arXiv subjects

Xinyu Zhang

Publications and source records attributed to Xinyu Zhang.

At least 19 recordsLinked to original sources

arXiv2026

AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization

Auto-bidding plays an essential role in online advertising, automatically adjusting bids for advertisers to optimize their commercial goals. The emerging AI-Generated Bidding (AIGB) paradigm widely adopts generative modeling to optimize bidding strategies, yet suffers from the limited mode coverage of offline datasets and inadequate task-state understanding, hindering effective exploration of optimal strategies. Large Language Models (LLMs), with prior world knowledge and reasoning capabilities, offer a promising approach to overcome these limitations. However, directly applying LLMs to auto-bidding tasks faces inherent challenges in limited numerical precision, hallucinations, and inference latency. To address these limitations, we propose AIGB-R1, a hierarchical self-evolving auto-bidding framework aiming to enhance AI-Generated Bidding via LLMs' Reasoning capabilities, comprising a high-level Planner module for macro-level strategy planning and a low-level Executor module for fine-grained decision-making. Building upon this, we design an experience-driven self-evolving loop, enabling autonomous strategy exploration and optimization from accumulated experience. We adopt a two-stage pipeline of offline pre-training and post-training alignment, and build an interactive bidding simulation environment for strategy rollout. Furthermore, we propose Decoupled Group Relative Policy Optimization (D-GRPO) to achieve end-to-end optimization via advantage decoupling. Experimental results on a large-scale public dataset demonstrate the effectiveness of AIGB-R1.

cs.LG↗

arXiv2026

WebHarbor: Docking Real Websites for Evolving GUI Agent Environments

GUI web agents require realistic and interactive environments that remain stable during both evaluation and training. Live website evaluation suffers from changing content, access restrictions, and limited coverage of authenticated workflows, while existing synthetic environments for training often lack functional depth and visual fidelity. We propose WebHarbor, a suite of 153 executable and dockerized website environments that faithfully reconstruct the visual interfaces, content, and functional workflows of real-world web services under deterministic control. We build WebHarbor through task-driven human-agent collaboration, where environments evolve as new tasks expand their functional scope, coding agents implement the required changes, and human contributors audit visual fidelity, functional correctness, and task validity. The resulting suite comprises 3,518 tasks spanning multi-step and post-authentication workflows, where existing works often fail to cover. Beyond reliable evaluation, WebHarbor provides high-fidelity synthetic web environments for cost-efficient reinforcement learning (RL). We demonstrate that a Qwen 4B-level GUI agent trained with RL in WebHarbor generalizes to unseen tasks and performs competitively with leading open-source GUI web agents across four benchmarks. The code and data are publicly available at https://github.com/aiming-lab/WebHarbor.

cs.AI↗

arXiv2026

Spectral Condition for $μ$P under Width-Depth Scaling

Generative foundation models are increasingly scaled in both width and depth, posing significant challenges for stable feature learning and reliable hyperparameter (HP) transfer across model sizes. While maximal update parameterization~($μ$P) has provided a principled solution to both problems for width scaling, existing extensions to the joint width-depth scaling regime remain fragmented, architecture- and optimizer-specific, and often rely on technically involved theories. In this work, we develop a simple and unified spectral framework for $μ$P under joint width-depth scaling. For deep residual networks whose residual blocks contain $k$ transformations, the framework specifies how the norms of weights and their per-step updates should scale with width and depth. It reveals a fundamental transition from $k=1$ to $k\geq 2$, {unifying previously disparate $μ$P formulations and identifying the CompleteP-style parameterization ($k\geq 2$)} as more appropriate for practical architectures with multi-transformation branches such as Transformers. Building on this framework, we derive a general recipe for implementing $μ$P across a broad class of optimizers by mapping spectral constraints to {concrete, optimizer-specific HP parameterizations}, recovering existing results and extending them to additional optimizers. {Finally, experiments on GPT-2 style language models show that the CompleteP-style parameterization ($k\geq 2$) achieves stable feature learning and robust HP transfer under width-depth scaling, whereas standard parameterization and the Depth-$μ$P-style parameterization ($k=1$) often fail to do so.} These results support the practical effectiveness of the proposed spectral framework.

