arXiv:2607.14005v1 Announce Type: new
Abstract: Driving-world generation has emerged as a core capability for scalable autonomous-driving simulation, yet existing methods remain limited in object-level controllability and long-horizon stability. We present M$^\text{4}$World, a Multi-view and Multimodal generative driving world model that synthesizes future surround-view video streams and synchronized LiDAR scans while supporting interactive object Manipulation and stable Minute-long streaming. Fine-grained object manipulation is realized through a flexible conditioning interface that supports explicit control over both the spatial layout and visual appearance of individual objects. Stable minute-long streaming, on the other hand, is achieved through a multi-stage training framework that enables online causal generation in only four denoising steps while maintaining coherent world dynamics throughout extended rollouts. Building on these components, we introduce an efficient few-clip post-training as well as a suite of visual reference-conditioned generation models, preserving general generation ability while allowing rare-case customization for long-tail controllability. To assess controllability beyond realism, we further introduce an automated VLM-based judging pipeline that evaluates scene-level condition adherence, view-wise object controllability, and cross-view object consistency. Comprehensive experiments show that M$^\text{4}$World consistently delivers high generation quality, precise controllability, and stable minute-long streaming. Together with downstream long-tail augmentation and scene editing, these results demonstrate the potential of M$^\text{4}$World for controllable, scalable driving simulation.
Science Journals
arXiv:2607.13296v1 Announce Type: new
Abstract: Nonlinear optical interactions are typically driven by coherent sources. Recently, however, incoherent sources have also been shown to offer new opportunities in nonlinear optics. Here we study the effects of the source incoherence in second harmonic generation. We derive and experimentally validate an analytic expression for the far-field coherence function at the second harmonic frequency, which is a generalization of the well-known van Cittert-Zernike theorem. Unlike the linear optics case, here the results depend not only on the source spatial distribution, but also on the nonlinear structure of the crystal. We further show, by performing Young's double-slit measurement, that the coherence at the second harmonic increases as the light propagates inside the nonlinear crystal. Our results establish nonlinearity as a degree of freedom for optical coherence, opening a new research area at the interface of statistical optics and nonlinear optics.
arXiv:2607.13985v1 Announce Type: new
Abstract: Active lenses based on metasurfaces have been demonstrated across broad applications, and their response speeds remain limited to the microsecond range. Here we report the experimental demonstration of a magneto-optical (MO) active lens based on a magnetic hologram formed on a magnetic garnet film. A Fresnel zone plate pattern was created using an array of permanent magnets, focusing a 633 nm laser beam of 12.5 mm diameter to a spot size of 0.7 mm at a focal length of 3.1 m with a modulation depth of ~91%. The focusing state was reversibly switched by applying an external magnetic field pulse. Through optimization of the magnetic field application system, a switching speed of 10.8 $\mu$s was achieved, surpassing the operating speeds of conventional active lenses. The intrinsic picosecond-scale response of the MO effect indicates that further improvement in switching speed is achievable, establishing this work as a proof-of-concept for ultra-fast active lenses.
arXiv:2607.13815v1 Announce Type: new
Abstract: Background: Replication studies in software engineering are increasingly common, yet their interpretation remains uncertain and inconsistent because assessments frequently rely on loosely defined or ad hoc criteria. Aim: This study aims to document how replication study outcomes are currently assessed in empirical software engineering, identify problems arising from inconsistent criteria and propose a principled framework for meaningful evaluation. Method: We conducted a systematic review of replication studies, with the search covering recent empirical software engineering replications (2021--2025). For each study, we extracted the criteria used to assess replication outcomes and analysed these for heterogeneity, logical consistency, and alignment with established statistical principles. Results: A total of 10 replication studies were located. The analysis reveals substantial heterogeneity in assessment practices, with contradictory criteria applied to similar data, limited acknowledgement of measurement uncertainty, and an absence of shared standards. We propose a principled framework grounded in statistical, methodological, and measurement considerations, and demonstrate its application through worked examples. Conclusions: Adopting consistent and transparent assessment principles would reduce ambiguity, improve comparability, and support more reliable evidence accumulation in software engineering replication research.
