arXiv subjects

Yi Zhang

Publications and source records attributed to Yi Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2026

WuYuEval: A Multi-Level Benchmark for Large Language Models in Solid Waste Management

Large language models (LLMs) are increasingly used as technical assistants, but their competence in solid waste management (SWM) remains difficult to assess because existing benchmarks emphasize general knowledge rather than professional decisions under engineering, environmental, and policy constraints. We introduce WuYuEval, a multi-level benchmark for evaluating LLMs in SWM across foundational knowledge, domain reasoning, and expert decision-making. After quality auditing, WuYuEval contains a Foundation Module with 4,590 closed-ended multiple-choice questions across six task types and eight domain categories, together with an Expert Module with 247 scenario-based open-ended questions involving multi-objective optimization, constraint trade-offs, and system design. For expert tasks, we combine anchor-calibrated LLM-as-a-Judge scoring with Elo-based pairwise comparison. Across 33 LLMs, performance varied widely. The leading model reached 94.64\% accuracy on the Foundation Module, but average accuracy still fell from 84.14\% on easy questions to 42.50\% on hard questions, with lower performance concentrated in calculation, experimental design, urban planning, and open-ended expert tasks. Reasoning-oriented Thinking modes improve most matched model pairs after auditing, but the gains depend on baseline capability and are not uniformly positive. These results suggest that visible deliberation helps only when it remains anchored to units, assumptions, and engineering constraints; otherwise, it may drift from decisive answer boundaries. WuYuEval therefore provides both an evaluation resource and an empirical basis for developing SWM-oriented foundation models with professional reasoning chains and explicit constraint control.

cs.CL↗

arXiv2026

UniRank: Benchmarking Ranking Models for Unified Sequential Modeling and Feature Interaction

Ranking is a core stage in online advertising and recommender systems. Modern ranking models increasingly unify sequential modeling and feature interaction, yet many advances rely on proprietary data, closed implementations, and large-scale industrial infrastructure. This setting limits reproducible comparison and hinders academic study of scaling laws, long-sequence modeling, and multi-task ranking. To address these limitations, this paper proposes UniRank, an open benchmark for ranking models that unify sequential modeling and feature interaction. UniRank uses chronological pointwise autoregressive supervision, standardizes evaluation across feedback tasks, and provides a PyTorch toolkit with Distributed Data Parallel training, operator optimization, mixed-precision training, attention optimization, and other efficiency techniques that reduce hardware requirements. We benchmark 15 representative unified ranking models on five large-scale public datasets from short-video, advertising, and e-commerce platforms, with the largest dataset containing over 700 million instances and the longest behavior sequence exceeding 10^5 interactions. UniRank provides a reproducible basis for comparing unified ranking models, studying scaling laws under limited compute, and narrowing the gap between academic and industrial ranking research. We believe UniRank benefits researchers, practitioners, and beginners through reproducible experiments, production-oriented evaluation, and accessible implementations. Code and data are available at https://github.com/salmon1802/UniRank.

cs.IR↗

arXiv2026

High-Energy Microresonator Soliton Generation

Kerr resonators generate stable frequency combs in a compact platform with applications in coherent communications, sensing, quantum information processing, and astrophysics. Ultrashort pulses can be generated, moreover, with a wavelength and repetition rate flexibility inaccessible by traditional mode-locked lasers, which is desirable for high peak-power applications including in biomedicine and materials processing. However, for these applications, single-pulse energies will need to be significantly improved beyond the fJ level characteristic of sources today. Here we describe and demonstrate a simple approach for increasing the pulse energy in anomalous dispersion Kerr microresonators. Through scaling laws based on a mean-field cavity model and supported by experimentally-accurate numerical simulations, we show that pulse energy scales strongly with output coupling if supported by sufficient drive power. In strongly over-coupled cavities, femtosecond pulses can be stabilized with pulse energy beyond the pJ level. With this theoretical framework, by pumping a 12 GHz Si3N4 cavity with 30% output coupling with 2.6-ps time-lens generated pulses, we observe stable 74-fs pulses with a record output pulse energy of 6 pJ. High energy Kerr resonators are anticipated to improve performance for current applications and compliment mode locked lasers for high peak power applications.

