arXiv subjects

Peng Zhao

Publications and source records attributed to Peng Zhao.

At least 19 recordsLinked to original sources

arXiv2026

P-SER: Physical Consistency-Constrained Scattering Expandable Representation for Class-Incremental HRRP Target Recognition

High-resolution range profiles (HRRPs) provide compact radar target representations but exhibit strong aspect sensitivity, complicating class-incremental recognition when historical data are unavailable. Changes in viewing aspect shift scattering-peak locations and amplitudes, while reflector groups may produce responses resembling ships. Existing dynamic expansion methods can also incur substantial parameter and memory overhead. To address these challenges, we propose P-SER, a Physical Consistency-Constrained Scattering Expandable Representation framework for exemplar-free class-incremental HRRP recognition. P-SER reconstructs dominant scattering points from complex-valued sum- and azimuth-difference-channel HRRPs and observation geometry. The reconstructed spatial structure, target extent, and energy distribution are used to derive class-level scattering prototypes and sample-level saliency priors, providing physically grounded constraints for feature learning and old-class knowledge retention. A primitive scatterer dictionary further characterizes local strong-scattering mechanisms and calibrates uncertain ship--reflector predictions at inference. Meanwhile, lightweight residual adapters are incrementally introduced into a shared HRRP backbone to learn new classes with limited model growth. Experiments on simulated and measured HRRP datasets show that P-SER delivers competitive recognition performance without retaining old-class exemplars, while maintaining physical consistency, interference robustness, knowledge retention, and model scalability.

cs.CV↗

arXiv2026

Polarimetric Scattering-Structure Contrastive Learning for Few-Shot HRRP Recognition

Few-shot high-resolution range profile (HRRP) recognition is challenged by strong aspect-dependent variation and sparse novel-class support. We propose polarimetric scattering-structure contrastive learning (PSS-CL), which supplies an episodic HRRP encoder with a structured physical training view. Dominant local responses are represented by their soft affinities to canonical polarimetric mechanisms, normalized range locations, and energy contributions. A symmetric cross-view objective transfers this scattering structure to the learned HRRP representation, while episodic prototype learning supports novel-class inference without the physical branch. We conduct Random-aspect and Cross-aspect experiments on a six-class measured dataset. In Cross-aspect 1-, 3-, and 5-shot recognition, PSS-CL achieves 81.73%, 86.92%, and 90.98% mean class accuracy (MCA), respectively, outperforming the best competing results by 5.62%, 3.09%, and 3.56%.

cs.CV↗

arXiv2026

Search-Aware On-Policy Self-Distillation with Sibling Alignment for Generative Recommendation

Generative recommendation retrieves items by autoregressively generating semantic IDs, typically via beam search. On-policy self-distillation (OPSD) enhances training supervision by transferring privileged self-teacher knowledge at student-generated states. However, beam-search retention depends on competition among prefixes: visited states differ in importance, and initially reliable decisions may approach the pruning boundary as training evolves. Moreover, conventional distribution matching couples target-token confidence with preferences among competing continuations, potentially weakening relational supervision when the self-teacher strongly favors the target. To address these challenges, we propose Search-Aware On-Policy Self-Distillation (SA-OPSD), a framework that transfers privileged self-teacher knowledge to the states visited during the student's own beam search. Specifically, we introduce Evolving-Boundary Sampling, which prioritizes search-critical states by considering both current pruning-boundary proximity and initial-reference retention-margin decline under a controlled distillation budget. Furthermore, we develop Confidence-Decoupled Branch Distillation, which separates target-probability calibration from informative sibling relations to better preserve the teacher's preferences among competing branches. Experiments on Industrial, Office, and Instruments demonstrate strong performance for SA-OPSD.

cs.LG↗

arXiv2026

LegoSR: A Unified Framework for Self-improving Agent-Guided Scientific Discovery

Equation discovery solvers often use different data interfaces, equation representations and evaluation criteria, making their results hard to compare. We present LegoSR, a unified framework for algebraic symbolic regression and differential equation discovery. LegoSR integrates seven solvers through a common equation representation and evaluation process, and supports searches constrained by partially known equations. Researchers can run solvers directly or use LegoSR Agent, which selects solvers and adapts searches using solver knowledge, intermediate results and lessons from past tasks. On noisy ordinary and partial differential equation benchmarks, LegoSR Agent recovers nearly as many equations as the union of seven expert-tuned solvers and outperforms a generic agent with the same model, tools and budget by up to 29 percentage points. Three examples of increasing difficulty show the agent's role growing from running a solver to choosing one and then to building a partially known equation from failed searches.

