Searcharxiv⌕ Search

arXiv subjects

Chenyang Zhao

Publications and source records attributed to Chenyang Zhao.

At least 19 recordsLinked to original sources
arXiv2026

Scientific Data Skills: Enabling Agent-Ready Scientific Data Services at Scale

Scientific data are increasingly used by AI agents, yet existing dataset representations provide limited support for reliable dataset discovery and interpretation, constraining their effective use in scientific workflows. This limitation arises because agents must search across heterogeneous repositories and reconstruct dataset-specific semantics and operating procedures from documentation designed primarily for human use. To address this limitation, we introduce the Scientific Data Skill (SciDSK), an agent-ready representation that packages dataset-specific knowledge and operational guidance as a reusable agent skill. A SciDSK integrates dataset descriptions, scientific context, file organization, task-specific usage procedures, quality checks, and provenance information while retaining the underlying data in its original repository. We define a structured SciDSK specification and develop a systematic construction pipeline that grounds each SciDSK in authoritative dataset records and associated supporting materials. We further establish the Scientific Data Skill Bank, a unified platform that publishes SciDSK resources across six scientific disciplines and supports package access, persistent identification, and traceability to source datasets. We evaluate SciDSK through a retrieval benchmark for dataset discovery and controlled cases for dataset interpretation. On the query retrieval benchmark, Agent-SciDSK achieves 80.77% Hit@1, exceeding Agent-Raw by 9.62 percentage points. Across controlled interpretation cases, the SciDSK condition satisfies 23 of 24 assessment criteria, compared with 22 under the web-page condition. These results indicate that SciDSK improves how agents locate and understand scientific datasets, providing a stronger foundation for actionable scientific data use.

cs.AI↗
arXiv2026

Emergence of cooperation: A reputation-modulated reinforcement learning

Reputation is widely recognized as a key mechanism for sustaining cooperation. However, most existing game-theoretic models treat reputation primarily as an external factor that modulates payoffs, interaction structures, or strategy update rules. In many social contexts, though, reputation operates primarily as information -- it shapes how individuals interpret their own experiences and assess the behavior of others. To bridge this gap, we propose a spatial prisoner's dilemma game grounded in the reinforcement learning paradigm, in which agents equipped with Q-learning integrate both individual and social information via a locally defined reputation metric to guide their decisions. Our results reveal that reputation-modulated learning significantly promotes the emergence of cooperative behavior, and we observe a discontinuous phase transition from full cooperation to full defection as the temptation increases. Cooperation spreads through the nucleation of cooperative clusters, whereas the disintegration of these clusters drives the system into an absorbing state of complete defection. Overall, this study demonstrates that reputation facilitates cooperation not only by providing direct incentives but also by reshaping the social information landscape that agents rely on for learning and adaptation.

physics.soc-ph↗
arXiv2026

Quantum Adaptive Sensing for Accelerated MRI

Compressed sensing accelerates MRI by reconstructing images from undersampled k-space, but performance depends strongly on sampling distribution. We propose an adaptive framework that selects Cartesian phase-encode lines sequentially using a fixed-cardinality quadratic unconstrained binary optimization (QUBO) formulation. The objective combines a preference for central k-space, signal-energy information from previously acquired measurements, and pairwise terms that encourage spatially dispersed sampling. The formulation is compatible with classical annealing and quantum-annealing hardware. Retrospective experiments used simulated eight-coil 3D MRI data; QUBO problems were solved with parallel tempering, and images were reconstructed with SENSE and total-variation regularization. At 20% and 10% sampling, the proposed method improved PSNR, SSIM, NMSE, and HFEN compared with the evaluated static Cartesian strategies, including variable-density Poisson-disc sampling, although gains varied with resolution, acceleration, and noise level. In a reduced-pool experiment, a D-Wave quantum-classical hybrid solver achieved reconstruction quality comparable to variable-density Poisson-disc sampling, demonstrating feasibility on current quantum optimization infrastructure. While these results do not establish quantum computational advantage, the direct QUBO representation provides a practical framework for adaptive MRI sampling and may benefit from future advances in quantum-annealing hardware. Prospective scanner validation and systematic quantum-classical benchmarking remain necessary.

