Search arXiv⌕ Search

arXiv subjects

Shuai Zhang

Publications and source records attributed to Shuai Zhang.

At least 37 records · Page 2Linked to original sources
arXiv2026

Aligning Performance with Contribution: Towards Contribution-Aware Fair Recommendation

Existing research on user fairness in recommender systems has developed diverse objectives. However, it has paid limited attention to a distinct distributive perspective: whether users' contributions to model learning should be reflected in the recommendation benefits they receive. We argue that, in addition to existing fairness protections, a fair system may account for the alignment between users' estimated contributions and the recommendation performance they receive. Such alignment can incentivize sustained and informative engagement, thereby supporting a sustainable recommendation ecosystem. To this end, we propose Contribution-Performance Fairness, a novel fairness perspective which requires recommendation performance to be aligned with estimated contribution across user groups and to remain equitable among users with comparable contributions within a same group. To instantiate this perspective, we introduce the Contribution-Performance Fair Recommender (CPFR), a framework applicable to different backbone recommenders. CPFR constructs ordered user groups from a training-dependent contribution considering interaction volume, loss alignment, and optimization intensity, and jointly optimizes recommendation accuracy with the two fairness requirements. A game-theoretic analysis shows that such alignment can strengthen contribution incentives and improve system-level recommendation accuracy under voluntary contribution. Experiments on three datasets and three backbone models demonstrate that CPFR achieves a strong accuracy--fairness trade-off under the proposed operational metric.

cs.IR↗
arXiv2026

Gate-tunable large second-order and third-order nonlinear Hall effect in topological insulator BiSbTeSe2

The nonlinear Hall effect (NLHE) holds significant promise for next-generation electronic device applications, and it is essential to explore and engineer a strong and highly tunable NLHE. In this work, we report large second-order and third-order NLHEs in dual-gated BiSbTeSe2 devices. Both the second-order and third-order NLHE are highly gate-tunable. The second-order NLHE exhibits a higher generation efficiency than those reported in previous topological insulator systems, and it can be stably observed up to room temperature. Meanwhile, the generation efficiency of the third-order NLHE is comparable to previously reported values. The scaling analysis indicates that the second-order NLHE is consistent with contributions from skew scattering and a term compatible with either the Berry curvature dipole or side jump, whereas the third-order NLHE is consistent with contributions from the Berry connection polarizability tensor and higher-order skew scattering. Our work establishes a promising experimental platform for the development of NLHE-based devices.

cond-mat.mes-hall↗
arXiv2026

CreativeFlow: A One-to-Many Analogical Relation Transfer Method for 3D Asset Generation

Inspired by cognitive science, we present CREATIVEFLOW, an analogical generation framework that explicitly models analogical divergent thinking to mitigate creative homogenization in text-to-3D pipelines. Our method derives a series of meaningful yet relationally similar source-target asset pairs, each featuring distinct geometric configurations. Expert evaluations demonstrate that our framework substantially enhances creative novelty and visual fascination. This workflow and its resulting assets establish a foundational dataset and benchmark for future relation-aware 3D model training.

cs.AI↗
arXiv2026

The Model Shape Behind Scaling Laws: A Gradient Superposition Perspective

Neural scaling laws guide resource allocation, but how should a fixed parameter budget be divided between depth and width? Motivated by inverse-width representation-error scaling under strong superposition, we study this allocation through the network's end-to-end response. The Neural Feature Ansatz links weight and gradient geometry, motivating the Frobenius norm $F$ of the input average gradient outer product (AGOP) as an observable of response strength and overlap. By reproducing Anthropic's toy-model double-descent experiment, we find that $F$ and test reconstruction loss follow an approximately affine relationship across the full sample-size sweep and share an interpolation peak. Through Gaussian minimum-norm regression experiments, we further show that this correlation arises from a fitted-noise amplification term shared by prediction error and $F$. By pretraining language models spanning different shapes and parameter budgets, we find a strong positive correlation between loss and $F$ among the best shapes selected by validation loss. This supports interpreting effective parameter allocation as preserving task-relevant computation while reducing the amplification of error responses. Our results identify error-response amplification as a mechanism linking gradient superposition to generalization, providing a functional account of depth-width allocation and efficient parameter utilization. The code is availale at: $\hyperlink{https://github.com/wenjie1835/Shape_Scaling}{https://github.com/wenjie1835/Shape_Scaling}$.

