arXiv preprints from January 1, 2026 through September 28, 2026 — 06:13:51 EST
Generative AI models reproduce the human biases in their training data and further amplify them through mechanisms such as mode collapse. The loss of diversity produces homogenization, which not only harms the minoritized but impoverishes everyone. We argue homogenization should be a central concern in AI safety. To meaningfully...
In 1987 Hiroshi Maehara conjectured that a graph can be represented by vectors considered adjacent when not orthogonal (a faithful orthogonal representation) in codimension the minimum degree of the graph. Without settling the conjecture, Làslò Lovàsz, Michael Saks, and Alexander Schrijver (LSS) showed that a codimension of vertex...
To realize low-latency spatial transmission system for immersive telepresence, there are two major problems: capturing dynamic 3D scene densely and processing them in real time. LiDAR sensors capture 3D in real time, but produce sparce point clouds. Therefore, this paper presents a high-speed LiDAR point cloud densification method to...
Disaggregating the generation and training stages in RL is widely adopted to scale LLM post-training. There are two critical challenges here. First, the generation stage often becomes a bottleneck due to dynamic workload shifts and severe execution imbalances. Second, the decoupled stages result in diverse and dynamic network traffic...
We prove that continuous spectrum- and commutativity-preserving maps to $\mathcal{M}_n(\mathbb{C})$ from the space of normal (real or complex) $n\times n$, $n\ge 3$ matrices with spectra contained in a given continuous-injection interval image $Λ\subseteq \mathbb{C}$ or $\mathbb{R}$ are (a) conjugations; (b) transpose conjugations, or...
Inspired by a personal dream about knowledge-sharing barriers in an everyday hardware project, this paper proposes a Jung-inspired "Dream Layer" for LLM companions, reframing controlled offline hallucinations as a resource for learning and relationship-building rather than a mere reliability bug. Drawing on Jung's notion of the...
Semi-supervised learning on real-world graphs is frequently challenged by heterophily, where the observed graph is unreliable or label-disassortative. Many existing graph neural networks either rely on a fixed adjacency structure or attempt to handle structural noise through regularization. In this work, we explicitly capture...
Personality assessment in career guidance and personnel selection traditionally relies on self-report questionnaires, which are susceptible to response bias, fatigue, and intentional distortion. Game-based assessment offers a promising alternative by capturing implicit behavioral signals during gameplay. This study proposes a...
The $I=\frac{3}{2}$ $πK$ $s$-wave scattering phase shift is computed by lattice quantum chromodynamics with $N_f=3$ flavors of Asqtad-improved staggered fermions. The energy-eigenvalues of $πK$ systems at one center of mass frame and six moving frames using moving wall source technique are used to get phase shifts by Lüscher's formula...
Learning-based 3D visual geometry models have benefited substantially from large-scale transformers. Among these, StreamVGGT leverages frame-wise causal attention for strong streaming reconstruction, but suffers from unbounded KV cache growth, leading to escalating memory consumption and inference latency as input frames accumulate....
We show that for the standard sinusoidal Winfree model, a coupling strength exceeding twice the maximal magnitude of the intrinsic frequencies guarantees the convergence of the system for Lebesgue almost every initial data. This is proven by first showing, via an order parameter bootstrapping argument, that the pathwise critical...
Multivariate time-series models achieve strong predictive performance in healthcare, industry, energy, and finance, but how they combine cross-variable interactions with temporal dynamics remains unclear. SHapley Additive exPlanations (SHAP) are widely used for interpretation. However, existing time-series variants typically treat the...
Single Image Super-Resolution (SISR) aims to recover high-resolution images from low-resolution inputs. Unlike SISR, Reference-based Super-Resolution (RefSR) leverages an additional high-resolution reference image to facilitate the recovery of high-frequency textures. However, existing research mainly focuses on backdoor attacks...
We consider Cesáro operator on the Hardy space $H^p(\mathbb{C}_+)$ in the upper half-plane for $1<p<\infty$. In \cite{AS} it was proved that for all $1<p<\infty$ the spectrum of the operator $V=\frac{2(p-1)}{p}C-I$ is located on the unit circle and in \cite{ABC1} the authors of this note showed that for $p=2$ operator $V$ is unitary....
The unstructured and irregular nature of points poses a significant challenge for accurate point cloud quality assessment (PCQA), particularly in establishing accurate perceptual feature correspondence. To tackle this, we propose the Multi-scale Implicit Structural Similarity Measurement (MS-ISSM). Unlike traditional point-to-point...
When software artifacts are generated by AI models ("vibe coding"), human engineers assume responsibility for validating them. Ideally, this validation would be done through the creation of a formal proof of correctness. However, this is infeasible for many real-world vibe coding scenarios, especially when requirements for the...
We show how quantum discreteness of spatial area is consistent with a unitary implementation of Lorentz boosts in an LQG type quantization of a diffeomorphism invariant reformulation of free scalar field theory on 2d flat spacetime known as Parameterized Field Theory (PFT). This consistency is a result of Hilbert space...
In this paper, for $1 \leq p, r < \infty$ we characterize those symbols $f$ so that the induced Hankel operators $H_f$ are $r$-summing from Fock spaces $F^p_α$ to $L^p_α$. The main result shows that the $r$-summing norm of $H_f$ is equivalent to the $\mathrm{IDA}^{κ, p}$-norm of $f$, where $κ$ is a positive number determined by $p$...
In recent years, the rapid development of Large Language Models (LLMs) has significantly enhanced natural language understanding and human-computer interaction, creating new opportunities in the field of robotics. However, the integration of natural language understanding into robotic control is an important challenge in the rapid...
Temporal knowledge graph question answering (TKGQA) involves multi-hop reasoning over temporally constrained entity relationships in the knowledge graph to answer a given question. However, at each hop, large language models (LLMs) retrieve subgraphs with numerous temporally similar and semantically complex relations, increasing the...
Amplify and forward (AF) relaying is a viable strategy to extend the coverage of sub-terahertz (sub-THz) links, but inevitably propagates noise, leading to cumulative degradation across multiple hops. At the receiver, optimal decoding is desirable, yet challenging under non-Gaussian input distributions (video, voice, etc), for which...
The adjacency-diametrical matrix (AD matrix) of a connected graph $G$ with diameter $d$, denoted by $AD(G)$, is the matrix indexed by the vertices of $G$ in which the $(i,j)$-entry of $AD(G)$ is $1$ if $d_G(v_i,v_j)=1$, is $d$ if $d_G(v_i,v_j)=d$, and $0$ otherwise, where $d_G(v_i,v_j)$ denotes the distance between the vertices $v_i$...
Video Individual Counting (VIC) is a recently introduced task aiming to estimate pedestrian flux from a video. It extends Video Crowd Counting (VCC) beyond the per-frame pedestrian count. In contrast to VCC that learns to count pedestrians across frames, VIC must identify co-existent pedestrians between frames, which turns out to be a...
We study the symmetries of lattice staggered fermions in 2+1d. Using the symmetries, we can place the system on any sheared torus or Klein bottle. These different backgrounds provide diagnostics of various 't Hooft anomalies associated with the crystalline symmetries. We then compare the lattice model to its continuum limit. The...
Large language models (LLMs) have revolutionized natural language processing, but their ability to process long sequences is fundamentally limited by the context window size during training. Existing length extrapolation methods often suffer from performance degradation or computational inefficiencies. We thereby use a unified...