arXiv:2606.12688v2 Announce Type: replace Abstract: We are entering a new era of composite model architectures that integrate diverse components such as vision encoders, language backbones, diffusion and flow heads, audio codecs, action generators, and world-model predictors. Such architectures underpin a broad class of multimodal models, including unified multimodal models, omni models, speech-language models, vision-language-action policies, and world models. However, existing model serving frameworks were built on narrow assumptions about model structure, making them ill-suited to accommodate this new architectural diversity. Here we present M*, a universal serving system for efficient serving of composite AI models. M* represents models as dataflow graphs, processing requests spanning diverse modalities and tasks as traversals over these graphs. The core insight is a modular abstraction that supports arbitrary composition of model components, flexible placement onto a physical cluster, and model-agnostic optimizations within a distributed runtime. We call this abstraction the Walk Graph and show how it can concisely capture composite models from a broad range of families. We instantiate M* on representative models and find that it achieves, on average, 20% lower end-to-end latency than vLLM-Omni for text-to-image workloads on BAGEL, while delivering up to 2.9x lower real-time factor and 2.7x higher throughput for text-to-speech workloads on Qwen3-Omni. M* also outperforms the V-JEPA 2-AC rollout baseline for robotic planning by up to 12.5x. Thus, our work paves the road towards more efficient serving of complex models with minimal developer effort.
Science Journals
arXiv:2606.16160v1 Announce Type: new Abstract: Accurately classifying cognitive load from functional near-infrared spectroscopy (fNIRS) signals remains a significant challenge due to temporal variability, inter-subject differences, and sensitivity to preprocessing choices. This study provides a comprehensive evaluation of EEGNet for fNIRS-based cognitive load classification by systematically examining the effects of temporal segmentation strategies (overlapping vs. non-overlapping), window lengths (10s, 20s, 30s), feature extraction methods (Analysis of Variance (ANOVA), Principal Component Analysis (PCA), Fast Independent Component Analysis (FastICA)), learning rate configurations (fixed and adaptive), and evaluation protocols (random split vs. subject-independent (SI)). Results from random-split experiments show that overlapping segmentation, combined with smaller fixed learning rates (0.01-0.001), yields the highest accuracies, due to temporal redundancy and dense sampling of hemodynamic transitions. However, SI evaluation reveals a substantial drop in accuracy, demonstrating limited generalization to unseen participants. Under SI evaluation, non-overlapping segmentation outperformed overlapping windows, with the best accuracy of 56.11% achieved using PCA features with a 20-second window and a 0.1 learning rate. These findings indicate that eliminating temporal redundancy helps the model learn more robust and generalizable representations of cognitive load across individuals. Although adaptive learning rate strategy improved training stability, it did not surpass the performance of optimally selected fixed learning rates. The study highlights the critical role of segmentation strategy and learning rate selection in improving model generalization and identifies methodological considerations essential for developing reliable, real-time, and SI cognitive load classification systems using fNIRS.
arXiv:2604.18827v2 Announce Type: replace-cross Abstract: Scaling data and artificial neural networks has transformed AI, driving breakthroughs in language and vision. Whether similar principles apply to modeling brain activity remains unclear. Here we leveraged a dataset of 3.1 million neurons from the visual cortex of 73 mice across 323 sessions, totaling more than 150 billion neural tokens recorded during natural movies, images and parametric stimuli, and behavior. We train multi-modal, multi-task models that support three regimes flexibly at test time: neural prediction, behavioral decoding, neural forecasting, or any combination of the three. OmniMouse achieves state-of-the-art performance, outperforming specialized baselines across nearly all evaluation regimes. We find that performance scales reliably with more data, but gains from increasing model size saturate. This inverts the standard AI scaling story: in language and computer vision, massive datasets make parameter scaling the primary driver of progress, whereas in brain modeling -- even in the mouse visual cortex, a relatively simple system -- models remain data-limited despite vast recordings. The observation of systematic scaling raises the possibility of phase transitions in neural modeling, where larger and richer datasets might unlock qualitatively new capabilities, paralleling the emergent properties seen in large language models. Code available at https://github.com/enigma-brain/omnimouse.