cs.LG↗
arXiv2026

Curvature integral on convex hypersurfaces

Let $M^n$, $n\ge3$, be the smooth, connected, complete, noncompact boundary of a closed convex set with nonempty interior in $\mathbb{R}^{n+1}$. We prove that its normalized scalar-curvature integral at infinity is at most $4πω_{n-2}(1-\operatorname{AVR}(M))$, except when, up to an ambient rigid motion, $M$ is the product of a smooth closed convex surface in $\mathbb{R}^3$ and $\mathbb{R}^{n-2}$. These exceptional products have zero asymptotic volume ratio and limit $8πω_{n-2}$. We also characterize the equality case for the $4πω_{n-2}(1-\operatorname{AVR}(M))$ bound. The proof uses exterior tube volumes together with the cone Steiner and Gauss--Bonnet formulas. An elementary comparison of intrinsic and Euclidean radial distances proves existence of the intrinsic limit and identifies its exact value from the cone coefficients. The same conclusions extend to orientable convex submanifolds of arbitrary codimension

math.DG↗
arXiv2026

A Dominant Self-Conditioning Direction Drives Repetition in Unconditional Continuous Diffusion Language Models

Continuous diffusion language models offer an alternative to autoregressive generation, but their generations may suffer from repetition. We find that unconditional generations from ELF, a recent family of continuous diffusion language models, are more repetitive than human text, while Gen-PPL, a common likelihood-based metric, gives lower perplexity to repetitive generations and can conceal this problem while biasing quality evaluation. Our analysis links this behavior to a self-conditioning feedback loop in which clean-embedding predictions are repeatedly carried into subsequent denoising steps, driving representations toward an effectively one-dimensional contractive attractor associated with repetition. Based on this mechanism, we introduce Attractor-Contrast-Escape (ACE), a training-free inference-time intervention that estimates a repetition direction by contrasting denoising paths trapped in repetition with paths relatively free of repetition and subtracts it from the self-conditioning feedback during denoising. Using a direction estimated only once on ELF-B, ACE reduces mean 4-gram self-repetition rate from 7.28% to 4.48%, while retaining competitive results on several text-quality metrics beyond Gen-PPL. The direction remains effective across ELF sizes and inference configurations, and ACE also generalizes to other unconditional self-conditioned continuous diffusion language models. These results identify self-conditioning feedback as a source of repetition in continuous diffusion language models and show that ACE can directly mitigate this repetition during inference.

cs.CL↗
arXiv2026

TCSAlgBench: Benchmarking Automated Proving for Research-Level Theoretical Computer Science

Large language models perform strongly on competition mathematics, but their research-level reasoning remains difficult to evaluate systematically. Theoretical computer science (TCS) connects algorithm design to explicit guarantees and fundamental limits, providing a setting for evaluating whether models can justify computational improvements with arguments humans can inspect. We introduce TCSAlgBench, a benchmark and reusable pipeline for natural-language proof discovery, comprising 398 theorem-level challenges from 138 STOC and COLT 2026 papers. Expert-designed rules complete paper-specific context, preserve computational assumptions and quantitative guarantees, and withhold constructions when discovering an algorithm is part of the task. For each task, prover systems receive theorem statements and access to cited prior work. The pipeline supports fresh, versioned challenge batches from newly released papers. We evaluate ten model configurations from four families under direct inference and prover-verifier discussion, and compare four agent workflows under matched model-call opportunities. All evaluations use the full benchmark. In the model comparison, GPT-5.6 Sol max achieves the highest five-run verifier-accepted coverage at 23.6% after 10-round discussion. Discussion and repeated sampling improve coverage. In the separate agent comparison using GPT-5.5 xhigh, decomposition improves coverage over discussion, and agentic planning achieves the highest five-run verifier-accepted coverage at 25.4%. TCSAlgBench provides a refreshable testbed for measuring progress in model reasoning and studying how agent workflows support research-level proof discovery.

cs.AI↗
arXiv2026

CitySTAR: Structured and Topology-Aware Reasoning for Open-Vocabulary Urban 3D Grounding

