arXiv:2507.07295v2 Announce Type: replace-cross
Abstract: How biological networks achieve robust control despite relying on imperfect, local information remains an important open question. Here, we identify thermodynamic constraints that can curtail non-equilibrium steady-state responses so severely that even crude, local feedback rules can achieve globally stable control without requiring precise network design or global information. Specifically, using Markov jump processes as a general framework for biophysical dynamics, we derive general non-equilibrium response constraints showing that for many classes of rate perturbations, steady-state responses have fixed signs across all driving strengths, so that near-equilibrium responses predict far-from-equilibrium behavior regardless of system complexity. These constraints clarify several biological phenomena: monotonicity is thermodynamically guaranteed whenever a perturbation acts on a single transition rate, and non-monotonic responses, as observed for example in transcription factor regulation, arise only when an input simultaneously modulates multiple rates. Even in this case, we identify a graph-theoretic concept termed ``coherence'' that allows for a restoration of monotonicity. We show how coherence naturally and generally emerges in classic biophysical models of adaptation, including E. coli chemotaxis, and transcription factor regulation when biological constraints on network parameterization are included. We next show that, within a control-theoretic framework, these constraints guarantee that simple linear feedback on small subsets of kinetic rates achieves globally stable tracking and adaptation without coordinated manipulation of many variables. For systems with one regulator, local stability implies global stability for arbitrary network topologies without fine tuning, revealing that non-equilibrium thermodynamics fundamentally constrains biochemical network responses.
Science Journals
arXiv:2605.26384v1 Announce Type: new
Abstract: At global scale, data-center electricity demand is growing faster than the grids that supply it, while system operators increasingly require large flexible loads that can adjust power within seconds to absorb variable wind and solar generation. For multi-megawatt AI/HPC facilities, the key unresolved question is practical and measurable: how quickly can the software stack translate a grid request into a real change in GPU power at the facility meter, where commitments are settled? We answer this on real hardware with GridPilot, a three-tier predictive controller operating across milliseconds, seconds, and hours, augmented by a deterministic safety-island bypass for fast response. On a three-GPU NVIDIA V100 testbed, GridPilot achieves a measured end-to-end trigger-to-target response of 97.2 ms, which is 6.9x faster than the 700 ms requirement of Nordic Fast Frequency Reserve. We further incorporate an instantaneous Power Usage Effectiveness (PUE) correction so dispatched commitments remain robust at meter level rather than only at IT load level. In replay experiments across six representative European grids (from Sweden to Poland), the PUE-aware controller closes 2.5-5.8 percentage points of cooling-overhead drag. GridPilot is released as open source and serves as a proof of concept that MW-scale AI/HPC demand can be engineered as controllable, grid-responsive flexibility by design.
arXiv:2605.26385v1 Announce Type: new
Abstract: Large-scale search, recommendation, and retrieval-augmented generation (RAG) systems typically employ a two-stage architecture: an early-stage ranker (ESR) generates a candidate set, which is subsequently re-ranked by a late-stage ranker (LSR). While there are many reinforcement learning (RL) methods for training the LSR, end-to-end training of the ESR has proven challenging. In particular, naive application of "vanilla" policy gradient (V-PG) is not scalable for candidate-set sizes relevant for practical use due to exploding variance. This issue arises because V-PG propagates the gradient to the joint probability of the candidate sets, ignoring the contribution of each specific item in the candidate set to the reward. To mitigate this issue, we propose a novel "credit-assigned" policy gradient (CA-PG), which computes gradients with respect to the probability that the target item is chosen in any candidate set, i.e. marginalizing over all candidate sets that contain it. Our theoretical analysis reveals that CA-PG significantly reduces the variance of V-PG by marginalizing over the specific composition of the candidate set, while preserving the ability to learn the correct ranking of items under a reasonably aligned LSR policy. Experiments on both synthetic and real-world data demonstrate that CA-PG improves the convergence speed and training stability for ESRs utilizing the canonical Plackett-Luce model, especially when the candidate-set size is large.
