXiv preprint describes three demonstrations, including a physical Jackal working with virtual robots, but offers no comparative performance test.
Preprint: The intent-gating configuration had in-corridor plans in the main replay, while geometry-only scoring detected more conflicts at matched false-positive rates.
Preprint: A modeled Round-Robin attack keeps a Sigries vulnerability window open 70% of the time and cuts estimated MTTF to about one second.
Preprint: Experiments and numerical modeling suggest that CeCo2Ga8 has strongly one-dimensional magnetic and electronic behavior.
Preprint: A finite mathematical framework separates perception lag from concealment and examines coupled instability, attacks and transient hierarchical amplification.
An autonomous coding agent reached 99.5% of a reference score in simulation; full-grid inference took 2.52 seconds versus 1,583 seconds for the reference, a reported speedup of roughly 600 times.
Preprint simulations link curvature to resonance shifts and stronger modeled near fields, while nonlocal calculations reduce peak amplitudes.
A preprint reports a robust half-quantized modular commutator at a single Dirac critical point, but not in every gapless model.
The study compares Langevin dynamics with no annealing and spin-flip annealing in circuit and behavioral simulations.
A mathematical preprint gives exact detection rules for stable and semistable line bundles, while stopping short of classifying every destabilizer.
Preprint: A formal construction maps selected open-system observables to Yang-Lee dynamics without post-selection, but it remains untested on quantum hardware.