3D grounding aims to localize target entities in complex scenes from natural language and plays a fundamental role in embodied perception and spatial reasoning. However, existing approaches mostly rely on feature similarity or direct matching, making it difficult to connect natural-language intent with the implicit semantic and geometric structures hidden in billion-scale urban point clouds. We reformulate city-scale 3D grounding as structured constraint reasoning, where description semantics are organized into computable cross-modal constraints over open-vocabulary 3D entities, attributes, and spatial relations. We present CitySTAR, a training-free framework for reasoning-driven urban 3D grounding. CitySTAR lifts raw billion-scale urban point clouds into a query-ready scene graph of open-vocabulary 3D instances, with CodeLLM-driven tools supplying multimodal evidence for node attributes and 3D spatial relations. It then models target-context topology with paired hypergraphs and performs bidirectional topology verification for structural disambiguation. Finally, a Reflective Cross-modal Grounding module integrates topology consistency and candidate-centered 2D visual evidence to make decisions over a metric-aware 3D context graph. To further support this setting, we introduce CitySTAR-3D, an enhanced benchmark that improves semantic coverage, instance completeness, bounding-box fidelity, and spatial-relation complexity in city-scale 3D grounding. Extensive experiments show that CitySTAR consistently improves open-world urban 3D grounding while maintaining strong interpretability and generalization.

cs.CV↗
arXiv2026

BiSegMamba: Efficient Bidirectional Tri-Oriented Mamba for 3D Medical Image Segmentation

Accurate 3D medical image segmentation requires both fine spatial detail and long-range volumetric context. Although Mamba provides efficient long-sequence modeling, volumetric formulations remain sensitive to spatial serialization order, repeated multi-orientation processing, and fixed aggregation of orientation-specific representations. We propose BiSegMamba, an efficient bidirectional tri-oriented Mamba framework that follows a compact-to-detail and local-to-global processing strategy. A progressive compacting stem shifts semantic reasoning to compact representations while retaining shallow high-resolution features for reconstruction, and a multi-scale spatial mixer captures local anatomical patterns. For global context modeling, volumetric features are serialized along three anatomical orientations and their reversed directions, the resulting six sequences are processed jointly by a shared Mamba operator, and reconciled using learned channel-wise forward-backward weighting. The restored orientation-specific features are subsequently integrated through input-dependent directional fusion. Experiments on an in-house carotid dataset and ACDC, BraTS2023, and AMOS-CT achieve Dice scores of 96.3%, 91.63%, 92.57%, and 89.03%, respectively. Compared with SegMamba-V2 under standardized $128^3$ profiling, BiSegMamba reduces the parameter count by 65.9% (138.77M$\rightarrow$47.38M) and inference latency by 53.6% (129.08$\rightarrow$59.86 ms), while increasing throughput by 115.7% (7.75$\rightarrow$16.71 samples/s). These results demonstrate a favorable accuracy--efficiency trade-off among volumetric Mamba-based segmentation models.

cs.CV↗
arXiv2026

SkyAnchor: Updating Metric-scale Aerial 3D Gaussian Scenes from Unposed Ground-View Sequences

We study how to update a pre-built aerial scene with a newly captured, unposed ground-view sequence. The aerial scene already contains a reliable metric Structure-from-Motion (SfM) reconstruction and a pre-trained 3D Gaussian Splatting (3DGS) model, whereas the ground-view sequence is collected later to add street-level appearance but has unknown camera poses and global scale. Registering this sequence to the aerial SfM reconstruction is challenging because single-image cross-view localization is brittle and long trajectories are prone to drift. To address these challenges, we present SkyAnchor, which treats the existing aerial scene as a fixed scaffold for ground-view registration and scene update instead of jointly reconstructing aerial and ground imagery from scratch. It first localizes short groups of consecutive ground frames against geometrically verified aerial support, producing sparse anchor poses. It then recovers the full ground trajectory with anchor-constrained submaps, fixing the front and rear anchor poses during incremental registration and bundle adjustment. Finally, it inserts filtered ground Gaussians while preserving the aerial view, followed by lightweight joint refinement. Experiments on seven real aerial--ground scenes show accurate metric ground trajectories and updated 3D Gaussian scenes with strong aerial- and ground-view rendering quality.

cs.CV↗
arXiv2026

Probing the anomalous symmetry-breaking in kagome material CsV3Sb5 via third-order nonlinearity

The kagome material has rapidly established itself as a research frontier in condensed matter physics, owing to its distinctive geometric structure and the rich array of unconventional physical phenomena. In the kagome AV3Sb5 (A = K, Rb, Cs) family, the charge-ordered state exhibits a remarkable characteristic, i.e., anomalous symmetry-breaking, which is tied to the topological nature of the electronic band structure. Here, we report the third-order nonlinear longitudinal and Hall responses that persist stably up to room temperature in the kagome material CsV3Sb5. Notably, the nonlinear responses demonstrate significant enhancement below the charge density wave (~ 77 K) order and anomalous symmetry-breaking (~ 39 K) state. The scaling analysis indicates that the third-order nonlinear transport is governed jointly by quantum geometric contribution and extrinsic scattering. This study realizes a giant third-order nonlinear response and provides a distinct method to detect anomalous symmetry-breaking in CsV3Sb5.