cs.ET↗
arXiv2026

Fine-grained CLIP fine-tuning with self-annotated region alignment

Contrastive Language-Image Pre-training (CLIP) has been shown to have limitations in its fine-grained dense feature representation, due to its pre-training focusing on matching the whole image to a text description. Considering the large data and computational burden in pre-training a vision-language model from scratch, a series of works aim to enhance the fine-grained ability of CLIP through a fine-tuning scheme. However, existing works suffer from a variety of limitations: additional region annotations are usually required, which limits the semantic diversity due to the predefined categories and leads to a large effort to process the training data; and they usually sacrifice CLIP's original ability for global visual representation. To bypass these limitations, we propose SFF-CLIP (Self-annotated Fine-grained Fine-tuning for CLIP), which only uses image-text pairs as input to boost the fine-grained representation ability in the CLIP fine-tuning, while maintaining the global visual-semantic consistency. Concretely, a run-time region-phrase alignment scheme is designed, which obtains concept phrases from the input sentence, and aligns them with corresponding extracted region-based features using text-specific heat maps. Extensive experiments demonstrate that SFF-CLIP leads to significant performance improvements on fine-grained dense feature representation, as well as maintaining the performance of the original CLIP on image-level tasks. Code will be released later.

cs.CV↗
arXiv2026

CLAP: Closed-Loop Training, Evaluation, and Release Control for Domain Agent Post-training

Domain agents often face noisy business data, uncertain post-training gains, offline/application mismatch, and adapter-release risk. This paper presents CLAP (Closed-Loop Agent Post-training), a closed-loop method that converts business data into structured SFT samples, decision-preference samples, holdout sets, risk diagnostics, and release-gate records. CLAP combines data validation, target/evidence normalization, reward/KL diagnosis, offline gates, and application-chain replay to decide whether an adapter is suitable for the target application chain. On five anonymized manufacturing-scenario batches, QLoRA-style LoRA-SFT yields modest average gains: overall score increases by 0.0098, pass rate by 0.0240, and evidence accuracy by 0.0280, while hallucination and wrong facts decrease. Yet only 3 of 5 batches improve, some batches regress, and GRPO exposes high KL risks. Application-chain replay further shows that RAG is necessary for factual extraction; under the same 3B backbone and 100 replay cases, an application-RAG-oriented LoRA-SFT adapter improves value, core fields, and answer-evidence doc/page matching over base+RAG, but increases latency. These results support managing domain-agent post-training through an integrated data-training-evaluation-release loop rather than relying on training completion or a single offline score.

cs.AI↗
arXiv2026

Efficient and Trainable Language Model Test-Time Scaling via Local Branch Routing

Test-time scaling improves language-model reasoning, but existing approaches often face a difficult trade-off: long chain-of-thought sampling remains single-threaded, while sentence- or solution-level search can be computationally expensive and hard to train end-to-end. We introduce Local Branch Routing (LBR), a token-level test-time scaling framework that expands a small local lookahead tree, forwards all sampled branches through the language model, and uses a lightweight router to select the depth-1 subtree to commit. By routing over the hidden states of candidate local futures, LBR allows each token decision to use evidence beyond the root next-token distribution while avoiding full solution-level search. The resulting prune-shift-grow decoding process preserves discrete branch identities and defines a tractable tree-trajectory likelihood: newly grown nodes are counted when first sampled, and router decisions are assigned explicit probabilities. This enables end-to-end reinforcement learning with verifiable rewards, jointly optimizing the base model and router under the same likelihood-ratio principle as discrete-token RLVR. On synthetic hierarchical-planning tasks, LBR shows that post-candidate hidden states provide useful routing evidence. On mathematical reasoning benchmarks, LBR improves both Pass@1 and Pass@32 over discrete chain-of-thought, vanilla discrete-token RLVR, and RL-compatible soft-token branching baselines. These results suggest that lightweight local branching offers an efficient, trainable, and discrete form of language-model test-time scaling.