cs.AI↗

arXiv2026

SparseGDN: Chunk Sparse in Linear Attention

Gated Delta Networks (GDNs) offer linear-time sequence modeling, yet long-context prefill remains costly due to dense intra-chunk computation, recurrent-state updates, and intermediate state storage. We introduce SparseGDN, a hardware-aware framework that combines structured intra-chunk approximation with input-dependent recurrent-event selection while producing outputs for every token. A block-adjacent solver reduces local Delta-solve computation, while a dynamic selector allocates recurrent-update budgets across head pairs. The GPU implementation materializes updates only for selected events and reconstructs token outputs using compact state snapshots and group-local interactions, reducing computation and memory traffic. We evaluate SparseGDN on Qwen3.5-4B and Qwen3.5-9B across RULER tasks and context lengths from 4K to 128K. Retaining approximately 29% of recurrent events on average, SparseGDN achieves 1.50-1.94x GDN-core prefill speedups over native FLA kernels, including online selection overhead.

cs.LG↗

arXiv2026

Dependence-Aware Risk Assessment and Optimal Design of AI-Agent Safeguard Systems

Large language model (LLM) agents increasingly rely on safeguards to prevent unauthorized actions while preserving legitimate service. However, existing risk assessments often focus on individual safeguard performance while overlooking dependence among safeguard decisions. This paper develops a dependence-aware framework for risk assessment and optimal design of multi-component AI safeguard systems, jointly selecting configurations and operating policies for sequential layered screening and selective verification. We establish stochastic comparisons of layered-screening error probabilities and invocation costs under lower-orthant ordering. For selective verification, we characterize exactly which policies have error probabilities and review costs that respond monotonically to stronger dependence under supermodular ordering. For a fixed first stage and reviewer, we show that the optimal routing map is a two-threshold rule in the first-stage likelihood ratio. An accompanying open-source software package, Dependence-Aware Policy Optimization Algorithms (DAPOA), facilitates the practical implementation of the proposed framework. Numerical experiments show that ignoring dependence can lead to inaccurate risk assessments and infeasible safeguard designs, with independence-based selections violating security requirements in six of eight scenarios. An empirical study using four LLM safeguards on benchmark-derived agent actions further demonstrates the practical implications of dependence modeling. Under the principal operating setting, dependence-aware selection lowers held-out total cost by 39.6% relative to independence-based selection.

cs.AI↗

arXiv2026

WebViSE: Scaling Visual Design Intelligence from the Open Web for LLM Website Generation

Large language models have made remarkable strides in website development, enabling complete and interactive web experiences to be built from natural-language instructions with far less effort. Yet, non-expert users often struggle to articulate sophisticated visual and interaction designs in natural language. Meanwhile, the open web already contains a vast reservoir of such design knowledge, embodied in mature website implementations. To this end, we introduce Website-derived Visual Skill Extraction (WebViSE), a framework that automatically abstracts these implementations into reusable Visual Skills and builds a structured Skill library. To scale the library, WebViSE organizes its Skills in a multi-parent DAG and jointly evolves Skill descriptions and DAG structure using Skill-conditioned Potential Queries, enabling efficient matching between diverse user instructions and relevant Skills. The auto-evolved structure outperforms non-hierarchical LLM retrieval baselines at only 39.09% of the estimated retrieval cost with Qwen3.8-Max. For any models, WebViSE can be directly injected as plug-and-play guidance, enabling diverse models to produce more visually compelling and aesthetically refined web pages. For open-source models, WebViSE can be used to build a data-synthesis pipeline that internalizes visual design intelligence into model weights. Experiments across multiple web development benchmarks and human evaluation consistently show that WebViSE improves the visual quality and interaction fidelity of artifacts. Overall, our work establishes an automated path for extracting, organizing, and transferring design intelligence from the open web to website-building models. We release all extracted Visual Skills, the associated Skill DAG, and our code at https://github.com/YifeiSun-0710/WebViSE.