arXiv:2606.16970v1 Announce Type: new Abstract: Different from deterministic rankers that seek to maximize relevance at top ranks, stochastic ranking policies instead estimate distributions over permutations, from which rankings are sampled, towards obtaining diversified or fair exposure. Such policies are commonly evaluated in terms of expected effectiveness postreranking. However, the randomness inherent in these policies gives rise to a fundamental but under-explored ex ante question: prior to applying stochastic reranking, how large can the induced variation in retrieval effectiveness be in the worst case? This paper presents a theoretical analysis of reranking risk, defined as the maximum absolute change in discounted cumulative gain (DCG) resulting from a permutation sampled from a stochastic reranking policy applied to a fixed retrieved list.We derive that this risk is governed by the distribution of the recall points in the initial retrieved list. We conduct experiments on submitted runs from the TREC Fairness 2022 track that employ stochastic reranking policies and empirically demonstrate that the effectiveness variations predicted by our theory closely approximate the observed changes in DCG.
arXiv:2606.04406v2 Announce Type: replace Abstract: We define bit-counting complexity classes, where the membership depends on the binary profile of the number of accepting paths of non-deterministic polynomial time Turing machines. We study the relationship between this new family of complexity classes and the classical complexity classes. We prove that the complexity class ${\bf PP}$ is contained in our comparison based bit-counting complexity classes ${\bf B_{|0|=|1|}P}$, ${\bf B_{|0|<|1|}P}$ and ${\bf B_{|0|>|1|}P}$. We further show that all of these complexity classes are Turing equivalent ${\bf P}^{\bf PP} = {\bf P}^{{\bf B_{|0|=|1|}P}}={\bf P}^{{\bf B_{|0|>|1|}P}}={\bf P}^{{\bf B_{|0|<|1|}P}}$. We also prove that complexity classes ${\bf NP}$ and ${\bf CoNP}$ are contained in both of our parity based bit-counting complexity classes ${\bf B_{|0| \oplus}P}$ and ${\bf B_{|1| \oplus}P}$.
arXiv:2606.13608v2 Announce Type: replace Abstract: Agent systems are advancing quickly across domains, but their evaluation remains fragmented. Most benchmarks rely on fixed, LLM-centric harnesses that require heavy integration, create test-production mismatch, and limit fair comparison across diverse agent designs. The root problem is the lack of an open, agent-agnostic assessment interface. We advocate Agentified Agent Assessment (AAA), where evaluation is performed by judge agents and all participants interact through standardized protocols: A2A for task management and MCP for tool access. Conventional benchmarking defines two separate interfaces, one for the benchmark and one for the agent, while AAA only needs one; this yields a generic, unified framework that separates assessment logic from agent implementation and enables reproducible, interoperable, and multi-agent evaluation. We further introduce AgentBeats as a concrete realization of AAA: we identify five practical operation modes that make standardized assessment compatible with real-world constraints on openness, privacy, and reproducibility. To evaluate our design at scale, we conduct two studies: a five-month open competition that drew 298 judge agents across 12 categories together with 467 subject agents from independent participants, showing that AAA applies across a heterogeneous range of benchmarks; and a case study on coding agents that confirms agentified evaluation preserves fidelity with the public record while surfacing previously missing head-to-head results, yielding research insights about agent design. Combining a community-scale field study and a controlled coding case study, we verify that AAA delivers coverage, practicality, and fidelity across heterogeneous scenarios at scale. Together, AAA and AgentBeats offer a clear path toward open, standardized, and reproducible agent assessment.
