MOTIF recommender reports gains, especially on cold-start items
Preprint results from three sparse datasets report modest overall gains and larger gains for items whose training interactions were masked.
Search
2,230 results
901–920
Simulated UAV routing proposal delivers more packets on time
This preprint reports about 96% cumulative packet arrival by the deadline for MAPPO-DM in modeled UAV networks, ahead of three multipath baselines, with evidence limited to simulation.
Screen-reader users covered more documents with linked pages
Preprint: In a 16-person comparison, direct document browsing reached more distinct documents, while the AI interface showed more connected exploration.
MTAR Reports Faster Training and Lower FID Than LlamaGen
Preprint: ImageNet tests report up to 0.95 lower FID and 39% faster training for MTAR than LlamaGen, with no change to inference.
Source checks identify 76 confirmed blends among 225 candidates
A preprint dated 27 August 2026 reports a 33.8% false-positive fraction among the candidates examined, with corrected source identifications and periods.
AI Physics Benchmark Finds Accuracy Falls at University Level
Preprint: OmniPhys tests multimodal models on Chinese physics questions, reasoning tasks and diagram generation.
Many cryo-EM runs outperform one-shot reconstruction on mixtures
Preprint: An ensemble of reconstruction jobs reached 97.4% accuracy on a 45-class synthetic test and separated four major ribosomal states from junk.
New framework links p-adic homotopy to algebraic geometry
Preprint: A theorem-driven study sets algebraic tests for when p-complete spaces and étale homotopy types can come from finite CW complexes.
Wireless method detects blocked RIS panels in moving-user tests
This preprint reports active-RIS identification probability above 95% and higher weighted sum rates under tested blockages and mobility.
A Math Preprint Shows a Proposed Boundary Eigenvalue Bound Can Fail
A constructed family of hemisphere metrics slips below the proposed threshold, while separate conformal and rotational analyses find obstacles to boundary-area counterexamples.
Agent-routing method posts higher task success with fewer request messages
Preprint simulations found that SPFR retained 97.63% of a global comparator's Utility while using 81.24 times fewer request-triggered messages.
A New Taxonomy Maps Construction Tasks for Robot Builders
Preprint: Researchers catalogue construction activities and test selected robot skills on a laboratory arm and hand.
Math preprint links stable-pair series to framed-sheaf formulas
This arXiv version-1 preprint reports proved local Hirzebruch identities, while its projective-plane results remain conditional on two conjectures.
AI coding prompts linked to higher costs in benchmark study
Preprint: In 2,700 controlled runs, prompt details and thinking effort were associated with large cost differences, and one cheap probe was associated with lower forecast error.
Radio observations hint at hidden micro-flares on AU Mic
This preprint reports variable 12 to 26 GHz emission, two clear flares and radio behavior the authors say is consistent with trapped energetic electrons.
Unseen Camera Paths Expose Weaknesses in 3D Reconstruction
Preprint: A five-scene benchmark found lower image quality when models were tested on camera trajectories absent from training.
Real-time spatial-audio vocoder reports higher spatial scores
Preprint: CSAVocoder reported higher spatial scores in baseline and listening comparisons, while audio-quality results were mixed.
In one five-tier failure simulation, Standard Retry had lowest success
Preprint: A review of 200 GitHub repositories was paired with a three-scenario simulation of four retry strategies.
Non-Hermitian lattice models show distinct Euler-topology phases
Preprint: Three theoretical models link integer Euler labels with Wilson-loop winding and spectral signatures; the two indices are not reliable in modeled critical phases.
Lightweight AI keeps simulated drone network near optimum
Preprint study reports that a bandit-based controller exceeded 90% of an optimal data-rate benchmark, but it has not yet been tested in hardware or the field.