cs.CL↗
arXiv2026

The one-step Shafarevich gap in embedding dimension five

Let $k$ be an algebraically closed field of characteristic zero and let $S=k[x_1,\ldots,x_5]$ with maximal ideal $\mathfrak m=(x_1,\ldots,x_5)$. For a codimension-$r$ subspace $Q\subset S_2$, set $I_Q=(Q)+\mathfrak m^3$. Then $S/I_Q$ has Hilbert function $(1,5,r)$. We prove that the translated one-step locus defined by these ideals is contained in the smoothable component for every $r\in{6,7,\ldots,15}$. We introduce a finite field differential rank certificate proving dominance, for $6\le r\le 14$, of the Erman--Velasco map $\operatorname{GL}*5\times (\mathbb A^5)^r\dashrightarrow \operatorname{Gr}(r,\operatorname{Sym}^2 k^5)$, $(g,a^{(1)},\ldots,a^{(r)})\mapsto g\cdot\langle q(a^{(1)}),\ldots,q(a^{(r)})\rangle$, where $q(a)=\sum*{i=1}^5 a_i y_i^2-\left(\sum_{i=1}^5 a_i y_i\right)^2$. The endpoint $r=15$ is handled separately by a flat degeneration of $21$ general reduced points to the fat point defined by $\mathfrak m^3$. Combined with the known small cases and with the known elementary components for $r=3$ and $r=5$, this gives the complete one-step classification in embedding dimension five: the one-step loci with Hilbert function $(1,5,r)$ are smoothable for all $r\neq 3,5$, and the cases $r=3,5$ are precisely the generically reduced elementary component cases. In this sense the one-step Shafarevich gap in embedding dimension five is completely resolved.

math.AG↗
arXiv2026

Prefilling-dLLM: Predictive Prefilling for Long-Context Inference in Diffusion Language Models

Diffusion large language models (dLLMs) re-encode the entire prefix at every denoising step, causing recomputation that scales quadratically with context length and becomes prohibitive for long-context scenarios. We propose Prefilling-dLLM, a training-free prefill-decode disaggregation framework for dLLMs that partitions the prefix into N chunks, caches their KV representations once, and selects the top-K most relevant chunks with intra-chunk token sparsity for decoding, showing that sparse prefilling can outperform dense attention while reducing per-step complexity from quadratic in the full sequence length to quadratic only in the decode length. On LongBench and InfiniteBench, Prefilling-dLLM achieves state-of-the-art quality among dLLM acceleration methods, and an attention kernel that parallelizes decoding over the non-contiguously cached chunk KV yields 9.1--28.0x speedup at 8K--32K contexts. We further show that beginning-of-sequence tokens prepended to each chunk act as periodic attention anchors that eliminate the lost-in-the-middle phenomenon. Code is available at https://github.com/menik1126/Prefilling-dLLM.

cs.CL↗
arXiv2026

Deformation Theory and Torus-Fixed Geometry of the Nested Hilbert Scheme of Points

In this paper, we study the nested Hilbert scheme $(\mathbb{A}^2)^{[n,n+1]}=\mathrm{Hilb}^{n,n+1}(\mathbb{A}^2)$ from a combination of deformation theory, torus actions, and Young diagram combinatorics. We first recall the scheme theory and functor basics needed to define Hilbert schemes. We then use a classic result on first-order deformations to identify $T_I(\mathbb{A}^2)^{[n]}\cong \mathrm{Hom}_{\mathbb{C}[x,y]}(I,\mathbb{C}[x,y]/I)$. For a nested pair $I\subset J$, with $\dim_{\mathbb{C}}\mathbb{C}[x,y]/I=n+1$ and $\dim_{\mathbb{C}}\mathbb{C}[x,y]/J=n$, the tangent space becomes a compatibility kernel $T_{(I,J)}(\mathbb{A}^2)^{[n,n+1]}\cong \ker(\mathrm{Hom}(I,R/I)\oplus \mathrm{Hom}(J,R/J)\to \mathrm{Hom}(I,R/J))$. The torus-fixed points are indexed by a partition $\lambda\vdash n+1$ together with a removable corner $c$ of its Young diagram. This corner is not only combinatorial, but also the monomial form of a one dimensional socle direction in $R/I_\lambda$. The blow-up map to $(\mathbb{A}^2)^{[n]}\times \mathbb{A}^2$ has fibres given by projective spaces of one-dimensional quotients of $J/\mathfrak m_pJ$, whose torus-fixed points are addable boxes of the smaller diagram. These two local fibres explain how the universal family, the blow-up geometry, and Young diagram combinatorics come together in the study of the local geometry of the nested Hilbert scheme of points. Finally, we derive the tangent weight formula at a fixed point $(I_\lambda,I_{\lambda\setminus c})$ in the torus convention used in the paper. Using the standard arrow basis, we show in the proof how the arm-leg weights are modified by the compatibility kernel through a shortening rule determined by $c$. A Macaulay2 verification computes the compatibility kernel from monomial syzygies and checks the weight formula for all partitions of size at most $16$.