arXiv:2601.16470v3 Announce Type: replace-cross Abstract: Nonlinear stochastic motion presents significant challenges for Bayesian particle tracking. To address this challenge, we propose a lifting framework that constructs a higher-dimensional linear stochastic representation of nonlinear state-space models. The resulting surrogates enable the use of computationally efficient linear filtering techniques while retaining a direct connection to the underlying nonlinear dynamics. The paper derives the necessary conditions for such transformations using Ito's lemma and variational calculus, and illustrates the method on a bistable cubic motion model, radial Brownian process model, and a logistic model with multiplicative noise. Simulations confirm that the transformed linear systems, when projected back, accurately reconstruct the nonlinear dynamics and, in distinct regimes of stiffness and singularity, yield tracking accuracy competitive with conventional filters, while avoiding their structural instabilities.
arXiv:2606.05594v2 Announce Type: replace Abstract: We introduce the Coverage Gap as a measurable distance between the public exposure of critical-infrastructure operators and their declared capability to coordinate vulnerability disclosure. We instantiate it against the 915 Chilean Operadores de Importancia Vital (OIVs) designated by the National Cybersecurity Agency (ANCI) under Ley 21.663 (Resolucion Exenta No. 87, 2025). Using a passive-only, OSINT-based method consistent with ISO/IEC 29147:2018 and Chile's computer-crimes safe harbour (Ley 21.459), we census the foundational disclosure-capability layer (Layer 1: a verifiable disclosure contact). Only 16 of 915 OIVs (1.7%) publish a verifiable RFC 9116 disclosure channel; all four major banks and both telecommunications incumbents lack one entirely. This compares with over 99% adherence under CISA Binding Operational Directive 20-01 (the U.S. federal Vulnerability Disclosure Policy directive; the email-authentication mandate is the separate BOD 18-01). On the secondary email-authentication axis, Chilean OIVs are comparatively strong: DMARC enforcement (quarantine or reject) is present for 146 of 915 designations (16.0%) -- equivalently 16.6% of the 882 measurable domains -- with any-DMARC at 28.0%, above the ~11% top-1M baseline of Tatang et al. (RAID 2021). End-of-life or known-vulnerable components affect an estimated 23.5% (Wilson 95% CI [12-38%]). We propose a four-stage remediation roadmap and release the open-source tool anci-oiv-resolver v0.6.0 (Apache 2.0) for independent reproduction of the OIV-domain mapping. This is a corrected version 2; the email-authentication re-anchor and benchmark-label fix are documented in the 'Changes in v2' note.
arXiv:2606.05878v2 Announce Type: replace Abstract: Foundation models mark a profound paradigm shift in time series modeling, with task-specific models being superseded by general-purpose zero-shot models. Yet, current approaches primarily focus on forecasting, while real-world time series are often irregularly and partially observed, requiring models that can jointly forecast, impute missing values, and handle degraded sampling conditions. To address these challenges, we introduce TS-ICL, a novel probabilistic In-Context Learning encoder--regressor Transformer that unifies forecasting and imputation. TS-ICL formulates time series tasks as timestamp-aligned regression and naturally incorporates covariates by training on synthetic dependency structures generated from a novel causal data prior. Empirically, TS-ICL achieves a new state-of-the-art in imputation, while remaining competitive with leading forecasting foundation models across both univariate and covariate-aware benchmarks. It shows particularly strong performance in forecasting with partially observed look-back windows.
arXiv:2606.05189v2 Announce Type: replace-cross Abstract: Biological neural systems achieve remarkable robustness and adaptability in uncertain environments through sparse, event-driven spike-based information processing and adaptive regulation. Inspired by this paradigm, this paper develops a neuromorhpic disturbance observer (NDO) and control framework that replaces conventional continuous-time signal representations with spike-timing encoding. Both disturbance estimates and control inputs are constructed via integrate-and-fire (IF) neuron dynamics from discrete spike events, yielding intrinsically event-driven updates. An adaptive-threshold triggering mechanism is inspired by spike-frequency adaptation (SFA), enabling history-dependent regulation of spike generation. Simulation results demonstrate that the proposed framework achieves neurally inspired robustness and adaptability, while the adaptive-threshold spiking scheme reduces spike events to 42.6% of the fixed-threshold case under noisy conditions.