arXiv:2604.09157v2 Announce Type: replace
Abstract: We introduce a novel learning framework for accelerated Monte Carlo (MC) dose calculation termed Energy-Shifting. This approach leverages deep learning to synthesize highly complex polyenergetic dose distributions directly from simple monoenergetic inputs under identical beam configurations. Unlike conventional denoising techniques, which rely on noisy low-count dose maps that compromise beam profile integrity, our method achieves superior cross-domain generalization on unseen datasets by integrating high-fidelity anatomical textures and source-specific beam similarity into the model's input space. Furthermore, we propose a novel 3D architecture termed TransUNetSE3D, featuring Transformer blocks for global context and Residual Squeeze-and-Excitation (SE) modules for adaptive channel-wise feature recalibration. Hierarchical representations of these blocks are fused into the network's latent space alongside the primary dose-map parameters, allowing physics-aware reconstruction. This hybrid design outperforms existing UNet and Transformer-based benchmarks in both spatial precision and structural preservation, while maintaining the execution speed necessary for real-time use. Our proposed pipeline achieves a Gamma Passing Rate exceeding 98% (3%/3mm) compared to the MC reference, evaluated within the framework of a treatment planning system (TPS) using 6MV TrueBeam Lineac Accelerator (LINAC) for prostate radiotherapy. These results offer a robust solution for fast volumetric dosimetry in adaptive radiotherapy.
arXiv:2605.26404v1 Announce Type: new
Abstract: Modern online services rely on third-party APIs for authentication, payments, communication, identity verification, fraud detection, observability, and fulfillment. These dependencies are outside the direct operational control of the application owner and may experience regional outages, throttling, latency spikes, quota exhaustion, or behavior changes that surface as user-visible failures. This paper presents configuration-driven dynamic API routing, an architecture for resilient third-party service integration based on pluggable factor lists, real-time telemetry, circuit breakers, bulkhead isolation, and a closed-loop decision engine. A factor list defines operation-specific hard gates and weighted scoring functions that evaluate candidate providers using live metrics, regional policy constraints, quota state, latency, cost, and incident signals. The router separates routing policy from application code, allowing operators to adapt vendor selection at runtime without redeploying applications. We formalize the factor-list model, describe a request-time routing algorithm, present the event pipeline that computes sliding-window provider health metrics, and analyze failover behavior under degraded-provider scenarios. We also describe an anonymized SMS verification case study in which manual vendor switching was replaced by automated routing driven by completion-rate telemetry.
arXiv:2605.02958v2 Announce Type: replace
Abstract: Representation Engineering analyses often characterize refusal using static directions extracted from terminal or pooled representations. We ask whether this view misses how refusal is constructed across layer-token positions. Using causal tracing, we identify a \textit{Refusal Trajectory}: a sparse upstream activation pattern that often persists even when attacks such as GCG suppress terminal refusal signals. Based on this observation, we propose SALO (Sparse Activation Localization Operator), a lightweight white-box detector that operates on raw hidden-state volumes from a selected layer window. Across Qwen, Llama, and Mistral models, SALO improves jailbreak detection on several attack families under a fixed XSTest-calibrated operating point. We further analyze static RepE-style baselines, ROI sensitivity, adaptive GCG attacks, and encoded-input boundary cases, clarifying both the promise and limitations of refusal-trajectory monitoring.
arXiv:2605.26376v1 Announce Type: new
Abstract: Hepatocellular carcinoma (HCC) is biologically heterogeneous, shaped by the interplay between hepatic functional reserve and tumor-related oncologic factors; thus, similar survival outcomes may reflect fundamentally different underlying biological processes. Prognostic modeling in HCC is informed by rich multimodal information from multiparametric MRI and radiology reports from routine clinical practice. Existing prognostic vision-language models (VLMs) learn a single entangled latent representation that blends hepatic and tumor-related factors, limiting both accuracy and biological interpretability. We present BioFact-MoE, a biologically factorized Mixture of Experts (MoE) framework that explicitly decomposes liver and tumor factors via biologically supervised experts within a residual MoE survival architecture. On a HCC cohort of N=588 patients (pretrained on 4,582 3D MRI image-report pairs), BioFact-MoE consistently improves survival prediction over all baselines across time horizons, achieving 12-, 18-, and 24-month AUCs of 75.33%, 75.85%, and 73.96%. Beyond scalar risk prediction, gated expert weights enable phenotype-aware risk stratification. Pathway-informed gating uncovers clinically meaningful treatment-associated survival heterogeneity. In held-out validation, hepatic and tumor embeddings show selective associations with liver function and tumor burden markers, respectively (p<0.05), without supervision. The code is available at https://github.com/jy-639/BioFact-MoE.