math.AG↗
arXiv2026

Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling

Tokens are the fundamental units of computation in modern autoregressive models, and generation length directly influences both inference cost and reasoning performance. Despite its importance, existing approaches model length primarily at the coarse sequence level. We introduce the Length Value Model (LenVM), a token-level framework that estimates the remaining generation length at every decoding step. By formulating length modeling as a value estimation problem and assigning a constant negative reward to each generated token, LenVM predicts a bounded, discounted return that is a monotone proxy for the remaining generation horizon. This value formulation provides annotation-free, dense, unbiased, and scalable supervision. Experiments on LLMs and VLMs show that LenVM supports exact control, continuous performance--efficiency steering, length prediction, and interpretation. On LIFEBench-token, it raises the exact-length score of Qwen2.5-7B-Instruct from $30.9$ to $64.8$ in one pass and reaches $83.6$ when combined with LCG. Near $200$ tokens on GSM8K, LenVM retains about $63\%$ Pass@1 versus $6\%$ under a hard token budget. Across other settings, LenVM improves matched-length quality over prompting and EOS calibration, while its length-prediction accuracy improves with scale and exceeds specialized predictors. Its token-level values also identify markers of shifts between longer and shorter trajectories. These results establish generation length as an effective scalable token-level value signal for control, prediction, and future value-based training.

cs.CL↗
arXiv2026

CodeComp: Structural KV Cache Compression for Agentic Coding

Agentic code tasks such as fault localization and patch generation require processing long codebases under tight memory constraints, where the Key-Value (KV) cache becomes the primary inference bottleneck. Existing compression methods rely exclusively on attention signals to estimate token importance, systematically discarding structurally critical tokens such as call sites, branch conditions, and assignments that are essential for code understanding. We present CodeComp, a training-free KV cache compression framework that incorporates static program analysis into LLM inference via Code Property Graph priors extracted by Joern. Across bug localization and code generation benchmarks, CodeComp consistently outperforms attention-only compression baselines under equal memory budgets, recovering the majority of full-context accuracy under aggressive KV cache compression, while matching the patch generation quality of uncompressed full-context inference and integrating seamlessly into SGLang-based agentic coding pipelines without model modification.

cs.CL↗
arXiv2026

MMSpec: Benchmarking Speculative Decoding for Vision-Language Models

Vision-language models (VLMs) achieve strong performance on multimodal tasks but suffer from high inference latency due to large model sizes and long multimodal contexts. Speculative decoding has recently emerged as an effective acceleration technique, yet its behavior in VLMs remains insufficiently understood. We introduce MMSpec, the first benchmark for evaluating speculative decoding in vision-language models. MMSpec contains 600 multimodal samples across six task categories and integrates ten representative speculative decoding algorithms under a unified evaluation framework. Our study reveals three key findings: (1) methods designed for text-only LLMs degrade in multimodal scenarios, (2) vision awareness becomes increasingly important at larger batch sizes, and (3) throughput speedup alone does not reliably reflect latency performance. Motivated by these findings, we propose ViSkip, a plug-and-play speculative decoding method that dynamically adapts speculation to vision tokens and achieves state-of-the-art performance.

cs.CV↗
arXiv2026

Balancing Multiple Objectives in Urban Traffic Control with Reinforcement Learning from AI Feedback