cs.AI↗

arXiv2026

A Simple and Unified Method for Gradient-Variation Regret

Gradient-variation online learning has attracted increasing attention for its connections to game theory and optimization. It has been studied in both non-stationary online learning, which aims to minimize dynamic regret against time-varying comparator sequences, and universal online learning, which seeks to adapt to unknown loss curvature. Prior work obtained gradient-variation bounds using different techniques tailored to each setting. We propose a simple and unified method based on surrogate losses with a quadratic penalty on decision movement, yielding gradient-variation bounds in both settings. For non-stationary online learning, the method induces an optimistic-to-gradient-variation reduction, enabling a broad class of algorithms to match the rate established by Zhao et al. (2024). For universal online learning, it yields a simple algorithm that matches the optimal universal regret bounds of Yan et al. (2024) while relaxing the extended smoothness assumption. Finally, we apply our method to online games and improve on existing guarantees.

cs.LG↗

arXiv2026

JetSpec++: Foresight Parallel Drafting with Self Refinements

Speculative decoding's effectiveness is governed by two quantities: the acceptance rate of drafted tokens and the cost of producing them. We show the two objectives can be co-optimized with algorithm-system co-design, and present JetSpec++, which retains a strong learned draft model while hiding its cost almost entirely. JetSpec++ combines three components: intra-block self refinement, which trains the draft model to iteratively correct its own parallel predictions within a block of tokens; inter-block foresight drafting, which begins the next block before the current one is verified; and pipelined drafting and verification, which overlaps the two stages so drafting's marginal wall-clock cost approaches zero. In comparison with state-of-the-art speculative decoding methods, JetSpec++ attains a wall-clock speedup of up to $7.45\times$ over autoregressive (AR) decoding and $1.54\times$ over DSpark, with up to 1,361 tokens per second for a single request on GB200, and an average acceptance length of 8.94 across seven benchmarks, while preserving the target model's output distribution. Our code and models are available at https://github.com/hao-ai-lab/JetSpec.

cs.CL↗

arXiv2026

Thermodynamic topological classification of magnetically charged slowly rotating Kerr black holes in nonlinear electrodynamics with a cosmological constant

We provide an explicit rotating black-hole realization of the predicted $\overline{W}^{1-}$ thermodynamic topological subclass by studying a slowly rotating, magnetically charged Kerr black hole in nonlinear electrodynamics (NLED). Using a reduced fixed-parameter off-shell potential, we derive the inverse-temperature curve analytically rather than relying solely on numerical plots. In AdS spacetime, this curve begins at a finite nonzero value, increases monotonically, and diverges at large horizon radius; it therefore contains exactly one defect with $w=W=-1$ and has no stable large-black-hole branch. Positivity of the analytic coefficient establishes this result throughout the admissible slow-rotation parameter domain. Under the same reduced prescription, the Kerr-AdS benchmark instead has two branches and belongs to the $W^{0-}$ class, establishing a clear topological distinction between the NLED-Kerr and classical rotating cases. In dS spacetime, the formally extended off-shell curve remains monotonic and any isolated zero has $w=-1$; because the parameter-dependent horizon-admissible range is not fixed in the present reduced analysis, no physical dS asymptotic subclass is assigned.

gr-qc↗

arXiv2026

Achieving room temperature plasticity in TiO2 ceramics by flash processing

Ceramics are indispensable materials for modern engineering and high-technology applications. However, the intrinsic brittleness at room temperature can cause catastrophic failure and severely limit their widespread applications. Here, we report a flash processing (FP) method based on the coupling of electric and thermal fields to significantly improve the room-temperature plasticity of TiO2 ceramics with a large plastic compressive strain of 15.4% during micropillar compression. This large room-temperature plasticity is attributed to the proliferation of stacking faults and twin boundaries during the loading process, which may be assisted by the high concentration of oxygen vacancies introduced by the FP method. Our findings suggest that FP can be a simple and effective method for achieving improved room-temperature plasticity of ceramic materials through defect engineering.