cs.LG↗

arXiv2026

Regional Anchor Routing for Multi-Object Motion Transfer

Compositional motion transfer requires a model to extract motions from multiple reference videos, and apply each motion to a user-specified object in one target scene. Existing methods normally struggle to reliably assign different reference motions to the correct target objects, thus a motion intended for one object may affect another. Meanwhile, some motion-specific tuning approaches are still facing another issue: they must run a new optimization whenever the reference video changes, making repeated transfers inefficient. To address these limitations, we propose RART, a framework that unifies object-wise motion extraction, target-region anchoring, and frame-aligned injection into a frozen video Diffusion Transformer for multi-object motion transfer. For each reference motion, RART extracts temporally ordered features from the masked reference video, and then conditions them on the target first-frame mask to form mask-anchored motion tokens. Frame-aligned motion cross-attention allows each generated frame to read only the motion tokens at the corresponding time slots, composing all object motions in one denoising process. During training, a motion-attention routing loss encourages each target region to attend to its corresponding motion tokens and reduces interference between objects. Therefore, RART can handle novel reference-target combinations without per-video optimization. Experiments show that RART achieves accurate motion preservation, efficient inference, and strong compositional generalization.

cs.CV↗

arXiv2026

PulmoWorld: An Asynchronous Multimodal Medical World Model for Longitudinal Pneumonia Dynamics

Pneumonia changes across respiration, inflammation, and organ function, while the measurements that describe it arrive at different times and are often incomplete. We introduce PULMOWORLD, a world model for longitudinal pneumonia dynamics that connects asynchronous state estimation, recursive simulation, and evidence-constrained planning. History-aware filtering separates state propagation from observation-driven correction, so a measurement updates the shared belief only when it is observed. That belief supports recursive forecasts of clinical values and modality representations, and summaries of future prognosis. Cap-MBRL searches imagined continuations inside an authorized action set and replaces a frozen fallback only when a candidate meets the risk budget and a predicted-cost margin. On the pneumonia cohort MASS, recursive forecasts are competitive with task-trained sequence models, and simulated future states retain prognostic information. Trained separately on public MIMIC-IV-CXR chest radiographs and clinical trajectories, this model improves prognostic discrimination over latest-value fusion in development and is competitive on long-horizon error on the locked pneumonia panel. In five known-mechanism MASS-Syn environments, Cap-MBRL lowers selection regret relative to direct fallback execution.

cs.CV↗

arXiv2026

Agent Memory as Recommendation: Recommender-System Design for Long-Term Memory of LLM Agents

Large language model (LLM) agents that serve users across many sessions must choose, for every request, the few stored memories that enter a limited context window. This choice largely decides whether the agent answers correctly, and it runs on every request at Web scale. This selection faces three challenges: a large and growing store, a per-request cost budget, and multi-session questions that need all of their evidence. Many memory systems build notes or graphs with hundreds to thousands of LLM calls per history, and each is designed on its own, so it is unclear which component yields the accuracy. These challenges mirror the properties of recommendation, so we formalize agent memory as budgeted slate recommendation, with the request as the user, stored memories as items and the context window as a slate. The formalization places memory systems in one design space of recommender cascades and factorizes accuracy into the completeness that each stage controls. Guided by it, we propose AgentMemRec, which adapts candidate generation, learning to rank and slate construction from recommender systems and selects memories without any LLM call. On multi-session questions of LongMemEval, AgentMemRec outperforms memory and context-selection methods, including A-Mem, HippoRAG 2, DynamicRAG and T-Mem, by 10 to 31 points of accuracy. It matches an LLM reranker that makes 64 calls per query at 3.2 times lower serving cost, and on LoCoMo it outperforms Mem0 and Zep without any LLM call at write time. The factorization attributes the gains to slates that more often hold all of the evidence, so recommender-system design offers an effective and efficient foundation for the long-term memory of LLM agents.