arXiv:2607.13179v1 Announce Type: new
Abstract: User Experience (UX) is recognized as a critical factor for the success of digital products, particularly in software startups, environments marked by time constraints, limited resources, and low maturity in design practices. Building Minimum Viable Products (MVPs) through low-fidelity prototyping represents a well-established strategy for rapid validation cycles at reduced cost. A systematic literature mapping, however, revealed gaps in the ecosystem of available tools: a predominance of general-purpose solutions adapted for prototyping, the absence of integrated methodological guidance, and the incipient use of Artificial Intelligence in the design process. This paper presents SoftBoard, a web-based tool for the creation and evaluation of low-fidelity prototypes in the context of MVP development. The tool integrates a prototype editor, team-based project organization, and a multi-agent system based on large language models that supports requirements elicitation and refinement, automates prototype generation, and evaluates interface quality based on usability heuristics. This integration aims to reduce reliance on prior UX expertise, standardize the prototyping process, and support teams in building MVPs aligned with user needs. As future work, a feasibility study with software professionals is currently underway.
arXiv:2607.13311v1 Announce Type: new
Abstract: Retrieval in the SQL setting has largely been studied as the task of finding, within a large collection of SQL statements, the statement that answers a natural-language question. At scale, however, a more fundamental retrieval problem precedes generation: schema retrieval, identifying the tables and columns a question requires in a database that may contain thousands of them, far more than fit in a model's context. We argue that this step warrants first-class evaluation. To this end, we recast five text-to-SQL datasets (Spider, BIRD, BEAVER, and two LiveSQLBench variants) as retrieval tasks at both table and column granularity, covering realistic and enterprise-scale schemas under two document representations, and we show that off-the-shelf text and code embedders transfer poorly to this setting. We then propose corpus-adaptive fine-tuning: natural-language queries are synthesized directly from the target schema corpus, granularity-aware hard negatives are mined, and a 305M-parameter embedder is fine-tuned contrastively. This procedure raises average recall@10 from 60.4 to 75.6 (nDCG@10 from 51.9 to 68.0), making the 305M model the strongest retriever under one billion parameters and competitive with state-of-the-art embedders of 4-8B parameters, more than an order of magnitude larger. The same recipe improves an 8B state-of-the-art embedder from 77.8 to 78.4 recall@10, matching the best result on the benchmark and indicating that the adaptation is backbone-agnostic. Leave-one-corpus-out experiments and a leakage audit show that these gains reflect a transferable schema-retrieval ability rather than memorization of the evaluation data. Our results establish schema linking as a standalone retrieval task and lightweight, label-free corpus adaptation as a practical route to deploying it at enterprise scale.
arXiv:2607.13313v1 Announce Type: new
Abstract: We demonstrate that an array of integrated coupled waveguides can be designed to simultaneously perform the spectral filtering and routing of light. The framework that we present here operates in a regime opposite to the shortcut-to-adiabaticity techniques, but we show it nevertheless also only requires smooth, slow, and small, modulations of the array's parameters. Our design strategy rely on the analogy which exists between the temporal evolution of an Hamiltonian and the propagation of light in an array of coupled integrated waveguides, mimicking spatially in photonics phenomena analogue to the electronic transitions. Interestingly, the chromatic dispersion restores in photonics the correspondence between spatial propagation effect and wavelength dependence. Therefore our framework is able to fully reproduces temporal effects such as resonant transitions, making it suitable for filtering applications. As a design involving modulations, it shares similarities with Bragg-Gratings (BGs): but it adds some extra features that make it complementary in terms of applications and functionality. First it adds routing capability with the seamless management of multi-waveguide configuration. And secondly the framework we present requires only wavelength-scale modulations, order of magnitudes slower than for BGs. The effects we present here may not interfere with BGs geometry and functionality, and could even be combined with the latter.