physics.optics↗

arXiv2026

Bayesian Signal Component Decomposition via Diffusion-within-Gibbs Sampling

In signal processing, the data collected from sensing devices is often a noisy linear superposition of multiple components, and the estimation of components of interest constitutes a crucial pre-processing step. In this work, we develop a Bayesian framework for signal component decomposition, which combines Gibbs sampling with plug-and-play (PnP) diffusion priors to draw component samples from the posterior distribution. Unlike many existing methods, our framework supports incorporating component-wise model-driven and data-driven priors into diffusion models in a unified manner. Moreover, the proposed posterior sampler allows component priors to be learned separately and flexibly combined for different decomposition tasks at inference time. Under suitable assumptions, the proposed Diffusion-within-Gibbs (DiG) sampler provably produces samples from the posterior distribution. We also show that DiG can be interpreted as an extension of a class of recently proposed diffusion-based samplers, and that, for suitable classes of sensing operators, DiG better exploits the structure of the measurement model. Numerical experiments demonstrate the superior performance of our method over existing approaches.

eess.SP↗

arXiv2026

WuYu-EnvLE-Bench: A Benchmark for Evaluating Large Language Models in Environmental Law Enforcement

Large language models (LLMs) are increasingly considered for environmental enforcement, but their ability to produce traceable enforcement decisions remains unclear. We introduce WuYu-EnvLE-Bench, a benchmark built from real enforcement cases, regulatory standards, and expert review. It contains 2,521 benchmark instances, 14 tasks, and 12 pollution-medium subdomains across pre-enforcement, in-enforcement, and post-enforcement workflows. Using Absolute Environmental Enforcement Score (AES) and Intelligent Enforcement Index (IEI), we evaluate open-source and closed-source LLMs across capability, response quality, and resource efficiency. Results show that LLMs perform well on rule-bounded tasks but remain unreliable in evidence-chain construction, contradiction detection, multi-source integration, and procedural judgment. Model scaling also shows diminishing returns: medium-sized models approach leading models in structured tasks, while larger models do not reliably overcome evidence-reasoning bottlenecks. WuYu-EnvLE-Bench highlights the need for evidence-grounded, rule-aware, and task-adaptive enforcement reasoning.

cs.AI↗

arXiv2026

STAC: When Innocent Tools Form Dangerous Chains for LLM Agents

As LLMs advance into autonomous agents with tool-use capabilities, they introduce security challenges that extend beyond traditional content-based LLM safety concerns. This paper introduces Sequential Tool Attack Chaining (\STAC), a novel multi-turn attack framework that exploits agent tool use. \STAC chains together tool calls that each appear harmless in isolation but, when combined, collectively enable harmful operations that only become apparent at the final execution step. At the core of \STAC is an automated, closed-loop pipeline that synthesizes executable multi-step tool chains, validates them through in-environment execution, and reverse-engineers stealthy multi-turn prompts that reliably induce agents to execute the verified malicious sequence. Using this framework, we generate and systematically evaluate 483 \STAC cases, featuring 1,352 sets of user-agent-environment interactions and spanning diverse domains, tasks, agent types, and 10 failure modes. Our evaluations show that state-of-the-art LLM agents are highly vulnerable to \STAC, with an average final attack success rate (ASR) of 91.2\% -- exceeding 90\% for all but one of the eight agents evaluated. We further perform defense analysis and find that existing prompt-based defenses provide limited protection. To address this gap, we propose a new reasoning-driven defense prompt that achieves the strongest initial-turn protection, cutting ASR by up to 28.8\%; however, this advantage erodes sharply under adaptive attacks, and an experience-based defense (ToolShield) proves more durable over sustained multi-turn interactions. These results highlight a crucial gap: defending tool-enabled agents requires reasoning over entire action sequences and their cumulative effects, rather than evaluating isolated prompts or responses.

cs.CR↗

arXiv2026

On Quantum Aspects of 1-Form Symmetries II: Bordism, Invertible Phases, and Anomalies

We study quantum anomalies associated with $U(1)$ 1-form symmetries from the perspective of invertible phases and bordism. We compute the oriented and spin bordism groups of the Eilenberg-Mac Lane space $K(\mathbb{Z},3)$ up to degree 8 using the Atiyah-Hirzebruch spectral sequence, resolving the relevant extension problems by geometric arguments and identifying both bordism invariants and geometric generators. We then relate these invariants to perturbative and global anomalies, and discuss physical examples and top-down constructions of the corresponding anomaly terms. For 5-dimensional theories, we find a new mixed perturbative anomaly between the $U(1)$ 1-form symmetry and spacetime diffeomorphisms, while for 7-dimensional theories we find a new $\mathbb{Z}_2$-valued discrete anomaly intrinsic to the $U(1)$ 1-form symmetry. We also discuss their boundary realizations and give new physical interpretations of these anomalies.