arXiv:2606.16163v1 Announce Type: new Abstract: Existing single image dehazing methods are often constrained by computational redundancy in pixel-level optimization and the lack of physical interpretability in implicit neural networks. These limitations hinder the balance between representation efficiency and reconstruction fidelity. To address these issues, we propose Dehaze-GaussianImage, the first zero-shot framework that introduces 2D Gaussian Splatting (2DGS) into the image dehazing domain to break the traditional pixel-grid processing paradigm. Distinct from static convolutional neural networks (CNNs) or Transformers, our approach models hazy images as continuous and dynamically evolvable anisotropic Gaussian fields. Specifically, we propose a novel reconstruction-decoupling zero-shot learning strategy that embeds the atmospheric scattering model into the Gaussian parameter space. This strategy drives Gaussian primitives to adaptively split, clone, and prune during optimization, achieving geometric-level decoupling of the transmission medium and clear textures. Furthermore, explicit structure-preserving constraints are introduced to suppress artifacts commonly caused by traditional physical priors. Experimental results demonstrate that the proposed method achieves state-of-the-art (SOTA) performance in a fully unsupervised manner with minimal parameters, highlighting the potential of explicit Gaussian representation for low-level vision tasks.
arXiv:2606.16561v1 Announce Type: new Abstract: The Internet of Things (IoT) is integral to modern cyber-physical systems. Quantitative cybersecurity assessment in IoT environments remains challenging due to heterogeneous system architectures, evolving threat landscapes, and the limited availability of reliable probabilistic exploitability data. Although Attack Tree Analysis (ATA) provides a structured framework for modelling potential attack paths leading to system compromise, conventional ATA quantification often relies on subjective expert judgement or heuristic scoring schemes, which can introduce uncertainty and reduce analytical reproducibility. This study introduces a data-driven probabilistic security framework for IoT-based safety-critical systems by integrating Model-Based Systems Engineering (MBSE), ATA, and empirical vulnerability data. In the proposed framework, SysML models capture system architecture, from which attack trees are derived. Vulnerabilities are mapped as Basic Attack Steps and assigned exploitation probabilities using the Exploit Prediction Scoring System (EPSS). The attack tree is then represented as a Bayesian Network, enabling probabilistic reasoning, diagnostic inference, and vulnerability criticality analysis. The framework quantifies system compromise probabilities, identifies likely causes of attacks, and prioritises mitigation strategies. By combining architecture-driven modelling with real-world vulnerability intelligence, it provides a rigorous, reproducible approach for cybersecurity risk assessment in complex IoT environments.
arXiv:2606.02957v3 Announce Type: replace Abstract: U.S. financial institutions subject to fair lending laws have been running algorithmic fairness programs for decades. Despite this long history, remarkably little is known about how these requirements operate in practice. In this paper, we offer the first empirical account of how financial institutions test for and mitigate algorithmic discrimination on the ground. In doing so, we shed light on how the regulatory design of fair lending law and regulation have shaped the policies, processes, and practices of fair lending programs. Drawing on 35 semi-structured interviews with participants across the fair lending ecosystem, we find that while financial institutions have a floor of fairness practices aimed at preventing discrimination in lending largely absent in other domains, the specifics of how firms test for discrimination and search for less discriminatory algorithms varies widely. We also find that regulatory supervision via fair lending examinations has been the key driver of compliance work, but that the practical impact of fair lending programs often depends on how well they can navigate competing business incentives, perceived legal tensions, and regulatory uncertainty. Ultimately, our findings highlight the unique role that supervisory authority has played in successfully fostering fair lending practices -- a regulatory design feature that is distinct from other areas of civil rights law and almost completely absent from recent policy proposals for dealing with algorithmic discrimination.