cs.IR↗

arXiv2026

Knowing When the Evidence Is Enough: Requirement-Directed Search with Horn-Closure Certificates

After each Web search, a search-augmented model must decide whether its evidence suffices: answer, search again, or abstain. Prevailing systems leave this to holistic model judgement, which can accept a context naming every entity yet lacking one required fact. We instead make sufficiency structural: a per-query requirement hypergraph states which facts jointly yield another fact. This raises two challenges: deriving requirements from the question alone, and linking each requirement to passages about the right entity. Certify therefore decomposes the question, searches per requirement, and accepts links only after entity checks. Horn closure over the links yields a certificate, the minimal sufficient passages and their derivation, or an abstention naming open requirements. Given correct retrieval, requirements and links, the decision is sound, and complete up to the entity check. Pooled over three corpora with unanswerable twins, Certify decides significantly more accurately than search agents, graph retrieval and sufficiency judges in default configurations. On the same evidence, it matches a proprietary model's sufficiency label without an additional model call. Over 21M Wikipedia passages, entity checks add 4 to 13 points of decision accuracy where gold evidence exists. Finally, answers from its certificate are 80% correct, against 57% from all retrieved passages.

cs.IR↗

arXiv2026

Cwikel estimates and Weyl's laws for $L\log L$ potentials on quantum tori

We prove an $L\log L$-endpoint Cwikel estimate on quantum tori $\mathbb T_θ^d$, i.e., for every $d\geq 2$ and $x\in L\log L(\mathbb T_θ^d)$, \begin{equation*} \big\|(1-Δ)^{-\frac{d}{4}}λ(x)(1-Δ)^{-\frac{d}{4}}\big\|_{\mathcal{L}_{1,\infty}} \leq C_d \Norm{x}_{L\log L(\mathbb T_θ^d)}, \end{equation*} where $λ(x)$ is the left multiplication by $x$. Our argument actually gives an abstract weak Hilbert--Schmidt estimate for multiplication operators in finite von Neumann algebras whose $L_2$-spaces admit orthonormal bases of unitaries. As a consequence, it yields a critical Sobolev embedding $W_2^{d/2}\hookrightarrow\exp(L_2)$ on quantum tori. When $θ=0$, our method also gives a simpler proof of the $L\log L$ Cwikel estimate on classical tori. As an application, we use the $C^*$-algebraic principal symbol map to prove Weyl's laws for sandwich operators \[ S(1-Δ)^{-d/4}λ(x)(1-Δ)^{-d/4}T, \] where $x\in L\log L(\mathbb T_θ^d)$ and $S,T\inΠ$, the $C^*$-algebra generated by $λ(C(\mathbb T_θ^d))$ and Fourier multipliers with continuous homogeneous symbols of degree zero. We also prove a semiclassical Weyl's law for critical Schrödinger operators with $L\log L$ potentials.

math.OA↗

arXiv2026

Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents

Open-ended tool agents can complete the same objective through different tool paths, and their quality depends jointly on the interaction trace, environment state, and final artifacts, so such tasks rarely admit an executable ground-truth reward. The central challenge is therefore to construct a reward from existing experience that exposes the current policy's failures while remaining stationary during optimization. We introduce Reward Engine, which, like a self-motivated student, first works out what success requires and then optimizes toward it. From the task and initial trajectories, it builds a RewardPack of testcases for required correctness and rubrics for graded quality, each executed as a check that returns a judgment with evidence. The frozen RewardPack then guides the agent to revise and select its outputs, and can also serve as a reward for reinforcement learning. On WorkBuddy Bench Office and Claw-Eval, the constructed reward largely agrees with the benchmark graders, not reversing their ordering on 82.3\% and 80.5\% of candidate pairs. Used to revise executions, it raises three-run consistency ($\mathrm{Pass}^{3}$) over Baseline from 45.0\% to 54.0\% and from 78.4\% to 83.7\%.