arXiv:2607.13823v1 Announce Type: new
Abstract: The problem is to select k columns with maximal volume from a real mxn matrix X. We show that the logarithm of the volume is a submodular function on columns of X, and for full column-rank matrices X with sufficiently large singular values, it is a non-negative non-decreasing function. As a consequence, traditional Businger-Golub QR with column pivoting is a greedy algorithm, with a relative error of at most 37 percent. In contrast, Gu-Eisenstat strong rank-revealing QR is a 1-interchange algorithm, with a relative error of at most 50 percent. The higher accuracy, under this metric, of the simple QR with column pivoting confirms its well known effectiveness in practice. For general matrices, we show that Businger-Golub QR is a greedy algorithm, and Gu-Eisenstat QR a 1-interchange algorithm for maximizing the trace of the upper triangular matrix in a QR factorization of X, with Businger-Golub again having a better error bound. This is extended to finding kxk submatrices of maximal volume in symmetric positive-definite matrices.
arXiv:2607.14006v1 Announce Type: new
Abstract: Penetration testing traditionally evaluates whether adversaries can exploit weaknesses in software, infrastructure, configurations, or operational controls to achieve security-relevant compromise. This paradigm remains necessary for AI-enabled systems, but it is no longer sufficient. In such systems, adversaries may influence prompts, retrieved content, sensor inputs, training data, memory, tools, or human-AI interaction loops to alter system behavior without directly compromising the underlying infrastructure. This paper reframes penetration testing for AI-enabled systems as objective-driven behavioral evaluation. We define an AI-enabled system as one in which learned models materially influence behavior affecting operational outcomes, and we define AI-enabled penetration as the feasible induction of AI-governed behavior that violates one or more operational objectives under an explicit threat model. This definition preserves conventional penetration testing while extending it to adversarial pathways such as prompt injection, indirect prompt injection, data poisoning, sensor manipulation, retrieval poisoning, tool misuse, and agentic misalignment. We further propose a testing workflow that identifies operational objectives, maps AI-governed behavior, analyzes adversarial influence surfaces, defines behavioral failure criteria, executes scenario-based tests, and reports evidence linking adversarial action to objective violation. A running example involving an AI-enabled security operations center assistant illustrates how penetration may occur through behavioral influence rather than infrastructure compromise. Together, the definitions, workflow, and example provide a technical framework for evaluating adversarial success in deployed AI-enabled systems.
arXiv:2607.13319v1 Announce Type: new
Abstract: High-speed off-road autonomy requires precise closed-loop control for a target vehicle while remaining robust across changing terrains. Recent forward kinodynamic (FKD) prediction foundation models suggest a promising path, starting from a generalist model and specializing it to the target platform. However, effective specialization remains challenging, as it often requires substantial real-world data, and models adapted to one setting can still overfit to specific terrains or driving regimes. We present OptCar (Optimized Car), a recipe for bridging the gap from generalist to specialist FKD models that preserves cross-terrain generalization while optimizing performance for a specific vehicle. $\texttt{OptCar}$ introduces a history-conditioned dynamics adaptation module that encodes recent state-action observations into a dynamics context token, and then fine-tunes the generalist model using limited real-world data together with targeted synthetic rollouts from environment-specific system identification. In closed-loop model predictive control (MPC) experiments across three terrains and an out-of-distribution cart-pulling task, the largest gains appear at 6~m/s, the highest speed evaluated and the regime in which slip dominates tracking error. On vegetation and dirt, the most slip-diverse terrain, OptCar reduces 6~m/s trajectory tracking error by roughly 55% relative to a fine-tuned AnyCar baseline, and remains the most accurate even when an unseen cart payload changes the dynamics. With only 5 minutes of real data per terrain, OptCar is competitive on road with a specialist trained on 30 minutes of road data, and substantially outperforms it once the terrain changes.
arXiv:2607.13325v1 Announce Type: new
Abstract: Quantum Processing Units promise speed-ups for selected computational problems, including combinatorial optimisation, but their industrial utility remains an open challenge. We study an industrial variant of the Job-Shop Scheduling Problem using quantum, quantum-inspired, and classical methods across three platforms: IBM Quantum, the D-Wave Quantum Annealer, and the Fujitsu Digital Annealer. By tailoring formulations to hardware-specific constraints, we show that hardware-software co-design is essential for solution quality and scalability. We benchmark all approaches against an exact classical solver and a MILP formulation, evaluating runtime, solution quality, and scalability. Our results indicate that quantum and quantum-inspired optimisation can support industrial solver selection, integration in classical workflows, modelling decisions, and early proof-of-concept development, while suggesting a potential path towards improved approximations for industrial scheduling.