hep-th↗

arXiv2026

HyMobileAgent: Data-Environment Co-Scaling for Efficient GUI Agents

As large multimodal models move from understanding content to operating on digital environments, mobile GUI has emerged as a challenging and consequential testbed for digital embodied intelligence. Mobile agents operate under three coupled constraints: precise perception of complex interfaces, scalable acquisition of high-quality interaction data, and robust long-horizon decision making under compounding execution errors. This report presents HyMobileAgent, a mobile GUI agent built on Hy3.0-VL-A3B, a vision-native foundation model featuring native any-resolution input, an A3B-scale deployment budget, and a 32K context window to model extended interaction histories. Rather than relying solely on model scaling, we develop a joint data and environment centric scaling framework to address the key bottlenecks of mobile interaction. Our framework integrates a GUI perception flywheel combining mock-interface synthesis, rejection sampling, and icon-specific augmentation; a knowledge pipeline that transforms tutorial videos into structured interaction data; a million-scale action data pipeline deployed across more than 2000 sandbox and real-device instances with automated failure attribution; the PhoneWorld Mock App Factory, providing a resettable training environment with 34 mock applications and over 34000 tasks; and a structured Planning-and-Reflection mechanism with explicit dead-loop detection for reliable long-horizon execution. We also introduce a progressive training recipe consisting of mid-training, supervised fine-tuning, and reinforcement learning with task-specific reward designs.

cs.CV↗

arXiv2026

Non-Hermitian Quantum Adiabatic Algorithm

Non-Hermitian systems offer new opportunities for quantum optimization and computation. Here, we show that non-Hermitian quantum adiabatic algorithms require not only a real, gapped spectrum, but also a stable pseudospectrum. We propose a novel framework by mapping non-unitary quantum circuits to local Hamiltonian paths, thereby preserving their optimization advantages and shallow depth. While a direct non-Hermitian extension of the Feynman-Kitaev construction suffers severe pseudospectral instability, our history-decoupled construction yields both a controlled pseudospectrum and a real, gapped spectrum. Using the CK benchmark family of maximum independent set problems, we demonstrate polynomial-evolution-time non-Hermitian adiabatic computation that remains robust against perturbations. We further discuss a feasible optical implementation using coupled waveguides with an auxiliary lossy channel. Our work establishes pseudospectral stability, alongside real, gapped spectra, as key principles and a practical route for non-Hermitian quantum adiabatic computation.

quant-ph↗

arXiv2026

Adaptive Fusion Self-supervised Learning for Recommendation

Self-supervised learning (SSL) has recently attracted significant attention in the field of recommender systems. Contrastive learning (CL) stands out as a major SSL paradigm due to its robust ability to generate self-supervised signals. Mainstream graph contrastive learning (GCL)-based methods typically implement CL by creating contrastive views through various data augmentations. Despite these methods are effective, we argue that there still exist several challenges. i) Data augmentation requires additional graph convolution (GCN) or modeling operations, significantly increasing time costs. Moreover, graph augmentation disrupts the intrinsic properties of the user-item graph by randomly removing nodes/edges, while feature augmentation applies noise to all nodes, neglecting their unique characteristics. ii) Existing GCL-based methods use traditional CL objectives to capture self-supervised signals. However, few studies have explored obtaining more beneficial CL objectives from more perspectives and have attempted to fuse the varying self-supervised signals from these CL objectives to enhance recommendation performance. To overcome these challenges, we propose Adaptive Fusion Graph Contrastive Learning (AFGCL) for recommendation. AFGCL exploits structural information naturally produced during graph propagation to construct contrastive representations. Specifically, we introduce an adaptive fusion strategy that estimates the contributions of different propagation depths to the primary recommendation task and adaptively combines their representations. Furthermore, we construct an explicit representation for each observed user--item interaction and propose a fused contrastive objective. Experimental results on three public datasets demonstrate the superior recommendation performance and training efficiency of AFGCL compared with state-of-the-art baselines.

cs.IR↗

arXiv2026

FAST Discovery of $μ$Jy Radio Pulsations from PSR J2238+5903, Providing a DM Distance Anchor for the Candidate TeV Halo 1LHAASO J2238+5900

We report the first detection of radio pulsations from PSR J2238+5903, a gamma-ray pulsar spatially coincident with the extended TeV source 1LHAASO J2238+5900. Our 3000 s FAST L-band observation reveals a weak periodic signal at the known Fermi-LAT spin period, with $P=162.76568$ ms and $\mathrm{DM}=247.5\pm3.0~\mathrm{pc~cm^{-3}}$. The signal is independently confirmed by both FFT-based and Fast Folding Algorithm searches. The radiometer equation gives a flux density of $S_{1250}\simeq3\,μ$Jy, placing PSR J2238+5903 among the faintest radio-detected Fermi pulsars. Interpreting the DM with Galactic electron-density models gives $d_{\rm DM}=7.4\pm3.9$ kpc. At this distance, the LHAASO WCDA 39\% containment radius corresponds to a characteristic diameter of $\sim132$ pc, and the $>1$ TeV luminosity is $L_{\rm TeV}\simeq7.1\times10^{34}$ erg s$^{-1}$, about 8\% of the pulsar's spin-down power. The radio DM thus provides the first pulsar-specific distance constraint for assessing whether 1LHAASO J2238+5900 is a young relic-PWN / TeV-halo transition system.