cs.AI↗

arXiv2026

In-Context Batch Black-box Optimization via Flow Matching

Batch black-box optimization is important for expensive systems where many candidates can be evaluated in parallel, yet only a few rounds of evaluation are affordable. In this regime, effective batch construction must identify multiple promising regions while avoiding redundant evaluations, and conventional Bayesian optimization can require increasingly difficult batch-selection procedures as the batch size grows. We introduce Flow Matching for Thompson-Sampling Refinement (FMTR), an amortized approach that learns offline a continuous, history-conditioned proposal distribution from which variable-size batches can be sampled. FMTR is trained offline by conditional flow matching on Thompson-sampling argmaxes refined toward high-value regions via a horizon-dependent Langevin pushforward map, adapting the exploration-exploitation tradeoff over the optimization horizon. At test time, the model parameters remain fixed and batch generation requires neither task-specific surrogate refitting nor numerical acquisition optimization. FMTR achieves strong performance against batch BO and in-context baselines on synthetic and real-world benchmarks, while substantially reducing the computational cost of batch selection.

cs.LG↗

arXiv2026

A practical route to amorphous solid-state electrolyte structures by hybrid reverse Monte Carlo with foundation potentials

Amorphous alkali oxychlorides are fast ion conductors, but their atomic structure remains unresolved. Total scattering sums all partial pair correlations into one curve that has no unique inversion, so the structure must be modeled. Melt-quench molecular dynamics builds a model from the interatomic potential alone, and reverse Monte Carlo (RMC) builds one from the measurement alone. The two have rarely been compared with all else held fixed. We benchmark both routes and two hybrid schemes on amorphous \ce{Li_{0.9}NbO_{0.9}Cl_{4.1}} against one neutron pair distribution function with one foundation potential. The hybrid schemes put the energy either in the acceptance rule (inline) or in short relaxations between blocks of refinement (alternating). Each refinement starts from a melt-quench glass or from a random packing of isolated units. The glass has the lowest energy and places the first-shell peaks correctly, but carries almost none of the oscillation beyond 6~Å (goodness of fit $R^{2} = 0.29$). Pure RMC fits the measurement to within its counting noise ($R^{2} = 0.99$), but at more than $1.3$~eV/atom above the glass, and niobium loses about half of its first-shell chlorine. Started from the glass, both hybrid schemes reach $R^{2} = 0.92$ within $0.1$~eV/atom of its energy and keep its coordination chemistry. Structures from the two inputs fit almost equally well and differ in energy by a tenth of the RMC penalty, yet they differ in whether their polyhedra are connected. The input is therefore part of the method. We recommend quenching a glass with a foundation potential and refining it against the measurement with a hybrid scheme. The resulting structure carries both the local chemistry and the medium-range order, and it is the starting point for modeling ion transport in these conductors.

cond-mat.mtrl-sci↗

arXiv2026

Physical Time as a Missing Dimension of Robot Policy Memory Dynamics

Existing robot policy memories typically evolve with observation sequences, implicitly treating each update as the same amount of physical time. Yet identical observation histories can span different durations and require different actions, exposing physical time as a missing dimension of policy memory. We introduce Temporal Flow Policy (TFP), which learns memory dynamics to retain past events and evolve the memory state over measured time intervals. Observations and prior memory determine the content and time scales of each transition; elapsed time determines how far the state advances. TFP conditions diffusion and flow-matching policies through this memory state and is trained with their native imitation objectives. We evaluate TFP with Diffusion Policy and $π_{0.5}$ through controlled comparisons of memory designs. These experiments show that successful timing depends on both measured intervals and retained history. TFP achieves higher average success than the recurrent and state-space memories we equip with the same intervals and than a variant that feeds the interval to a learned gate. Memory readouts recover the time since an event and motion information after occlusion. Changing the supplied intervals systematically delays or advances actions, linking time inputs to behavior. Real-robot tasks requiring timing or memory, together with LIBERO evaluations, demonstrate benefits in physical and general manipulation. These results suggest viewing robot memory not as stored observation history, but as a state that retains past events and evolves with physical time.