arXiv:2607.14008v1 Announce Type: new
Abstract: In this paper, we introduce Lighthouse RL, a sample-efficient reinforcement learning (RL) approach for analog circuit sizing. Traditional methods lack generalization across different performance targets, while standard RL approaches waste resources exploring unpromising regions. Our method addresses these inefficiencies through a strategic reset strategy that initializes episodes from high-performing configurations discovered during training, called "lighthouses". These states, which are closer to the target objectives, guide exploration toward promising regions. When compared to RL and Bayesian optimization methods from the literature, we demonstrate the effectiveness of our approach on a 2D benchmark problem and on two analog circuits, showing significant improvements in sample efficiency (up to 1.72x faster), optimization performance (100% vs. 0-87% success rate), generalization (75% vs. 0-50% extrapolation success), and objective maximization. This efficiency is particularly valuable for computationally expensive black-box optimization problems, and our reset strategy can be used as a plug-and-play enhancement for any RL-based optimization approach.
arXiv:2607.13843v1 Announce Type: new
Abstract: Efficiently breaking time-reversal symmetry at the subwavelength scale remains a cornerstone challenge for advanced electromagnetic wave manipulation. This work presents a rigorous analytical framework, based on cylindrical wave expansion, to investigate and optimize the nonreciprocal scattering of transverse electric waves by magnetically biased plasmonic rods. An intuitive metric is introduced to quantify the breaking of time-reversal symmetry via the asymmetric lifting of degeneracy between azimuthal modes of opposite angular momentum, hosted by the gyrotropic particles. Leveraging this metric, a comprehensive mapping of the multiparametric space of operational frequency, cyclotron frequency, and cylinder optical size isolates regimes of maximum nonreciprocity. A detailed multipolar decomposition reveals that this extreme behavior stems from the phase-matched asymmetric excitation and interference of localized electric and magnetic dipole modes. Moving from individual, isolated meta-atoms to collective photonic systems, the optimized cylinders are arranged into a periodic grating. Under oblique incidence, the combination of geometric asymmetry and magnetic mode splitting, forces the metasurface to transmit light in a totally different way when excited by opposite sides. The reported findings and design principles offer a versatile blueprint for the development of dynamically tunable flat-optics isolators, directional transceivers, and advanced wavefront routers.
arXiv:2607.13336v1 Announce Type: new
Abstract: Recent diffusion-based video generation models have enabled high-quality personalized video customization through both tuning-based pipelines, which fine-tune a video diffusion model, and reference-based pipelines such as image-to-video generation. However, these capabilities raise serious concerns about personal privacy, identity ownership and intellectual property protection. Existing anti-customization works focus on protecting images, while protection for videos against both reference- and tuning-based customization remains largely underexplored. Protecting videos in this setting raises three challenges: (i) Image-level perturbations, optimized frame by frame, cannot survive temporal compression by 3D video VAE. (ii) A video-level perturbation optimized on a single video is vulnerable to temporal editing and fails to protect unseen videos. (iii) Temporally inconsistent perturbations are not robust to temporal attacks. To address these challenges, we propose Temporally Consistent Universal Adversarial Perturbations (TC-UAP), the first protection method against both reference- and tuning-based video customization. TC-UAP optimizes an identity-level multi-frame UAP over sliding windows from multiple videos, accounting for local temporal dependencies induced by temporal compression in video VAE and enabling a single perturbation to protect unseen videos of varying lengths. Moreover, we introduce intrinsic temporal modeling and an extrinsic surrogate temporal-attack loss, which make the perturbation temporally consistent and robust to unseen temporal attacks. Empirically, quantitative and qualitative results show that TC-UAP achieves the strongest identity protection compared with existing methods under both reference- and tuning-based video customization, and remains robust under multiple unseen temporal attacks.