cond-mat.str-el↗
arXiv2026

LLMZero: Discovering Adaptive Training Strategies for RL Post-Training via LLM Agents

RL post-training strategies are dataset-dependent and reveal a recurring empirical pattern: capacity parameters accumulate monotonically across stages, while regularization parameters predominantly oscillate in response to shifting training dynamics. This distinction highlights a potential flaw in fixed training schedules: by forcing all parameters along rigid paths, they fail to capture the dynamic exploration-exploitation tradeoffs that regularization must track. We uncover this through LLMZero, an agentic system that optimizes training trajectories via tree search by diagnosing pathologies at each checkpoint and proposing coordinated multi-parameter transitions. Across four diverse GRPO tasks, LLMZero discovers strategies that improve over the base model by 9% to 140% and over grid search by 6% to 15% (relative), consistently outperforming random search and a skill-based agent under a matched compute budget. The capacity--regularization asymmetry is consistent across all four tasks, offering a candidate design heuristic for multi-stage training.

cs.LG↗
arXiv2026

PeopleSearchBench: Evaluating AI-Powered People Search Platforms with Criteria-Grounded Verification

AI-powered people search platforms are increasingly deployed for recruiting, sales prospecting, and professional networking, yet no standardized benchmark exists for their rigorous evaluation. We present PeopleSearchBench, an open-source benchmark comprising 119 multilingual queries across four scenarios: corporate recruiting, B2B sales prospecting, expert search, and influencer discovery. A central contribution is Criteria-Grounded Verification, an evaluation methodology that decomposes each query into explicit, independently checkable criteria and verifies each returned individual via live web search, producing factual relevance judgments rather than subjective LLM-as-judge scores (Cohen's kappa = 0.84 with human annotators). We evaluate four architecturally diverse platforms along three complementary dimensions---Relevance Precision, Effective Coverage, and Information Utility---and find that multi-source search agents significantly outperform single-domain systems, particularly in influencer discovery where the performance gap is largest. Platform rankings are robust across ablations on scoring thresholds, dimension weights, and judge models. All code, queries, and evaluation prompts are publicly available.

cs.AI↗
arXiv2026

Post-Newtonian dynamics of charged compact binaries

We investigate the dissipative dynamics of charged compact binaries in Einstein-Maxwell theory. By evaluating the mass and electric multipole moments, we compute the gravitational and electromagnetic fluxes {through next-to-leading order in the post-Newtonian expansion}. Using the flux-balance equations, we derive the evolution of the orbital angular frequency for quasi-circular inspirals. We further analyze circular orbit stability in charged black-hole binaries and quantify how the charge-to-mass ratios affect the inspiral dynamics.

gr-qc↗
arXiv2026

Certified Multi-Turn Robustness for LLM Safety via Compositional Bounds and Safety Persistence

Large language models (LLMs) are vulnerable to multi-turn jailbreak attacks that progressively manipulate conversation context. Existing certified robustness methods are limited to single-turn inputs; naive multi-turn composition yields bounds that degrade exponentially in the number of turns. We introduce Multi-Turn Certified Robustness (MTCR), a framework that models conversational safety via State-Adversarial MDPs and defines $k$-turn certified robustness as the worst-case safety probability across $k$ adversarial turns. MTCR comprises: (i) compositional certification via embedding-space mode decomposition, yielding tighter certified lower bounds than naive multiplication; (ii) $(α,β)$-safety persistence, improving the degradation rate from $\underline{p}^{k}$ to $β^k$ (with $β> \underline{p}$) and yielding interpretable horizon estimates; (iii) matching information-theoretic upper bounds establishing tightness; and (iv) a unified algorithm combining these results. Experiments on six LLMs under $ε$-bounded and Crescendo-style attacks confirm that empirical safety consistently exceeds the certified bounds.

cs.AI↗
arXiv2026

Multi-Feature Riemannian Hypergraph for Online Test-Time Adaptation of Motor Imagery Brain-Computer Interface