arXiv:2605.26465v1 Announce Type: new
Abstract: Local Differential Privacy (LDP) has become the de facto standard for privacy-preserving data collection in large-scale systems, in particular for the purpose of estimating frequencies. However, the current research landscape lacks a systematic and principled way to compare LDP protocols. The parameter $\varepsilon$ of LDP is considered the measure of privacy, but it only bounds worst-case distinguishability. Other comparisons rely on utility-driven analyses, where mechanisms are ranked based on their ability to preserve data utility for a given privacy budget $\varepsilon$. Both such kinds of comparisons fail to account for the strength of protocols against diverse attacker models. In this paper, we propose a framework for analyzing LDP frequency estimation protocols through the lens of Quantitative Information Flow (QIF). By modeling LDP mechanisms as probabilistic channels, we leverage the concept of refinement (Blackwell ordering) to establish more principled classifications. This approach allows us to determine when one protocol is intrinsically superior to another for all possible adversaries, and to discuss the implications for utility. In particular, our analysis uncovers cases where protocols previously deemed "optimal" are, in fact, incomparable with, or strictly dominated by, other protocols. We provide a formal QIF-based treatment of seven state-of-the-art protocols, including Generalized Randomized Response (GRR), local hashing variants (BLH, OLH), unary encoding schemes (SUE, OUE), and Thresholding with Histogram Encoding (THE). This perspective bridges the gap between the LDP and formal methods communities and enables principled, adversary-aware reasoning about locally private systems.
A Dataset of Robot-Patient and Doctor-Patient Medical Dialogues for Spoken Language Processing Tasks
arXiv:2605.26747v1 Announce Type: new
Abstract: Large Language Models (LLMs) have brought huge improvements to Artificial Intelligence (AI), which can be applied to general-purpose tasks. However, their application to textual or spoken medical consultations is still an open research problem. This paper proposes MeDial-Speech, a novel speech dataset for training and evaluating Med-AIs that can carry out consultations with patients. It was collected in realistic environments from robot-patient and doctor-patient dialogues, contains 111+ hours of speech data (without data augmentation), and covers four health conditions: Lewy body dementia, heart failure, shoulder pain, and angina. In addition, we propose a dialogue benchmark via sentence selection (with 20 options) to evaluate three state-of-the-art LLMs: GPT-5 mini, DeepSeek-V3, and Claude Sonnet 4. Experimental results reveal that Claude Sonnet 4 is the best in sentence selection, with 71.1% accuracy using manual transcriptions and 74.7% using automatic transcriptions, and that all LLMs are highly overconfident in their probabilistic predictions, regardless of selecting correct or incorrect sentences in medical dialogues. This dataset is free of charge for non-commercial purposes at: https://huggingface.co/datasets/hcuayahu/MeDial-Speech
arXiv:2602.10450v2 Announce Type: replace
Abstract: Optimization modeling underpins decision-making in logistics, manufacturing, energy, and finance, yet translating natural-language requirements into correct optimization formulations and solver-executable code remains labor-intensive. Although large language models (LLMs) have been explored for this task, evaluation is still dominated by toy-sized or synthetic benchmarks, masking the difficulty of industrial problems with $10^{3}$--$10^{6}$ (or more) variables and constraints. A key bottleneck is the lack of benchmarks that align natural-language specifications with reference formulations/solver code grounded in real optimization models. To fill in this gap, we introduce MIPLIB-NL, built via a structure-aware reverse construction methodology from real mixed-integer linear programs in MIPLIB~2017. Our pipeline (i) recovers compact, reusable model structure from flat solver formulations, (ii) reverse-generates natural-language specifications explicitly tied to this recovered structure under a unified model--data separation format, and (iii) performs iterative semantic validation through expert review and human--LLM interaction with independent reconstruction checks. This yields 223 one-to-one reconstructions that preserve the mathematical content of the original instances while enabling realistic natural-language-to-optimization evaluation. Experiments show substantial performance degradation on MIPLIB-NL for systems that perform strongly on existing benchmarks, exposing failure modes invisible at toy scale.
