New autocorrelation test shows promise at higher time lags
Preprint simulations found mixed performance at lag one but stronger results at selected higher lags, while one solar-intensity series showed several significant lags.
Search
2,230 results
1361–1380
Preprint Narrows the Boundary for Bundles on Quadrics
A mathematical analysis fixes the threshold for odd-dimensional quadrics, narrows the even-dimensional case and identifies non-homogeneous examples.
Preprint Outlines a Route From Visual Learning to General Intelligence
A conceptual synthesis proposes prediction, generation, memory and action as pieces of visual intelligence, but reports no experiment demonstrating the idea.
Turing jump hierarchy restarts after a uniform limit step
Preprint: A formal study finds that ordinary jump closure reaches the arithmetical ideal, while a uniform limit oracle reopens the hierarchy.
Code-driven AI prototype links world state to controllable video
Preprint: A coding agent updates executable worlds while a video model renders their state, but the reported evidence remains qualitative.
Preprint examines bilayer graphene island tunneling and scattering
Tight-binding simulations report associations between valley Hall mode tunneling, impurity resonances and gate-pattern misalignment.
AI tuning coincided with stronger polymer print results
Preprint study on one printer found stronger later batches, but its tests did not establish an advantage over manual tuning.
AI Method Helps Separate Image Signals From Confounding
A preprint study reports more reliable effect estimates in simulations and recovers much of the performance lost to synthetic age confounding in MRI data.
Neural model processes 2.1 million electromagnetic soundings in 25 seconds
Preprint: Held-out tests showed close solver agreement; matched posterior tests found similar but not identical behavior, while the full-sampler comparison was limited by incomplete mixing.
Traffic-video system takes second place, but slips on FETV
Preprint: TAU-Agent combines retrieval, video captions and object tracking to answer questions about transport footage, with uneven results across benchmarks.
Rankings of Tested AI Fact-Checking Systems Vary Across Four Benchmarks
Preprint: A computational comparison found that retrieval and veracity results varied by dataset, while tested language models scored higher with gold-annotated than TF-IDF-retrieved evidence.
Teen Gamers Say Discord Extends Play as Risks Span Platforms
Preprint: Interviews with 16 U.S. teenagers describe Discord as a social hub for gaming, alongside reports of scams, account takeovers and moderation gaps.
Neutron-star model finds finite clusters of magnetic fluxtubes
A preprint reports local vortex pinning and type-1.5-like clustering in a two-dimensional model without phase-gradient entrainment.
Curated AI-image study reports 18 defect-free people and hand images
A 2026 arXiv preprint used human annotations to map defects in AI-generated compositional images and test a repair workflow.
Review Maps Trade-offs in On-Policy Math Self-Distillation
Preprint: A non-experimental review of on-policy self-distillation in mathematical reasoning reports higher scores in some cited comparisons alongside lower functional diversity and longer optimization steps.
A softer lattice may favor charge-density-wave lock-in
An arXiv preprint models how lattice strain and charge-density-wave mismatch can be shared, but its predictions still need experimental tests.
SparseGPT stays closest to dense model in interpretability test
Preprint: In the main 50% gemma-2-2b comparison, SparseGPT and WANDA stayed closer to the dense model than magnitude pruning.
New theorem gives quasimap spaces a global critical-locus model
A 2026 arXiv preprint states the result for arbitrary Nakajima varieties, chosen points and degree data.
Shenzhen simulation finds travel burden varies by resident profile
Preprint: A six-household model completed every scheduled activity, while average ratings for healthcare access and trip chaining trailed travel burden and activity feasibility.
Praxist AI system reports 100% success in rocket simulation
Preprint: Praxist also earned more medals than Claude Code in one MLE-bench sweep, while other evaluations produced mixed results.