cond-mat.mtrl-sci↗

arXiv2026

JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting

Speculative decoding (SD) accelerates autoregressive Large Language Models (LLMs) by drafting multiple tokens and verifying them in parallel, but it faces a scaling limitation: increasing the draft budget improves speed only when acceptance remains high and drafting overhead stays low. This ceiling has been difficult to break because prior head-based SD methods face a causality-efficiency dilemma. Autoregressive drafters produce path-conditioned candidates that are effective for tree speculative decoding with higher acceptance length, but their drafting cost grows with tree depth. Bidirectional block-diffusion drafters generate all positions in one pass, but their branch-agnostic marginals can form individually plausible yet mutually inconsistent trees, wasting budget and reducing acceptance. We propose JetSpec, a head-based SD framework that combines one-forward drafting efficiency with branch-wise causal conditioning. JetSpec trains a causal parallel draft head over fused hidden states from the frozen target model, producing candidate trees whose scores align with the target model's autoregressive factorization. This enables JetSpec to convert larger draft budgets into longer accepted prefixes and higher end-to-end speedup. Across math, coding, and chat benchmarks on dense and MoE Qwen3 models, JetSpec consistently outperforms bidirectional-head and tree-based SD baselines. On H100 GPUs, JetSpec achieves up to 9.64x speedup on MATH-500 and 4.58x on open-ended conversational workloads, with further latency gains demonstrated through vLLM integration under realistic serving loads. Our code and models are available at https://github.com/hao-ai-lab/JetSpec.

cs.CL↗

arXiv2026

Dynamic Regret via Discounted-to-Dynamic Reduction with Applications to Curved Losses and Adam Optimizer

We study dynamic regret minimization in non-stationary online learning, with a primary focus on follow-the-regularized-leader (FTRL) methods. FTRL is important for curved losses and for understanding adaptive optimizers such as Adam, yet existing dynamic regret analyses are less explored for FTRL. To address this, we build on the discounted-to-dynamic reduction and present a modular way to obtain dynamic regret bounds of FTRL-related problems. Specifically, we focus on two representative curved losses: linear regression and logistic regression. Our method not only simplifies existing proofs for the optimal dynamic regret of online linear regression, but also yields new dynamic regret guarantees for online logistic regression. Beyond online convex optimization, we apply the reduction to analyze the Adam optimizers, obtaining optimal convergence rates in stochastic, non-convex, and non-smooth settings. The reduction also enables a more detailed treatment of Adam with two discount parameters $(β_1,β_2)$, leading to new results for both clipped and clip-free variants of Adam optimizers.

cs.LG↗

arXiv2026

On the Relation Between Interval Regret and Dynamic Regret

Non-stationary online learning has attracted much attention in recent years, as static regret is insufficient to guide algorithm design in changing environments. To address this limitation, interval regret and dynamic regret have been introduced as two representative performance metrics that strengthen static regret in complementary directions. Interval regret requires an online algorithm to achieve competitive static regret over every local time interval, whereas dynamic regret evaluates performance against an arbitrary sequence of time-varying comparators. Despite their importance, the relation between these metrics has long remained unclear. Prior work has often regarded interval regret as the stronger notion, based on the intuition that local guarantees should naturally induce global guarantees. Consequently, it is widely conjectured that an algorithm with optimal interval regret should automatically attain optimal dynamic regret. In this paper, we first establish a negative result that refutes this intuition of a metric-level implication. Specifically, for both convex and curved functions (including exp-concave and strongly convex functions), we show that there exist instances in which an algorithm with optimal interval regret nevertheless fails to achieve optimal dynamic regret. We then show how to leverage local adaptivity to obtain optimal dynamic regret. In particular, optimal dynamic regret can be attained by invoking an interval regret minimization process over an enlarged Euclidean ball containing the original convex feasible domain and using a suitable domain-converted surrogate loss. This reduction applies to both convex and curved functions. As a byproduct, we obtain the first proper and efficient algorithm with optimal dynamic regret for exp-concave functions, improving prior results while significantly simplifying the analysis.