arXiv:2605.26373v1 Announce Type: new
Abstract: We study adversarial online learning with hidden-convex losses, i.e., nonconvex losses that become convex after a nonlinear reparameterization. Ghai, Lu and Hazan (2022) proved that, under geometric and smoothness assumptions, online gradient descent (OGD) on such nonconvex losses approximately simulates online mirror descent (OMD) on the underlying convex losses with a suitable regularizer, yielding $\mathcal{O}(T^{2/3})$ regret. They left open whether the optimal $\Theta(\sqrt{T})$ regret from online convex optimization can be recovered in this hidden-convex setting. We answer this question affirmatively. More specifically, via a sharper discrete-time algorithmic equivalence argument, we prove that OGD achieves $\mathcal{O}(\sqrt{T})$ regret under the same assumptions, matching the optimal worst-case rate for adversarial online convex optimization. We also address another open question of Ghai, Lu and Hazan (2022) by clarifying the geometry required for this algorithmic equivalence. We replace the diagonal-Jacobian sufficient condition with a necessary-and-sufficient Hessian compatibility condition, thereby expanding the class of admissible reparameterizations. We complement our tight regret bound with a lower bound showing that the Hessian compatibility assumption is essential for OGD; when it fails, we construct a smooth reparameterization and an adversarial sequence of hidden-convex losses for which OGD suffers $\Omega(T)$ regret. Finally, we extend our analysis to one-point bandit feedback and prove a $\mathcal{O}(T^{3/4})$ expected regret bound for bandit OGD with spherical smoothing, matching its classical rate on convex losses.
arXiv:2605.26806v1 Announce Type: new
Abstract: High-fidelity numerical simulations of compressible flow past a rapidly rotating cylinder are used to investigate the evolution of aerodynamic loads and flow instability over a wide range of Reynolds numbers (Re = 1000 to 6000). The study reveals a transition from periodic vortex shedding to complex multi-mode oscillatory states, with a critical bifurcation identified near Re = 5650. Spectral analysis of lift and drag signals shows the emergence and interaction of multiple dominant frequencies, accompanied by amplitude modulation and nonlinear mode coupling in the post-bifurcation regime. To model these highly nonlinear dependencies, data-driven approaches are systematically explored using a database of 101 high-fidelity simulations (1 million core hours). Polynomial regression provides baseline fits but fails to capture localized fluctuations near bifurcation. Bayesian regression frameworks employing B-spline and Gaussian radial basis functions improve flexibility and uncertainty quantification, with spline-based models demonstrating superior performance in capturing piecewise nonlinear trends. Artificial neural networks (ANNs) are then developed as high-capacity surrogate models, achieving excellent predictive accuracy for maximum lift coefficient and instability onset time, while maintaining reasonable fidelity for the more challenging drag coefficient. Beyond regression, the ANN is further evaluated as a generative model to reconstruct flow behavior at unseen Re. A hierarchical refinement strategy is introduced, and results show that when trained on high-fidelity data, ANN-based models can serve as efficient and reliable surrogates for complex fluid dynamics problems.
Certified Purity for Cognitive Workflow Executors: From Static Analysis to Cryptographic Attestation
arXiv:2605.01037v3 Announce Type: replace
Abstract: We present a certified purity architecture that converts governance enforcement in cognitive workflow systems from a runtime convention into a structural capability boundary. A prior three-layer governance architecture proves governance completeness, provenance completeness, and the impossibility of ungoverned effects, conditional on the pure module constraint: that step executors cannot perform effects. That constraint was enforced by module import graph analysis, which is insufficient against adversarial bypass on the BEAM virtual machine. This paper closes the gap through four mechanisms: (1) a restricted WebAssembly compilation target where effect-producing instructions are structurally absent; (2) purity certificates, cryptographically signed proofs binding executor binaries to their import classifications; (3) a runtime verification gate that rejects uncertified executors before they enter the governance pipeline; and (4) portable governance credentials via remote attestation for cross-organizational verification. We prove four theorems: structural purity by construction, bypass elimination for all five BEAM bypass classes, certificate integrity, and gate completeness. The guarantee holds relative to an explicit Trusted Computing Base. Evaluation on four implemented executors shows verification latency of 39--42 us, full plan cycle under 400 us, runtime overhead under 0.4% of a 100 ms HTTP request, and zero determinism divergences across repeated invocations.