arXiv:2607.13845v1 Announce Type: new
Abstract: Biometric template protection (BTP) must secure stored templates while tolerating intra-class variations. Existing methods rely on protected-domain similarity matching, error correction, or predefined-template mappings, potentially retaining exploitable similarity structures, introducing helper-data risks, depending on artificial targets, or coupling protection to specific modalities. Storing only cryptographic hash digests eliminates directly comparable representations and conceals pre-hash templates, but hash-based exact-match verification requires genuine samples to generate identical intermediate templates before hashing. Identity classification is naturally suited to this requirement because it maps variable biometric samples to stable and discriminative identity-level outputs. Based on this insight, we propose Multiple Permuted-Label Classifier Encoding (MPLCE). Through classifier-specific label permutations, MPLCE assigns each identity different labels across multiple classifiers. The predicted labels are encoded and concatenated to form an intermediate template, preventing repeated encodings of a single identity label and enlarging the effective candidate space while preserving classification consistency. The template is randomized with an application-specific XOR string and cryptographically hashed, enabling exact-match verification without error correction codes or biometric-dependent helper data. Using modality-specific classifiers, MPLCE retains the same template generation and protection procedure across modalities. On four face and two iris datasets, MPLCE achieves competitive performance, including a GAR of 98.61\% at a FAR of 5.51\(\times\)10\textsuperscript{-5}\% on YTF and a GAR of 99.10\% at a FAR of 0.00\% on CASIA-Iris-Lamp. Security analyses and attack evaluations support its irreversibility, revocability, and unlinkability under the threat model.
arXiv:2607.13970v1 Announce Type: new
Abstract: A novel analytical framework for power system strength was recently introduced in the IEEE Transactions on Power Systems, providing a unified formulation for assessing voltage and frequency strength. Building upon this formulation, this paper addresses a series of practical challenges for translating the theoretical framework into a real-world application. In particular, simplified analytical solutions for network-wide bus-level strength metrics are provided, together with compact expressions to capture the impact of relevant devices on strength. In addition, novel normalized strength metrics at a device level are defined, enabling the comparison of strength across different systems. A strength source model is introduced to study the behavior of devices under varying strength conditions. Finally, the framework is implemented in a real-world study case, demonstrating its applicability and potential as a practical tool for a comprehensive strength assessment.
arXiv:2607.13346v1 Announce Type: new
Abstract: Alignment faking is dangerous because a model can appear compliant under monitoring while preserving behavior it would reveal when unmonitored. When no scratchpad is visible, behavior alone cannot distinguish strategic from genuine compliance. We ask whether hidden states reveal what outputs hide.
We run a 13-model sweep for naturally-emerging faking, then probe and steer hidden states on the two models that fake. Natural faking appears only in Qwen3-32B (+18.2pp) and Llama-3.1-8B (+24.4pp at n=10, p<10^-15), while explicit scratchpad self-reports are rare (a Claude Opus 4 judge flags faking reasoning in 1.4% of Qwen and 0.0% of Llama compliant scratchpads). Both fakers show an asymmetric refusal residue: monitored compliance shifts toward refusal (d=0.32 Qwen, d=0.33 Llama, both p<0.0001), while refusal representations do not move (p=1.00).
Per-sample detection is model-conditional. Under leakage-free leave-one-query-out probing, detection survives on Llama (AUROC 0.87) but collapses to chance on Qwen (0.43), and cross-model transfer is null. Steering over 2,000 runs barely changes compliance (|h|<0.08), so the detected direction can flag faking but does not by itself control it.