cs.LG↗

arXiv2026

Universal Dynamic Portfolios

Cover's Universal Portfolio (Cover, 1991) matches the performance of the best constant rebalanced portfolio in hindsight. We generalize this framework to compete with an arbitrary comparator sequence $\mathbf{u}_1,\ldots,\mathbf{u}_T$, leading to a dynamic regret minimization problem for the log loss where existing methods break down due to potentially unbounded gradients. The log loss is exp-concave, a curvature property that classically yields fast rates for static regret, yet we show that this advantage generally disappears in the dynamic setting. In particular, a linear-loss-type $\sqrt{TP_T}$ dependence is unavoidable, where $P_T=\sum_{t=2}^T\lVert\mathbf{u}_t-\mathbf{u}_{t-1}\rVert_1$ is the standard path length. This limitation stems from the coarse nature of $P_T$, which obscures finer spatial and temporal structure of the comparator sequence. We therefore introduce two structure-aware measures---the Jensen-Shannon distance for spatial structure and the JS$^q$-path length for temporal structure---under which faster rates are attainable when the comparator sequence has favorable structure. To achieve sharp bounds for both measures simultaneously, we develop Universal Dynamic Portfolio, a parameter-free method that combines a new Dirichlet Hedge algorithm with a fixed-share update, while retaining a near-optimal $P_T$ guarantee in the worst case. Finally, under an additional bounded-gradient assumption, we show that OPS admits the faster $T^{1/3}P_T^{2/3}$ dynamic regret rate over all comparator sequences. We attain this rate with a tractable proper algorithm that applies more broadly to general online exp-concave optimization over arbitrary compact convex domains.

cs.LG↗

arXiv2026

AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters

Speculative decoding, in which a lightweight draft model first generates a draft sequence that is then verified by the target model, has become a prevalent paradigm for accelerating large language model inference. Recent work such as DFlash further boosts drafting efficiency by leveraging diffusion drafters, whose parallel denoising mechanism enables draft generation in a single forward pass. In this work, we uncover a central pitfall of diffusion drafters: the global dependency, arising from bidirectional attention and KV injection in diffusion drafters, is a double-edged sword. On one hand, it endows the model with parallel generation and global contextual modeling capabilities; on the other hand, this global dependency introduces high variance at both the domain-level and the token-level: acceptance rates fluctuate substantially across different domains, and the acceptance probability varies across draft positions. To tackle this issue, we propose the AdaFlash framework, comprising two components: (i) an on-policy distillation (OPD) algorithm with reverse-KL divergence tailored for diffusion drafters, which delivers stable convergence and continuously adapts the drafter to the deployment distribution; and (ii) an adaptive length head that dynamically adjusts the candidate length on the fly, substantially lowering the verification cost of the target model. Experiments demonstrate that AdaFlash consistently improves the speedup during deployment, with especially significant gains under high-concurrency, achieving up to 66% higher average throughput than previous state-of-the-art methods. Our code is available at https://github.com/ZinYY/AdaFlash.

cs.LG↗

arXiv2026

On the Convex Transform Order of Order Statistics

Let X1,...,Xn be independent exponential random variables with arbitrary positive, not necessarily equal, rates, and let Y1,\ldots,Yn be independent and identically distributed exponential random variables. We prove that, for every order statistic, the homogeneous order statistic is smaller than its heterogeneous counterpart in the convex transform order. Our results show that, relative to the homogeneous benchmark, heterogeneity stretches upper quantiles more strongly than lower quantiles, while the homogeneous order statistic ages faster in the convex-transform sense. We further extend the comparison to a proportional-hazards family that includes Weibull, Lomax, and Burr XII distributions. We give explicit conditions under which the comparison is preserved and show that the shape restriction is sharp within the Weibull family. These results characterize how component heterogeneity changes the distributional shape of order statistics beyond effects on location or scale.

stat.ME↗

arXiv2026

Frequency collisions in parametrically modulated superconducting circuits

Superconducting circuits are a leading platform for scalable quantum computing, where parametric modulation is a widely used technique for implementing high-fidelity multi-qubit operations. A critical challenge, however, is that this modulation can induce a dense landscape of parasitic couplings, leading to detrimental frequency collisions that constrain processor performance. In this work, we develop a comprehensive numerical framework, grounded in Floquet theory, to systematically analyze and mitigate these collisions. Our approach integrates this numerical analysis with newly derived analytical models for both qubit-modulated and coupler-modulated schemes, allowing us to characterize the complete map of parasitic sideband interactions and their distinct error budgets. This analysis forms the basis of a constraint-based optimization methodology designed to identify parameter configurations that satisfy the derived physical constraints, thereby avoiding detrimental parasitic interactions. We illustrate the utility of this framework with applications to analog quantum simulation and gate design. Our work provides a predictive tool for co-engineering device parameters and control protocols, enabling the systematic suppression of crosstalk and paving the way for large-scale, high-performance quantum processors.

quant-ph↗