arXiv:2606.13295v2 Announce Type: replace-cross Abstract: In the era of Explainable Artificial Intelligence, there is a renewed focus on single trees for their ease of interpretation. This paper introduces Simultaneous Latent Budget Trees, a probabilistic machine learning framework for classification trees in the presence of a stratification factor such as a temporal, spatial, or demographic variable, acting as a control variable or potential confounder. Standard tree growth procedures are not designed to optimize a conditional split rule. A model-based split rule is proposed in which child nodes are interpreted as latent components of a simultaneous mixture model, such as the Simultaneous Latent Budget Model and its constrained versions, fitted to the parent node. Mixing parameters drive the observations, differently for each group, to the child nodes whereas latent budgets parameters update the response classes profile of each level of the control variable. Parameters are estimated by least squares considering a neural network perspective of the model. An informative tree structure can be interactively visualized with interpretation aids on the node and the paths, including visual pruning and decision tree selection procedure. Suitable measures are proposed to handle an unbalanced response class distribution. The proposed methodology is applied to investigate gender-related differences in disease progression of Amyotrophic Lateral Sclerosis. The SLBT library with the various tree-based algorithms is available in the linked GitHub repository.
arXiv:2606.08151v2 Announce Type: replace Abstract: Modern large language model (LLM) agents do not simply need longer contexts; they need decision-relevant evidence at the moment of action. We study decision-aware context selection: ranking retrieved files, tests, traces, rules, and memories by their expected effect on an agent's next action rather than by semantic similarity alone. We present the Counterfactual-Inspired Context Layer (CICL), which builds an instance context graph, estimates decision-oriented utility for candidate units, and compresses selected evidence into typed memory cards. The same schema can be instantiated with hosted LLM judges, local surrogates, or lightweight rankers, making the selection protocol auditable across model choices. On 50 SWE-bench Verified file-retrieval instances, Qwen3.6-Plus reranking of BM25 top-50 candidates improves hit@1 from 0.58 to 0.78 and MRR@10 from 0.634 to 0.790, with all 2,500 judgments parseable. Controlled diagnostics show that CICL identifies action-critical evidence: removing the top-utility semantic unit reduces F1 from 0.245 to 0.000. In selected-then-compressed mode, memory cards save 44.93 tokens per query while preserving selected evidence. CICL provides a practical layer for measuring, ranking, and compressing decision-critical context for tool-using agents. Code is available at https://github.com/stephen-guan-researcher/CICL.
arXiv:2606.16171v1 Announce Type: new Abstract: Multi-fuel compression ignition (CI) engines offer superior power density and fuel flexibility. However, achieving consistent and optimal combustion phasing across a wide range of operating conditions remains a major challenge, particularly in the presence of modeling uncertainties. This paper presents a novel, data-driven real-time uncertainty compensation framework for combustion control in multi-fuel CI engines. The proposed approach introduces a pseudo-engine speed that enables dynamic adaptation of control inputs in response to uncertainty affecting the engine. To model the underlying combustion process, a Gaussian Process Regression (GPR) model is first trained on available input-output data, capturing the nonlinear and fuel-dependent behavior across varying operating conditions. Control inputs are then synthesized through model inversion of the learned GPR surrogate and augmented with an uncertainty compensator designed to mitigate deviations caused by dynamic variations in operating conditions and model inaccuracies. This integrated control strategy allows for real-time input corrections within a finite number of combustion cycles. Theoretical analysis establishes finite-time convergence guarantees for the proposed controller. Simulation results demonstrate that the proposed method steers the combustion phasing to the desired value in real-time, providing a scalable and adaptive control solution for multi-fuel CI engine operation.
arXiv:2606.16173v1 Announce Type: new Abstract: High-quality time series forecasting is pivotal for real-world decision-making. However, traditional point-wise metrics often fail to reveal complex temporal patterns and align poorly with human intuitive preferences. While the ''LLM-as-a-Judge'' paradigm has revolutionized text evaluation by providing flexible, human-aligned judgment, its application to time series remains largely unexplored. In this paper, we leverage Vision-Language Models (VLMs) as judges for time series forecasting, harnessing their ability to comprehend time series plots grounded in textual information. Specifically, we propose a novel framework integrating micro- and macro-level judgments informed by contextual information to evaluate time series forecasting. To this end, we introduce TimeVista, a comprehensive VLM-as-a-Judge benchmark comprising 5563 time series samples paired with detailed evaluation rubrics. Extensive meta-evaluations demonstrate that VLMs are highly reliable judges, achieving significantly higher consistency with human preferences than conventional metrics. Building upon our benchmark, we comprehensively assess recent Time Series Foundation Models (TSFMs) under the VLM-as-a-Judge paradigm. Our results demonstrate that VLMs serve as robust and interpretable judges, providing a comprehensive, human-aligned standard for evaluating time series models.