Standard residualized probing leaks across folds and reaches AUROC 0.63 on a control where no faking can occur; naive linear probes reach a meaningless AUROC 1.0; and conventional MLPs overstate detectability by 0.2-0.3 AUROC. For future alignment-faking detection work, we release a five-control measurement framework: multi-token extraction, refuse-vs-refuse confound checks, per-fold residualization, leave-one-query-out evaluation, and orthogonality-constrained probing.
arXiv:2607.13919v1 Announce Type: new
Abstract: Nonnegative Matrix Factorization (NMF) is a fundamental tool in unsupervised learning, which approximates a nonnegative matrix by the product of two low-rank nonnegative factors. The Kullback-Leibler (KL) divergence is best suited to measure the data to model discrepancy when the decomposed data sample follows a Poisson distribution, which is the case for count datasets such as term-document matrices or images. Most KL-NMF algorithms in the literature minimize a separable majorant of the loss to find their next iterate. We argue that this method has reached its limits and propose to use instead the second-order Taylor expansion of the loss, leading to a Newton-type method. We minimize this non-separable surrogate by proposing a generalization of the well-known HALS algorithm. This yields an efficient KL-NMF algorithm which provably converges and which competes favorably with state-of-the-art algorithms on a large variety of datasets.
arXiv:2607.14013v1 Announce Type: new
Abstract: Covert communication enables Alice to transmit to Bob while making the transmission difficult for Willie to detect. We study a scalar Gaussian covert-overlay model in which Alice's low-power covert signal is superimposed on an aggregate public component generated by Alice or other trackable sources. Willie is given all physically obtainable side information, including protocol details, timing, pilots, channel estimates, and calibration information, and subtracts his best estimate of the public component before testing. Covertness is imposed on the resulting residual through a relative-entropy constraint with budget $\delta$ conditioned on Willie's side information. In the stationary case, the residual under no covert transmission has variance $\sigma_0^2=\sigma_W^2+\sigma_e^2$, where $\sigma_W^2$ is Willie's receiver-noise variance and $\sigma_e^2$ is the irreducible cancellation error. Over $n$ channel uses, the maximal reliably transmissible covert payload is $R_C^\star\sqrt{n}(1+o(1))$ bits, where $R_C^\star=\frac{\sigma_0^2}{\sigma_B^2\ln 2}\sqrt{\delta}$, and $\sigma_B^2$ is Bob's receiver-noise variance. Thus, the square-root-law (SRL) constant is governed by the variance at Willie's actual detector input, not by receiver noise alone. Low-power Gaussian signaling achieves this constant, and a matching converse establishes first-order optimality within the conditioned additive Gaussian innovation model. For known time-varying conditioned residual variances, we also derive the first-order allocation, which assigns more covert power to larger residual variances. The results require a Gaussian post-cancellation null residual with known conditioned variance; non-Gaussian residuals and fixed non-vanishing variance uncertainty are outside the scope of this paper.
arXiv:2607.14041v1 Announce Type: new
Abstract: Historical Manchu OCR must accommodate various visually distinct writing styles, including regular script, running script, and the semi-cursive chancery hand used in palace memorials, despite limited labeled data. We study a multi-expert system that reuses checkpoints from an iterative fine-tuning process as domain specialists and uses a lightweight page-level image classifier to dispatch pages by visual style. When the checkpoint pool lacks a suitable specialist, we train an additional expert for that domain. On three frozen test sets, the routed system matches the selected specialist for each style at two-decimal precision: 0.30 percent CER on regular script, 1.57 percent on memorials, and 4.83 percent on running script. The router achieves 99.3 percent page-level domain accuracy and matches the domain-label oracle at the same precision. Two of the three selected specialists were not trained specifically for their final domain; only the running-script expert was trained with that domain as its target. We report the evaluation protocol, router design, and per-page predictions to make the comparison reproducible.
arXiv:2607.13562v1 Announce Type: new
Abstract: Knowing when to say "I don't know" is fundamental to human judgment, yet AI assistants offer a fluent answer to almost any question. In five experiments (N = 3,132; four preregistered, one direct replication), participants answered difficult questions and could always decline to respond. We engineered the questions so that AI advice was wrong, separating AI use from its accuracy. Merely having access to AI nearly eliminated participants' willingness to suspend judgment, and this held whether the advice was actively requested or simply displayed. Consequently, participants answered more questions but were correct about a third as often as when AI was unavailable-yet their confidence nearly doubled. Incentivizing accuracy and penalizing inaccuracy led participants to seek and follow AI advice less, answer more accurately, and suspend judgment more often, though still far less than when AI was unavailable. As AI suggestions grow ubiquitous and unsolicited, they may not simply affect answer accuracy; they may even alter the metacognitive threshold at which people decide whether they know enough to answer.