Reward design has been one of the central challenges for real world reinforcement learning (RL) deployment, especially in settings with multiple objectives. Preference-based RL offers an appealing alternative by learning from human preferences over pairs of behavioural outcomes. More recently, RL from AI feedback (RLAIF) has demonstrated that large language models (LLMs) can generate preference labels at scale, mitigating the reliance on human annotators. However, existing RLAIF work typically focuses only on single-objective tasks, leaving the open question of how RLAIF handles systems that involve multiple objectives. In such systems trade-offs among conflicting objectives are difficult to specify, and policies risk collapsing into optimizing for a dominant goal. In this paper, we explore the extension of the RLAIF paradigm to multi-objective self-adaptive systems. We show that multi-objective RLAIF can produce policies that yield balanced trade-offs reflecting different user priorities without laborious reward engineering. We argue that integrating RLAIF into multi-objective RL offers a scalable path toward user-aligned policy learning in domains with inherently conflicting objectives.

cs.AI↗
arXiv2026

CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use

AI agents are increasingly used to solve real-world tasks by reasoning over multi-turn user interactions and invoking external tools. However, applying reinforcement learning to such settings remains difficult: realistic objectives often lack verifiable rewards and instead emphasize open-ended behaviors; moreover, RL for multi-turn, multi-step agentic tool use is still underexplored; and building and maintaining executable tool environments is costly, limiting scale and coverage. We propose CM2, an RL framework that replaces verifiable outcome rewards with checklist rewards. CM2 decomposes each turn's intended behavior into fine-grained binary criteria with explicit evidence grounding and structured metadata, turning open-ended judging into more stable classification-style decisions. To balance stability and informativeness, our method adopts a strategy of sparse reward assignment but dense evaluation criteria. Training is performed in a scalable LLM-simulated tool environment, avoiding heavy engineering for large tool sets. Experiments show that CM2 consistently improves over supervised fine-tuning. Starting from an 8B Base model and training on an 8k-example RL dataset, CM2 improves over the SFT counterpart by 8 points on tau^-Bench, by 10 points on BFCL-V4, and by 12 points on ToolSandbox. The results match or even outperform similarly sized open-source baselines, including the judging model. CM2 thus provides a scalable recipe for optimizing multi-turn, multi-step tool-using agents without relying on verifiable rewards. Code provided by the open-source community: https://github.com/namezhenzhang/CM2-RLCR-Tool-Agent.

cs.AI↗
arXiv2026

On-Chip Erbium-Doped Tantalum Oxide Microring Hybrid Cavity Single-Mode Laser

We demonstrate a high-performance, single-mode Er:Ta2O5 microring laser monolithically integrated on a silicon platform via a customized Damascene process. The Er:Ta2O5 gain medium exhibits a low propagation loss of 0.73 dB/cm and a high intrinsic Q-factor of 5.03 x 105. By utilizing a hybrid cavity_consisting of a microring coupled to a U-shaped waveguide at two symmetric points_we exploit the Vernier effect to achieve robust longitudinal mode selection. Under a non-resonant 1480 nm pumping scheme, the laser yields a side_mode suppression ratio (SMSR) of 53.3 dB and a narrow linewidth of 9.5 pm. A slope efficiency of 2.76 % is achieved_the highest reported to date for Er:Ta2O5 lasers_with a lasing threshold of 3.3 mW. Furthermore, stable single-mode tuning is demonstrated across a temperature range of 18_68 celsius, consistently aligning with theoretical transfer matrix models. This work provides a scalable pathway for high-efficiency, tunable on-chip light sources, bridging the gap for monolithic active-passive integration on the tantalum oxide photonic platform.