cs.RO↗

arXiv2026

Endpoint Schatten class properties of commutators on quantum tori

We study endpoint Schatten class properties of commutators $[I_θ^ε,M_x]$ on quantum tori of dimension $d\geq2$, where $I_θ^ε$ is the Riesz potential of order $ε$ and $M_x$ denotes the left multiplication operator by an operator $x$ on quantum tori $\mathbb T_θ^d$. For $-\frac{d}{2}<ε<1$, $ε\neq0$, and $x\in L_1(\mathbb T_θ^d)$, we prove that $[I_θ^ε,M_x]\in \mathcal L_{\frac{d}{1-ε},\infty}$ if and only if $x\in\dot W_{\frac{d}{1-ε}}^1(\mathbb T_θ^d)$. We also determine the singular value asymptotics of this commutator for $x\in\dot W_{\frac{d}{1-ε}}^1(\mathbb T_θ^d)$. The necessary Sobolev condition holds throughout $1-d<ε<1$, $ε\neq0$. For $ε\leq-\frac{d}{2}$, we characterise Hilbert--Schmidt membership, with a logarithmically weighted Fourier condition at $ε=-\frac{d}{2}$. In order to get the singular value asymptotics of commutators, we prove singular value asymptotics for pseudodifferential operators of arbitrary negative order on quantum tori using $C^*$-algebraic principal symbols. This asymptotic result confirms Conjecture~7.1 of McDonald and Ponge (\emph{Adv. Math.} 412 (2023), 108815).

math.OA↗

arXiv2026

PA-LOCO: Learning Perturbation-Adaptive Locomotion for Quadruped Robots

Numerous locomotion controllers have been designed based on Reinforcement Learning (RL) to facilitate blind quadrupedal locomotion traversing challenging terrains. Nevertheless, locomotion control is still a challenging task for quadruped robots traversing diverse terrains amidst unforeseen disturbances. Recently, privileged learning has been employed to learn reliable and robust quadrupedal locomotion over various terrains based on a teacher-student architecture. However, its one-encoder structure is not adequate in addressing external force perturbations. The student policy would experience inevitable performance degradation due to the feature embedding discrepancy between the feature encoder of the teacher policy and the one of the student policy. Hence, this paper presents a privileged learning framework with multiple feature encoders and a residual policy network for robust and reliable quadruped locomotion subject to various external perturbations. The multi-encoder structure can decouple latent features from different privileged information, ultimately leading to enhanced performance of the learned policy in terms of robustness, stability, and reliability. The efficiency of the proposed feature encoding module is analyzed in depth using extensive simulation data. The introduction of the residual policy network helps mitigate the performance degradation experienced by the student policy that attempts to clone the behaviors of a teacher policy. The proposed framework is evaluated on a Unitree GO1 robot, showcasing its performance enhancement over the state-of-the-art privileged learning algorithm through extensive experiments conducted on diverse terrains. Ablation studies are conducted to illustrate the efficiency of the residual policy network.

cs.RO↗

arXiv2026

Towards Merging-conditional Backdoor Unalignment against Large Language Models

Model merging has become widely adopted for enhancing the task-specific performance of Large Language Models (LLMs) without additional training. While the specialized model from third-party platforms may undergo safety audits (e.g., backdoor detection), we reveal a novel and critical threat: One-sided Merging-Conditional Backdoor (OMCB). OMCB remains dormant and undetectable within one standalone model but activates upon merging, triggering safety unalignment. We first theoretically analyze the existence and geometric location of OMCB, revealing that the parameter shift crossing safety decision boundaries is the main cause of backdoor activation. Building upon this insight, we propose SMOKE, a meta-learning-based attack framework. SMOKE implants OMCB via a bi-level optimization strategy: the inner loop synthesizes surrogate meta-vectors through random low-rank initialization and finite-step maximization, while the outer loop optimizes the target model by simulating the merging process. Furthermore, preliminary analysis shows that the injected OMCB has bounded generalization loss under unknown merging configurations. Extensive experiments across diverse model architectures and merging techniques validate the effectiveness of SMOKE and its resistance to existing backdoor detections, highlighting a significant yet previously overlooked vulnerability in the model merging paradigm.