arXiv:2605.24461v2 Announce Type: replace
Abstract: The electric power supply for AI data centers is now the most significant bottleneck in the race toward Artificial General Intelligence, surpassing even the constraint of AI accelerator availability. To our knowledge, this paper is the first to describe the end-to-end power management process for a hyper-scale AI datacenter; from early power planning to accommodate next-generation accelerators 6--12 months before their general availability, to tuning power settings after large scale deployment, and finally to dynamic, runtime power management for evolving workloads. We present detailed power measurements for a 150 MW datacenter hosting a cluster of 83K GB200 GPUs. We share insights from building this state-of-the-art AI cluster. We hope this work encourages practitioners across the industry to share their own experiences as well.
arXiv:1909.08210v4 Announce Type: replace
Abstract: A restricted Boltzmann machine (RBM) is a two-layer neural network with shared weights and has been extensively studied for dimensionality reduction, data representation and recommendation systems in the literature. The traditional RBM requires a probabilistic interpretation of the values on both layers and a Markov chain Monte Carlo (MCMC) procedure to generate samples during the training. The contrastive divergence (CD) is efficient to train the RBM but its convergence has not been proved mathematically. In this paper, using a maximum a posteriori (MAP) estimate and the expectation maximization (EM) algorithm, we show that the CD algorithm without MCMC is convergent for the conditional likelihood object function. Another key contribution in this paper is the reformulation of the RBM into a deterministic model. Within the reformulated RBM, the CD algorithm without MCMC approximates the gradient descent (GD) method. This reformulated RBM can take the continuous scalar and vector variables on the nodes with flexibility in choosing the activation functions. Numerical experiments show its capability in both linear and nonlinear dimensionality reduction, and, for the nonlinear dimensionality reduction, the reformulated RBM can outperform principal component analysis (PCA) by choosing the proper activation functions. Finally, we demonstrate its application to vector-valued nodes for the CIFAR-10 dataset (color images) and the multivariate sequence data, which cannot be configured naturally with the traditional RBM. This work not only provides theoretical insights regarding the traditional RBM but also unifies the linear and nonlinear dimensionality reduction for scalar and vector variables.
arXiv:2605.26911v1 Announce Type: new
Abstract: LLM-generated peer reviews are increasingly common at major venues, yet their deficiencies are hard to detect because they are uniformly fluent and well-structured. Existing work either classifies authorship without judging quality, or scores quality with features designed for human-written reviews; no prior system detects deficiencies in LLM-generated reviews at the level of individual defect types. To bridge the gap, we introduce TADDLE, a Tool-Augmented Agent for Detecting Deficient LLM-Generated Peer Reviews, together with the first expert-annotated benchmark for this task. Our benchmark comprises 1,800 reviews on 50 ICLR 2025 papers, multi-label-annotated by 18 domain experts against a taxonomy of six defect categories (plus a non-deficient label). TADDLE decomposes detection into four specialized analysis tools -- Verify, Correct, Complete, and Transform -- orchestrated by an agent; an integrator synthesizes their outputs into binary and multi-label classifications via two-stage semi-supervised learning. Extensive experiments show that TADDLE performs strongly on both binary detection and the multi-label classification task. We release the benchmark and code at https://github.com/AquariusAQ/TADDLE.
arXiv:2605.26860v1 Announce Type: new
Abstract: The continuous spontaneous localization (CSL) model is one of the most promising approaches to address the wave function collapse problem in the measurement process of standard quantum mechanics. In this work, the effect of the CSL model on a Bragg atom interferometer was investigated. A Bragg interferometer achieving high fringe contrast of 99$\%$ has been demonstrated, maintaining this performance level at interrogation time up to $T=60~\mathrm{ms}$. The primary factors responsible for fringe contrast loss in the atom interferometer were systematically analyzed and corrected. This improvement established a new upper limit of $\lambda_{\rm CSL}=1.27\times10^{-5}~\mathrm{s}^{-1}$ at $r_C=10^{-5}~\mathrm{m}$ for the CSL collapse rate, representing approximately 4 times enhancement over previous atom-interferometric constraints.