arXiv:2606.16217v1 Announce Type: cross Abstract: Exact diagonalization of quantum lattice Hamiltonians returns floating-point eigenvalues whose accuracy is not certified: rounding error, eigensolver behaviour, and ill-conditioning can corrupt a result without warning. We present \texttt{symveig}, a pure NumPy/SciPy package that computes rigorous, machine-checkable enclosures of all eigenvalues of a Hermitian matrix, with an optional symmetry-sector decomposition. For a matrix that commutes with an abelian conserved quantity diagonal in the working basis (for example the total magnetization of a spin model), the package verifies each symmetry sector independently. Because each sector block is much smaller than the full matrix, this yields enclosures that are both tighter (by a factor of $3$-$9$ across system sizes $L = 4$-$12$) and dramatically faster (a wall-clock speedup of up to $130\times$ at $L = 12$) than verifying the full matrix, while never forming or diagonalizing it. Every enclosure half-width is a guaranteed upper bound on the distance from a computed eigenvalue to the nearest true eigenvalue under IEEE~754 round-to-nearest arithmetic, obtained by explicit floating-point error analysis with no heuristic slack. The implementation requires neither INTLAB nor MATLAB, bringing rigorously verified eigenvalue enclosure into the standard scientific-Python stack used in computational physics. We validate the package on $1$D Heisenberg (open and periodic), $J_1$-$J_2$ Heisenberg, and $2$D Heisenberg lattices, confirming that every computed eigenvalue is contained in its enclosure across all tested configurations.
arXiv:2508.19610v3 Announce Type: replace Abstract: Question-and-answer platforms such as Stack Overflow are an important way for software developers to share and retrieve knowledge. However, reusing poorly understood code can lead to serious problems, such as bugs or security vulnerabilities. To better understand how code comments affect the perceived helpfulness of Stack Overflow answers, we conducted an online experiment simulating a Stack Overflow environment (n=91). The results indicate that both block and inline comments are perceived as significantly more helpful than uncommented source code. Moreover, novices rated code snippets with block comments as more helpful than those with inline comments. Interestingly, other surface features, such as the position of an answer and its answer score, were considered less important. Moreover, the content of Stack Overflow has been a major source for training large language models. AI-based coding assistants such as GitHub Copilot, which are based on these models, are changing the way Stack Overflow is used. However, our findings have implications beyond Stack Overflow. First, they may help to improve the relevance also of other community-driven platforms, which provide human advice and explanations of code solutions, complementing AI-based support for software developers. Second, since chat-based AI tools can be prompted to generate code in different ways, knowing which properties influence perceived helpfulness can lead to more targeted prompting strategies to generate readable code snippets.
arXiv:2606.06834v2 Announce Type: replace Abstract: High-grade gliomas integrate into neural circuits through functional synapses with neurons, raising the question of which noncoding elements shape synaptogenic gene expression in tumor cells. The regulatory program written across the dark genome, what we call the $\textit{dark regulome}$, is the natural substrate to probe, and sequence foundation models offer a zero-shot route through in-silico mutagenesis (ISM); yet likelihood-based scoring is tautologically coupled to local sequence predictability, leaving the regulatory interpretation underdetermined. Across three architecturally distinct foundation models (Caduceus-Ph, HyenaDNA, Enformer) and 30,448 dark genome elements at 92 glioma-relevant loci, we introduce a residualization-and-permutation diagnostic that separates predictability-driven from regulation-driven RIS variance. A sharp 10kb proximal-regulatory horizon survives every control we apply, but the LM-derived element-class hierarchy does not: a six-feature linear baseline matches Caduceus top-decile membership at AUC $= 0.985$. Cross-architecture decomposition cleanly separates a sequence-predictability layer (the two language models co-rank long well-predicted transposable elements) from a regulatory-output layer (Enformer alone retains residual cCRE-discriminative signal), with literally zero overlap between the two top-100 lists. Conservation, brain cis-eQTL, and STRING-PPI cross-checks then anchor what biology survives: top-100 elements across all three models are $3.3\times$ enriched per model for matching brain eQTLs ($p_\mathrm{emp} < 5\times 10^{-3}$), while a tempting transposable-element regulatory layer and a striking NRXN1+NLGN1 protein-pair convergence both fail proper permutation tests once those tests are constructed. We deliver the diagnostic as a general methodological tool for any ISM-based regulatory study.