cs.CR↗

arXiv2026

When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization

Weight-only post-training quantization (PTQ) relies heavily on reconstruction loss minimization to preserve model quality at low precision. We show that the weights favored by minimizing this loss need not yield better model performance on new tasks. In fact, we find that lower reconstruction loss can even degrade model performance on the same calibration data. Our analysis further shows that weights with lower reconstruction loss on calibration data can have higher loss than other weights when the distribution of input activations changes. Motivated by these observations and our analysis, we propose Distributionally Robust Quantization (DRQ), a post-hoc refinement process that minimizes worst-case reconstruction loss over a constrained set of input activation distributions. DRQ refines the integer codes representing quantized weights within the existing quantization grid, keeping quantization parameters and inference operators unchanged. Extensive experiments show that DRQ improves models quantized by six representative PTQ methods, including AWQ, GPTQ, and ParoQuant, and delivers gains across both dense and mixture-of-experts large language models. These results establish DRQ as a general post-hoc refinement framework for weight-only PTQ, achieving better downstream performance without adding inference overhead.

cs.AI↗

arXiv2026

TutorLoop: Regulating Student Learning Behaviors via Sensor-in-the-Loop Generative Feedback

We present TutorLoop, a sensor-in-the-loop system that regulates student learning behaviors by delivering adaptive feedback based on real-time cognitive states. Unlike prior large language model (LLM) tutors that directly depend on scenario-specific content, TutorLoop operates on sensor-derived signals captured via webcams. Moreover, unlike direct cognitive-to-feedback mappings that are short-sighted, the system employs a deep reinforcement learning (DRL) agent to optimize the feedback type across the entire learning process. Finally, another LLM tutor refines feedback into human-like, context-aware messages. We evaluate TutorLoop in a large-scale user study (N=187), where a model trained offline is directly applied to a new learning task without retraining. Results show that TutorLoop provides less frequent yet more effective interventions, improving attention, reducing workload, increasing engagement, and ultimately enhancing learning outcomes. These findings highlight the potential of closed-loop, sensor-driven feedback for scalable human-AI integrated systems to support learning.

cs.HC↗

arXiv2026

Confidence-Ordering Reversal under Contextual Priors in Neural Decoding

Contextual priors improve neural-to-language decoding by reshaping candidate scores. However, confidence is read from the same reshaped scores, so the errors a prior leaves behind can become more confident with no change in accuracy to reveal it. We study how a prior shapes confidence in speech retrieval on MEG-MASC and MOUS using local decoding scores, a contextual prior combined by additive shallow fusion, and the fused top-two margin as confidence. Among initially incorrect predictions, we find a confidence-ordering reversal: a larger margin makes a repair more likely when the correct candidate starts near the top of the local ranking, but less likely when it starts lower. On MEG-MASC, pooled correctness AUROC is 0.87, yet AUROC separating repairs from residual errors falls from 0.70 at initial ranks 2-3 to 0.39 at ranks 21-50. Errors starting beyond rank 20, inside the reversed region, make up 46.6% of all post-fusion errors. We propose a score-level account: a repair must first close the correct candidate's initial deficit, limiting its final margin, whereas a residual error can build a large margin between two incorrect candidates. A causal intervention that changes only the fusion weight moves the reversal to deeper ranks as predicted. Under a word-level LM prior, it keeps moving after accuracy gain peaks, so a weight chosen for accuracy does not settle confidence. Reading local and prior scores separately improves selective decoding: the decoder answers on 74.5% of windows instead of 56.7%, while 92% of output sets still contain the correct candidate. Confidence after contextual fusion should retain the local and contextual evidence behind each prediction, not just the fused scores. Project website: https://confidencereversal.github.io/; Code: https://github.com/AmadeusFake/NeuDecodingConfReversal

cs.AI↗