astro-ph.HE↗

arXiv2026

Pelican-VLA 0.5: Attending Before Acting Benefits Generalization

In this report, we present Pelican-VLA 0.5, a unified VLA model that integrates vision-language understanding, future-frame generation, and action prediction within a single architecture. Pelican-VLA 0.5 achieves attention-level generalization: without object annotations, segmentation masks, attention supervision, or task-specific fine-tuning, its action pathway already focuses on the manipulation-relevant object and contact region. This behavior persists across unseen scenes and unseen robot embodiments, and is substantially stronger than in other open-source VLA baselines. We verify that this ability originates from the learnable Bottleneck Token inserted between perception and action: by routing task-relevant visual information through a compact bottleneck, the tokens interface induces manipulation-centric attention during pre-training and remains effective across different policy structures, including a MoT-style architecture.

cs.RO↗

arXiv2026

From Execution to Education: A Bloom-Aligned Framework for Measuring Educational Control in LLMs

We introduce a Bloom-aligned framework for measuring educational control in Large Language Models (LLMs): the ability to preserve a task's instructional intent while shifting its cognitive demand toward specified learning objectives. We apply this framework to programming tasks in computer science education to study the gap between solving tasks and adapting them for learners. Using revised Bloom's Taxonomy as an operational scale of cognitive demand, we evaluate two intervention settings: general difficulty control, where models are asked to make tasks harder or easier, and Bloom's control, where models are asked to target higher or lower Bloom's levels. We evaluate a matched Qwen3-Next model pair, comparing Qwen3-Next-80B-A3B-Instruct with Qwen3-Coder-Next across 2,520 tasks from three benchmarks. The framework reveals a robust directional asymmetry: both models reliably increase cognitive demand, but struggle to lower it. We further characterize these outcomes with semantic-delta clustering and layer-wise Fisher's Discriminant Ratio probing. Within this controlled comparison, the general model shows clearer middle-layer separability for both general difficulty and Bloom-control contrasts, whereas the coder model shows weaker separability for general difficulty and a deeper peak for Bloom-control contrasts. These results show that strong execution performance does not automatically entail Bloom-aligned educational control.

cs.CL↗

arXiv2026

EvoOMG: An Evolution-Oriented Multi-Agent Guidance Framework for Heterogeneous Legacy-and-MLO Wi-Fi Networks

The gradual deployment of Wi-Fi 7/8 multi-link operation (MLO) will lead to long-term coexistence between legacy non-MLO stations (STAs) and MLO-capable STAs in WLANs. This mixed deployment makes throughput optimization challenging because legacy STAs follow single-link contention and transmission, whereas MLO-capable STAs can exploit multiple links with richer access opportunities. Existing learning-based methods usually treat such networks as homogeneous systems and directly map the current observation to a complete MAC action, which cannot faithfully represent both legacy single-link and MLO multi-link behaviors. To address this issue, we propose EvoOMG, an evolution-oriented multi-agent guidance framework for heterogeneous legacy-and-MLO Wi-Fi networks. EvoOMG reformulates throughput optimization as a standard-constrained staged multi-agent decision problem. Each agent encodes recent channel, queue, contention, and transmission histories, first generates contention guidance, and then produces aggregation guidance conditioned on the preceding access stage and standard-specific feasibility constraints. This autoregressive design follows the Wi-Fi MAC order of ``contention before transmission'' while preserving distinct protocol behaviors of legacy and MLO-capable STAs. NS-3 evaluations show that EvoOMG improves scheduled goodput, convergence stability, and MLO link utilization over static enhanced distributed channel access (EDCA), one-step MADDPG, and independent-learning baselines, achieving substantial performance gains in representative mixed-standard scenarios.

cs.NI↗

arXiv2026

A Quadratic Vertex Threshold for Isolated Cliques in the Minimum Degree Kruskal-Katona Problem for 3-Uniform Hypergraphs