arXiv:2606.16175v1 Announce Type: new Abstract: Longitudinal personal albums are weak-schema multimodal databases: noisy perceptual records whose key facts require joins across faces, text, timestamps, locations, and repeated events. Existing visual, video, document, and lifelog benchmarks test sub-problems, but not album-scale profile reconstruction with social identity binding and evidence citation. Benchmarking this task is difficult because the ground truth needed for evaluation--owner profiles, social graphs, face-name maps, and evidence provenance--is private state that real albums cannot safely release. We introduce PAL-Bench, a controlled benchmark for evidence-grounded reconstruction under a public-record contract. Its Evidence Compiler builds latent private worlds, programs target-level evidence paths, renders album pixels, re-measures them through perception pipelines, and exports audited public/private views. Agents receive only perception-derived public records; targets, identifier maps, and evidence paths remain hidden. PAL-Bench contains 50 synthetic users, 36,659 public photo records, and 2,799 targets over owner facts, identities, and relations. A privacy-preserving audit with 10 participants confirms that PAL-Bench evidence structures match real private albums, though equivalent releases remain privacy-prohibitive. Across seven systems and two compute-matched diagnostics, a seven-metric protocol reveals a gap between plausible profile summarization and faithful social reconstruction: systems recover some owner facts but struggle with recurring identities and evidence citation. PAL-TRACE, a reference framework that freezes identity bindings before owner-fact mining, performs best but leaves hard identity resolution far from solved. PAL-Bench provides a testbed for perceptual entity resolution, multimodal data integration, temporal evidence aggregation, and provenance-aware structured prediction.
arXiv:2606.15418v1 Announce Type: new Abstract: We solve, exactly, the problem of minimum settling-time PI control of a pure delay process K e^{-Ls} under the hard time-domain constraint of zero overshoot, y(t) <= 1 for all t. The closed loop is a neutral delay system whose step response is piecewise polynomial on the delay segments, with geometrically decaying jump discontinuities at the segment boundaries t = kL. The constrained optimum is characterized by an equioscillation-type contact structure whose active contacts sit at echo boundaries: kink maxima grazing the setpoint, jumps landing on the settling-band edge, and boundary troughs anchored to it. The number of contact equations equals the number of gains, so the optimum is exactly computable for every band delta. In a closed-form regime, delta in [(3-2 sqrt2)/4, (3-2 sqrt2)/2] approx [4.29%, 8.58%], the optimal gains are independent of delta: K Kp = 1 - sqrt2/2, K Ki L = sqrt2/2, and the optimal settling time is Ts*(delta) = (4 - sqrt2 - 2 sqrt(delta)) L. Outside this window the optimum solves an explicit two-equation polynomial system per regime, and Ts*(delta) is a staircase with exact flats at integer multiples of L from jump-landing pinning. As delta -> 0 the optimal gains converge to K Kp = e^{-2}, K Ki L = 4 e^{-2}, the generic multiplicity-induced-dominancy (GMID) point of the neutral quasipolynomial. The GMID response satisfies the hard constraint and uniquely maximizes the decay rate; yet at every finite delta the delta-adapted optimum strictly beats the fixed GMID tuning, by about 40% at delta = 2%. The MID point is thus the limit of the optimal gains without ever being the optimal tuning. A numerical extension to first-order-plus-time-delay plants quantifies the speed/robustness trade across Ms in [1.39, 1.76].