In clinical motor imagery brain-computer interface (MI-BCI) decoding, cross-day transferability and online operation remain two critical challenges. Hypergraphs can improve transferability by capturing higher-order sample relationships, yet existing hypergraph-based methods for online emotion recognition neglect the cross-day benefits of Riemannian geometry widely adopted in EEG transfer learning. To bridge this gap, we propose the Multi-feature Riemannian Hypergraph (MRieHy), a framework tailored for online test-time adaptation in MI-BCI decoding that leverages Riemannian geometry to strengthen cross-day transferability. MRieHy first computes Riemannian means of covariance matrices from cross-day training data to align multi-day distributions. It then constructs a hypergraph over covariance matrices using Riemannian distance, complemented by a second hypergraph over deep features built with cosine similarity. The two hypergraphs are fused via adaptively learned combination weights, jointly optimized with the label projection matrices. During online testing, MRieHy maintains a first-in-first-out buffer of recent samples, performs Riemannian alignment on the buffered data, and decodes with the learned hypergraph. Extensive experiments on a private four-class ECoG dataset and two public four-class EEG datasets validate that MRieHy achieves notable performance gains over state-of-the-art baselines.

cs.LG↗
arXiv2026

Optimizing CUDA like a Human: Micro-Profiling Tools as Expert Surrogates for LLM-Based GPU Kernel Optimization

We present KernelPro, a closed-loop multi-agent system that automatically generates, profiles, and iteratively optimizes GPU kernel code by integrating large language model (LLM) code generation with hardware profiler feedback and pluggable bottleneck detection tools. KernelPro introduces four contributions: (1) a semantic feedback operator that encodes expert heuristics as pluggable micro-profiling tools, transforming raw hardware metrics into actionable natural language guidance; (2) a two-stage tool invocation architecture where roofline-based bottleneck classification filters which specialized analysis tools execute, combining kernel-level (ncu), instruction-level (SASS), and system-level (nsys) profiling; (3) a domain-adapted MCTS with progressive widening, asymmetric branching, log-reward calibration, dead-end pruning, and search memory for cross-iteration learning; and (4) direct CuTe source-level code generation via autonomous code search over the CUTLASS/CuTe codebase. On KernelBench, KernelPro achieves geometric mean speedups of 2.42x/4.69x/5.30x on Levels 1/2/3, establishing state-of-the-art performance across all difficulty levels. On VeOmni's expert-optimized MoE training kernels, KernelPro achieves 1.23x over hand-tuned Triton by generating a from-scratch raw-CUDA+CuTe Hopper WGMMA kernel. Ablation studies demonstrate that each design component independently and significantly improves optimization quality: micro-profiling tools (p < 0.0001 vs raw metrics), MCTS search (26% higher geometric mean vs greedy, p = 0.004), and proactive tool orchestration (23% improvement, p = 0.035). Finally, KernelPro is the first CUDA kernel coding agent to optimize energy efficiency beyond the speed-only focus of prior systems, demonstrating an 11.6% measured energy reduction at matched speed.

cs.LG↗
arXiv2026

DualKV: Shared-Prompt Flash Attention for Efficient RL Training with Large Rollouts and Long Contexts

Modern RL post-training methods such as GRPO and DAPO train on N response sequences of R tokens sampled from a shared prompt of P tokens, but standard FlashAttention replicates all P prompt tokens N times across both forward and backward passes -- duplicating compute and memory on identical hidden states. In large-rollout, long-context RL training (N>=16, P>=8K), this redundancy dominates the policy update cost. We observe that in decoder-only models, causal masking makes prompt representations invariant across sequences at every layer, so all per-token operations (norms, projections, MLP) and attention can process the prompt once -- a property not yet exploited at the kernel level for training. We propose DualKV, the first FlashAttention kernel variant that eliminates shared-prompt replication during RL training, via (1) fused CUDA forward and backward kernels that iterate over two disjoint KV regions -- shared context and per-sequence response -- in a single kernel launch, and (2) a data-pipeline redesign in veRL that repacks N(P+R) tokens into P+NR tokens per micro-batch, extending the token reduction from attention to the entire model by a factor rho = N(P+R)/(P+NR). DualKV is mathematically equivalent to standard attention and introduces no approximation. On Qwen3-8B GRPO training with 8xH100 GPUs (N=32, 8K-context), DualKV achieves 1.63--2.09x policy-update speedup, enables 2x larger micro-batches, and raises MFU from 36% to 76%. Similar gains hold for DAPO (2.47x speedup, 77% MFU). At 30B MoE scale on 16xH100, DualKV achieves 3.82x policy-update and 3.38x end-to-end step speedup over FlashAttention (which requires 4-way Ulysses sequence parallelism to avoid OOM). DualKV also extends to hybrid sliding/global attention with head dimension 512 (which FA2 does not support) and integrates with Ulysses sequence parallelism, demonstrated on Gemma-4-31B GRPO at 64K context.

cs.LG↗