physics.optics↗
arXiv2026

MMFormalizer: Multimodal Autoformalization in the Wild

Autoformalization, which translates natural language mathematics into formal statements to enable machine reasoning, faces fundamental challenges in the wild due to the multimodal nature of the physical world, where physics requires inferring hidden constraints (e.g., mass or energy) from visual elements. To address this, we propose MMFormalizer, which extends autoformalization beyond text by integrating adaptive grounding with entities from real-world mathematical and physical domains. MMFormalizer recursively constructs formal propositions from perceptually grounded primitives through recursive grounding and axiom composition, with adaptive recursive termination ensuring that every abstraction is supported by visual evidence and anchored in dimensional or axiomatic grounding. We evaluate MMFormalizer on a new benchmark, PhyX-AF, comprising 115 curated samples from MathVerse, PhyX, Synthetic Geometry, and Analytic Geometry, covering diverse multimodal autoformalization tasks. Results show that frontier models such as GPT-5 and Gemini-3-Pro achieve the highest compile and semantic accuracy, with GPT-5 excelling in physical reasoning, while geometry remains the most challenging domain. Overall, MMFormalizer provides a scalable framework for unified multimodal autoformalization, bridging perception and formal reasoning. To the best of our knowledge, this is the first multimodal autoformalization method capable of handling classical mechanics (derived from the Hamiltonian), as well as relativity, quantum mechanics, and thermodynamics. More details are available on our project page: MMFormalizer.github.io

cs.CL↗
arXiv2026

ScienceDB AI: An LLM-Driven Agentic Recommender System for Large-Scale Scientific Data Sharing Services

The rapid growth of AI for Science (AI4S) has underscored the significance of scientific datasets, leading to the establishment of numerous national scientific data centers and sharing platforms. Despite this progress, efficiently promoting dataset sharing and utilization for scientific research remains challenging. Scientific datasets contain intricate domain-specific knowledge and contexts, rendering traditional collaborative filtering-based recommenders inadequate. Recent advances in Large Language Models (LLMs) offer unprecedented opportunities to build conversational agents capable of deep semantic understanding and personalized recommendations. In response, we present ScienceDB AI, a novel LLM-driven agentic recommender system developed on Science Data Bank (ScienceDB), one of the largest global scientific data-sharing platforms. ScienceDB AI leverages natural language conversations and deep reasoning to accurately recommend datasets aligned with researchers' scientific intents and evolving requirements. The system introduces several innovations: a Scientific Intention Perceptor to extract structured experimental elements from complicated queries, a Structured Memory Compressor to manage multi-turn dialogues effectively, and a Trustworthy Retrieval-Augmented Generation (Trustworthy RAG) framework. The Trustworthy RAG employs a two-stage retrieval mechanism and provides citable dataset references via Citable Scientific Task Record (CSTR) identifiers, enhancing recommendation trustworthiness and reproducibility. Through extensive offline and online experiments using over 10 million real-world datasets, ScienceDB AI has demonstrated significant effectiveness. To our knowledge, ScienceDB AI is the first LLM-driven conversational recommender tailored explicitly for large-scale scientific dataset sharing services. The platform is publicly accessible at: https://ai.scidb.cn/en.

cs.IR↗
arXiv2025

On-chip quadratically nonlinear photodetector

Involving deterministically nonlinear photoresponse in on-chip photodetector is intriguing to develop sophisticated functions in photonic integrated circuits, such as in-sensor computing and optoelectronic mixing, though the corresponding devices are still lack of sufficient investigation. Here, we demonstrate an on-chip quadratically nonlinear photodetector (QNPD) by configuring an InSe p-i-n homojunction on a silicon waveguide. Telecom-band light guiding in the waveguide couples with the InSe evanescently and is frequency up-converted into visible light via InSe's second-harmonic generation (SHG), which is subsequently absorbed by InSe and finally generates photocurrent under the built-in electric field of the p-i-n homojunction. Governed by these sequential processes, the on-chip QNPD presents a quadratic function between photocurrent and optical power. Thanks to the efficient SHG and well-established homojunction in InSe, the QNPD reaches a high normalized responsivity of 37.1 A/W2 and low dark current of 1 pA, representing greatly improved performances among reported nonlinear photodetectors. Benefiting from the extra SHG process, the on-chip QNPD intrinsically incorporates light-light interactions, enabling straightforwardly monitoring all-optically mixing signals electrically. As an example, an array of 16-pixel QNPDs was designed to implement a fully single-shot on-chip autocorrelator without requirement of bulky optics and external cameras, which precisely measures picosecond pulses with high sensitivity of 6.1*10-10 W2.

physics.optics↗