arXiv:2606.15419v1 Announce Type: new Abstract: Objective: To enhance the accuracy, interpretability, and robustness of large language models (LLMs) in medical question answering (MedQA). Method: We designed a multi-agent peer-reviewed reasoning method in which multiple LLM agents independently generate chain-of-thought reasoning with candidate answers, then act as peer reviewers to evaluate each other's reasoning for factual correctness and logical soundness. The highest-rated reasoning chain is selected to produce the final answer. Experiments were conducted with five state-of-the-art LLMs (Llama-3.1-8B, Qwen2.5-7B, Phi-4, DeepSeek-LLM-7B, GPT-oss-20B) on three benchmark datasets: HeadQA, MedQA-USMLE, and PubMedQA. Performance was compared against single-model chain-of-thought reasoning and chain-of-thought-based majority voting. Results: Peer-reviewed reasoning consistently outperformed both baselines. The best model combination achieved an average accuracy of 0.820 across datasets, exceeding the strongest single model (0.777) and majority voting ensembles (up to 0.789). The method also scaled effectively with more participating models, while peer assessments reliably distinguished high- from low-quality reasoning chains. Conclusion: The proposed multi-agent peer-reviewed reasoning method enables LLMs to act as both solvers and evaluators, yielding superior performance in MedQA. By emphasizing reasoning quality rather than answer agreement alone, this approach improves accuracy, interpretability, and robustness, offering a promising direction for trustworthy biomedical AI systems.
arXiv:2606.08395v2 Announce Type: replace Abstract: Prime numbers are traditionally studied through numerical, probabilistic, and analytic frameworks. In this work, we introduce the concept of a prime event language, in which arithmetic phenomena are represented as symbolic event sequences and analyzed using tools from information theory and stochastic processes. Using all primes up to $N = 5 \times 10^9$ (234,954,223 primes), we construct event languages based on twin-prime occurrences and record prime-gap events. We investigate their statistical properties through finite-order Markov models, train/test validation, mutual-information analysis, and information-horizon measurements. For the Twin Prime Event Language, first-order Markov modeling reduces test-set cross entropy from 0.325350 bits to 0.319949 bits, corresponding to an information gain of approximately 0.0054 bits. This gain survives out-of-sample validation and therefore reflects genuine statistical structure rather than overfitting. Mutual-information analysis independently confirms the Markov results and shows that measurable dependence is concentrated almost entirely at lag 1. The mutual information decreases from approximately $5.96 \times 10^{-3}$ bits at lag 1 to approximately $5.07 \times 10^{-7}$ bits at lag 2 (approximately 11,700-fold reduction), representing a reduction of more than four orders of magnitude. Beyond lag 2, residual information fluctuates near the statistical noise floor. These results indicate that prime event languages are neither perfectly memoryless nor strongly predictable. Instead, they exhibit weak but reproducible short-range statistical structure characterized by first-order dependence and an effective information horizon of approximately one event.
arXiv:2606.16267v1 Announce Type: cross Abstract: Photo-induced ferroelectricity in the quantum paraelectric SrTiO$_3$ involves the dynamical interplay between a coherently driven Ti-O stretching vibration and multiple structural degrees of freedom, including antiferrodistortive rotations, strain, and the polar mode instability. In the high-temperature cubic phase, in the absence of average antiferrodistortion, time-resolved X-ray diffuse scattering has evidenced a correlation between a photo-induced reduction in antiferrodistortive fluctuations and the emergence of ferroelectric order. Here, we complement these measurements with time-resolved elastic X-ray diffraction in the low-temperature tetragonal phase, in which antiferrodistortive fluctuations are small but a finite average rotation has set in. In this phase, we observe a long-lived reduction of the equilibrium antiferrodistortive rotation angle. A unified theory of the nonlinear lattice dynamics based on first-principles calculations describes the dynamics in both high-temperature cubic and low-temperature tetragonal phases, providing a basis for light-induced ferroelectricity in SrTiO$_3$.