arXiv:2607.13683v1 Announce Type: new
Abstract: An LLM agent's real-task performance is shaped as much by the harness around its model as by the frozen model itself: its prompts, injected knowledge, runtime control, and configuration. In deployment the harness is often the only lever available, so improving it automatically is the natural way to raise performance without touching the weights. The hard part is not generating changes but knowing which one truly helped. Self-generated feedback is noisy, and an apparent gain can be a measurement artifact or an edit that merely overfits the tasks it was tuned on. We present a self-evolving agent-harness framework that separates proposing changes from crediting them: a language model diagnoses failures and proposes patches, while all sampling, measurement, and significance testing are owned by deterministic code, so every credited improvement is trustworthy by construction. Patches populate a gated, categorical quality-diversity archive (GSME) keyed on the (WHERE x WHY) pathology an edit addresses rather than the tasks it fixes, an anti-overfitting inductive bias; generalization is measured on a sealed test scored only after evolution. Across seven domains with a frozen open-weight model, the harness is train-selected and scored once on a sealed test; its credited gains there are +9 to +15.5pp and retain 86-147% of the training gain, evidence they generalize rather than overfit. The winning patch tracks the model's dominant pathology, not its size or family: changing the model can change the pathology and the patch, while the same pathology-to-patch match recurs across two model families. What transfers is the diagnose-and-credit loop, not any specific harness.
arXiv:2607.14044v1 Announce Type: new
Abstract: By 2030, 59 of every 100 workers will need reskilling or upskilling, yet the average time to close an enterprise skills gap grew from roughly 3 days in 2014 to 36 days in 2018. Most current frameworks accelerate single stages of upskilling programs and generally lack industry validation. We present an end-to-end framework that applies AI acceleration across five stages of knowledge acquisition, content development, content review and verification, teaching, and assessment development; with a strong focus on both production and learning efficiency. Three strong external signals validates the framework: the US National Association of State Boards of Accountancy reviewed and approved an upskilling program built on the framework for continuing-professional-education credits; 3 learners followed the program and passed the NVIDIA Certified Professional in Agentic AI exam in a significantly short amount of time, with 14 more in progress; the program's knowledge base supports complex downstream analysis such as the production of a robust 1,267 risk item dataset for managing multi-agent AI system risks.
arXiv:2607.14045v1 Announce Type: new
Abstract: This manuscript explores the integration of Large Language Models (LLMs) into the field of qualitative and mixed-methods social network analysis (SNA). We argue that the primary focus of this integration should be on enhancing the depth and rigor of qualitative SNA, rather than on replacing human researchers with automated systems. We begin by outlining the core principles of qualitative and mixed-methods SNA, emphasizing the importance of understanding the meaning of ties, the role of narratives, and the significance of relational identities. We then discuss how LLMs can be used as powerful tools to augment this work, from assisting with data collection and coding to supporting theory-building and abductive reasoning. We also address the limitations and ethical challenges of using LLMs in this context, including issues of bias, hallucination, and the need for reflexivity. We conclude with a series of research designs and practical recommendations for researchers who want to integrate LLMs into their work in a thoughtful and responsible way.
arXiv:2607.14057v1 Announce Type: new
Abstract: Expanding on the recent development of adaptive local kernel methods for approximating the action of linear operators, a local estimate of the error and an adaptive procedure for approximating solutions to the Poisson equation is developed. The error estimate is used in the midst of the adaptive procedure to determine locations in the solution domain where decreasing the spacing between nodes can decrease the error in the solution. The approach described here is essentially "meshless", with only local Delaunay triangulations leveraged as a convenience to provide information on the local geometry of the node set. Once this information is utilized, it is discarded. The experiments performed show close agreement between the error estimate and actual absolute forward error in the approximate solution along with a comparison to existing indicators for refinement. The combination of the adaptive procedure and error estimate provides an automated technique to resolve localized features of a solution without the, often intractable, expense of uniform refinement across the entire solution domain.