arXiv:2605.27358v1 Announce Type: new
Abstract: Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on-device LLMs. We first formulate an on-device MoE scaling law that jointly optimizes MoE architecture under mobile memory and compute constraints, identifying an on-device sweet spot - moderate sparsity with fine-grained and shared experts - that is simultaneously memory and compute-optimal. Building on the derived architectures, we train MobileMoE with a four-stage recipe covering pre-training, mid-training, instruction fine-tuning, and quantization-aware training, all on open-source datasets. Across 14 benchmarks, MobileMoE matches or exceeds leading on-device dense LLMs with 2-4$\times$ fewer inference FLOPs, and matches or surpasses the state-of-the-art MoE OLMoE-1B-7B with up to 60% fewer parameters. To bridge the last mile to mobile deployment, we provide the first efficient MoE inference on commodity smartphones with comprehensive on-device profiling. At comparable INT4 weight memory, MobileMoE-S delivers $1.8$-$3.8\times$ faster prefill and $2.2$-$3.4\times$ faster decode than the dense baseline MobileLLM-Pro.
arXiv:2605.26196v1 Announce Type: new
Abstract: Trusted Execution Environments (TEEs) protect confidentiality and integrity of trusted applications by creating an isolated environment for executing code. Prior work has shown that users may feel more comfortable sharing data when they know it will be protected by a TEE, especially if they understand what a TEE is. In this study, we evaluated text-based explanations introducing TEEs to non-experts. We analyzed existing TEE explanations to develop candidate explanations and evaluated them via vignette scenarios with 966 crowdworkers. The explanations that enhanced understanding most were non-technical ones that highlighted specific threats that can be prevented by a TEE. Surprisingly, even the explanations that enhanced understanding had little effect on willingness to use the TEE-enhanced technology. These results provide insights into ways to communicate technical security concepts more effectively but also suggest that explaining security technology might not be enough to address users' privacy concerns.
arXiv:2605.26796v1 Announce Type: new
Abstract: Incentive-based load curtailment unlocks critical demand-side flexibility but is hindered by the limited knowledge of private user parameters and the inherent nonsmoothness of responses due to physical device constraints. We address this via a constrained bilevel optimization framework and propose the Bi-ZOL (Bilevel Zeroth-Order Learning) algorithm. Unlike conventional black-box methods, Bi-ZOL exploits the bilevel structure to decompose the hypergradient, integrating the exact analytical information of the SO's objective with a zeroth-order estimate of the unknown response sensitivity. This structural decomposition-based learning method mathematically smoothes the nonsmooth response landscape and reduces hypergradient estimation error. We provide theoretical convergence guarantees to an approximate stationary point and demonstrate through simulations that Bi-ZOL achieves near-optimal performance.
arXiv:2605.27313v1 Announce Type: new
Abstract: Demographic information is often used to model annotator perspectives in subjective tasks such as hate speech detection, but its benefit is inconsistent: it improves performance in some settings and behaves as noise in others. This paper asks when demographic features help. We analyze demographic gain as a function of both data split properties and modeling frameworks. For data splits, we measure annotator disagreement, namely how often annotators assign different labels to the same example, along with training size and train-test demographic coverage. We find that demographic gains concentrate in regimes with low training disagreement, high test disagreement, fine-grained ambiguity measurement, sufficient training data, and greater demographic overlap. Motivated by these regimes, we introduce a gated demographic residual model that treats demographics as a selective adjustment to text-only predictions. Experiments on MHS and POPQUORN show that this design is effective, especially on high disagreement or low confidence examples. Overall, our results suggest that demographics should not be assumed useful by default; their value depends jointly on the data regime and the modeling framework.
arXiv:2605.26275v1 Announce Type: new
Abstract: Automatic prompt engineering (APE) rewrites prompts to improve downstream task performance, but existing APE loops treat the optimizer itself as a fixed pipeline. We port the code-as-action paradigm of CodeAct (Wang et al., 2024a) to APE and propose SPEAR (Sandboxed Prompt Engineer with Active Roll-back), a free-form agentic optimizer with four tools -- evaluate, python, set_prompt, finish -- that decides autonomously how and when to use them. The distinctive tool is the Python sandbox: the optimizer writes and executes arbitrary Python on the current evaluation DataFrame, performing structural error analysis (confusion matrices, error clustering, per group metrics) the agent itself authors. Two guardrails turn the long-horizon agent into a monotone-improving optimizer: auto-rollback on metric regression, and an optional guard metric floor. We evaluate on three industrial LLM-as-judge suites (13 judge tasks across recruiter-intake, conversational-memory, and query-refinement systems) plus seven BBH tasks and GSM8K. SPEAR wins every industrial task on the primary metric ($\kappa$ 0.857 vs 0.359 on tool-selection; F1-macro 0.815 vs 0.763 on filter-relevance; $\kappa$ 0.254 vs 0.218 on the hardest extraction dimension). On BBH-7 SPEAR averages 0.938 accuracy vs GEPA 0.628 and TextGrad 0.484. Ablations show the Python tool is the largest single lever on complex judge tasks ($\Delta \approx +0.79\kappa$ on the 5-class tool-selection judge, $\Delta \approx +0.35\kappa$ on the hardest extraction dimension when removed); its irreplaceable contribution is class-pair confusion aggregation that a long-context LLM cannot extract reliably from the raw eval DataFrame.