Given a set $X$ and an integer $t$, let $\mathcal{F}$ be a family of $k$-subsets of $X$. The Kruskal-Katona theorem states that if $|\mathcal{F}|\geq \binom{t}{k}$, then $|\partial_{k-1}\mathcal{F}|\geq\binom{t}{k-1}$. The minimum degree version of this problem asks: if $δ(\mathcal{F})\geq \binom{t}{k-1}$, how small can $|\partial_{k-1}\mathcal{F}|$ be? In this article, for the case $k=3$, we prove that, for every sufficiently large integer $t$, every extremal hypergraph for this problem contains an isolated copy of $K_{t+1}^3$ whenever $|X| \geq ct^2 + o(t^2)$, with the constant $c = 1 + \sqrt{928/33}$. Our proof uses a graph transformation that regularizes the neighborhood structure of extremal graphs, reducing the problem to a counting argument on the neighbors of a disjoint clique family. This gives a quadratic-order threshold for the every-extremal version of the problem, compared with the cubic-order threshold of Füredi and Zhao [SIAM J.\ Discrete Math.\ 36(4), 2022].

math.CO↗

arXiv2026

Cross-Domain Radiomap Prediction for Multi-Scatterer Environment: A Spherical-Wave-Based Approach

Radiomap prediction has found extensive applications in network planning and optimization. The radiomap is implicitly determined by electromagnetic (EM) wave propagation. The evolution of wireless communication toward higher frequency spectra has highlighted the effect of mesoscopic (i.e. wavelength-comparable scale) scatterers, which is negligible in the sub-6 GHz spectrum. To address this challenge, we develop a linear forward channel model to capture the propagation behavior in the multi-scatterer environment. The proposed model utilizes spherical-wave mode expansion to track the source radiation pattern, including scattering from a single scatterer and interactions among multiple scatterers. Both phenomena are represented as a superposition of spherical-wave modes, effectively capturing the multipath effect from a wave-based perspective. This forward model is then employed to formulate an inverse optimization problem, where the scattering responses and scatterer locations are jointly learned from sparse field measurements. Simulation results demonstrate that the proposed model accurately reconstructs and extrapolates the radiomap in both the spatial and the beam domain, further enables frequency-domain channel interpolation.

cs.IT↗

arXiv2026

Shadows of Uniform Hypergraphs under a Minimum Degree Condition

Given a set $X$ and an integer $t$, let $\mathcal{F}$ be a family of $k$-subsets of $X$. The Kruskal--Katona theorem implies that if $|\mathcal{F}|\geq \binom{t}{k}$, then $|\partial_\ell\mathcal{F}|\geq\binom{t}{\ell}$. The minimum degree version of this problem asks: if $δ(\mathcal{F})\geq \binom{t}{k-1}$, how small can $|\partial_\ell\mathcal{F}|$ be? We call a hypergraph \textit{extremal} if it achieves the minimum value of $|\partial_\ell \mathcal{F}|$ subject to the degree condition $δ(\mathcal{F}) \geq \binom{t}{k-1}$. Füredi and Zhao [SIAM J. Discrete Math. 36(4), 2022] proved that for $k=3$, $\ell=2$ and $t\ge 2$, every extremal hypergraph contains an isolated copy of $K_{t+1}^3$ when $|X| > \frac{1}{4}(t+1)^2(t+2)$. In this article, we study the general case $k > \ell \geq 2$. By developing a hypergraph transformation that combines shifting operations with antilexicographic compression, we prove that, for every integer $t\ge k-1$, there exists an extremal hypergraph containing an isolated copy of $K^{k}_{t+1}$ whenever $|X| > \frac{1}{4}(t+1)^2\binom{t-1}{\ell-2} + 3t+1$. In the case when $k=3$ and $\ell=2$, this gives the threshold $\frac14(t+1)^2+3t+1$, which is smaller than $\frac14(t+1)^2(t+2)$ for every $t\ge3$; for $t=2$, the two thresholds give the same integer condition on $|X|$.

math.CO↗

arXiv2026

Probabilistic Robustness in Medical Image Classification

Deep learning (DL) has shown strong performance in medical image classification, but its trustworthy deployment remains challenging in safety-critical clinical settings, where prediction errors under perturbations may lead to severe consequences. Existing studies mainly focus on adversarial robustness (AR) from a worst-case perspective; however, such settings may be less representative of real medical applications. In this work, we investigate probabilistic robustness (PR) as a more practical measure of model trustworthiness. To this end, we construct a set of natural corruption settings for medical image classification and systematically evaluate commonly used DL models on MedMNIST v2 dataset. Our study provides a statistically grounded perspective on assessing the trustworthiness of DL models, thereby supporting their more trustworthy deployment in medical imaging applications.

cs.CV↗