arXiv:2605.27354v1 Announce Type: new
Abstract: Model internals encode rich information about how a large language model (LLM) processes its training data; however, post-training data engineering largely relies on external signals and ignores rich intrinsic signals lying in model internals. We propose SAERL, a data engineering framework for LLM reinforcement learning (RL). It models three intrinsic data properties: diversity, difficulty, and quality, using model internals extracted with Sparse Autoencoder (SAE), an advanced mechanistic interpretability tool. Each property grounds a concrete data engineering operation: SAE-space clustering with moderate batch mixing for batch diversity control, a difficulty proxy for easy-to-hard curriculum ordering, and a quality probe for data filtering. SAERL improves average accuracy by 3.00% over vanilla GRPO and reaches target accuracy with 20% fewer training steps on Qwen2.5-Math-1.5B, with consistent gains across model scales and RL algorithms. Experiments show that SAE transfers effectively across model families and scales, serving as a lightweight and reusable data engineering tool. These results demonstrate that model internals are a powerful and practical source of signals for post-training data engineering.
arXiv:2507.00637v3 Announce Type: replace
Abstract: Cyberattacks on enterprise networks exploit complex dependencies among infrastructure, services, and applications, which challenge traditional analysis methods that focus on attack paths or network topology in isolation. In this study, we introduce a novel probabilistic multilayer modelling framework, based on influence propagation in networks, that integrates attack graphs with the communication network topology, enabling a service-centric impact analysis of cyberattacks. Our method captures both the vulnerability exploitability and network connectivity, allowing us to assess the likelihood of attack propagation and cumulative impacts across interconnected services. By integrating standard vulnerability metrics (such as CVSS) with the network-level connectivity probabilities, the framework provides a cohesive view of the dynamics of cyberattacks. We validate this approach using a realistic case study of an enterprise network, demonstrating its ability to determine critical nodes, vulnerabilities, and service dependencies that significantly influence attack outcomes. Our findings show that integrating network and attack graph perspectives offers more actionable insights into risk assessment and mitigation planning, advancing the analysis of cyberattacks in complex networked environments.
arXiv:2605.26699v1 Announce Type: new
Abstract: Anthropogenic underwater noise from offshore wind turbines is a growing environmental concern, particularly with the large-scale deployment of bottom-fixed and floating devices. This study presents a physics-based vibroacoustic framework to predict operational underwater noise emissions from offshore wind turbines and compares monopile-supported and floating configurations for a 10 MW turbine. The methodology combines time-domain aero-hydro-servo-elastic simulations with a frequency-domain acoustic formulation based on equivalent dipole sources and Green's function solutions, accounting for underwater confinement between the free surface and seabed through the method of images. Results show that floating configurations exhibit enhanced low-frequency acoustic emissions, producing up to 15% higher OASPL than the monopile structures under equivalent water depths for frequencies below 10 Hz due to additional rigid-body motions, while monopile structures radiate more efficiently at higher frequencies associated with drivetrain excitations. Significant differences in the spatial distribution and directivity of the radiated sound field are also observed, with floating platforms displaying more complex three-dimensional radiation patterns and stronger direction-dependent variations, reaching approximately 20-25 dB in the 100-1000 Hz band, compared with the smoother and nearly axisymmetric response of monopile configurations. Water depth strongly influences propagation regimes and overall sound levels, with shallow-water floating configurations showing variations of up to 7% in OASPL relative to deep-water cases. The proposed framework enables quantification of vibro-acoustic noise and provides a predictive tool for assessing underwater acoustic impacts during the design phase, supporting environmentally informed offshore wind turbine design and future regulatory and monitoring strategies.