arXiv:2606.13984v2 Announce Type: replace-cross Abstract: Classification and Regression Trees (CART) constitute one of the most influential paradigms in statistical learning. Although a variety of impurity measures have been proposed for different statistical models, these criteria are typically introduced on a case-by-case basis and analyzed separately. In this paper, we study CART through the lens of Bregman divergences. This perspective places the classical least-squares criterion, Poisson deviance, Kullback-Leibler-type losses, and other impurity measures associated with exponential-family models within a common framework. As a result, key ingredients of the CART methodology -- including node representatives, impurity measures, and split selection rules -- can be expressed and analyzed through general properties of convex functions rather than through separate model-specific constructions. Beyond the algorithmic formulation, we investigate theoretical properties of Bregman-based CART procedures. In particular, we analyze how geometric properties of the generating convex function influence impurity reductions and stability of recursive partitions. We also establish consistency results within the proposed framework, providing a unified theoretical treatment for a broad family of CART type procedures. Our results provide a geometric interpretation of impurity-based tree construction and show that many classical CART impurity criteria admit a common interpretation within a Bregman framework.
Science Journals
arXiv:2606.24187v2 Announce Type: replace Abstract: Long video understanding remains a daunting challenge for Multimodal Large Language Models (MLLMs) due to the excessive computation and memory footprint. Thus, keyframe selection is often adopted to mitigate this shortcoming, which however still suffers from low flexibility and high noise due to its hard sampling principle. In this paper, we define video frame selection as a problem of Quasi-Gaussian Sampling, and propose an adaptive and training-free approach termed AdaQ. Inspired by the 3-$\sigma$ rule of Gaussian distribution, the objective of AdaQ is to achieve the optimal 3-$\sigma$ interval for different examples, i.e., a smaller 3-$\sigma$ interval for the local query and a larger one for the global query, thereby facilitating robust and adaptive frame sampling. To validate AdaQ, we apply it to four MLLMs with three embedding models. The extensive experimental results not only show its obvious performance gains over the default MLLMs and the SOTA keyframe selection methods, e.g., helping Qwen3-VL-8B outperform GPT4o by 15.8% on average by using only 64 frames, but also confirm its superior robustness and high efficiency for long-video understanding, e.g., only 1 hyper-parameter needs to be set.
arXiv:2606.24326v2 Announce Type: replace-cross Abstract: Given a graph class $\mathcal{G}$, the limiting density of $\mathcal{G}$ is defined as $\delta(\mathcal{G})=\lim_{n\to\infty} \mathsf{ex}(\mathcal{G},n)/n$ where $\mathsf{ex}(\mathcal{G},n)$ is the maximum number of edges of a graph in $\mathcal{G}$ on $n$ vertices. The limiting density $\delta(\mathcal{G})$ is known to be a rational number when $\mathcal{G}$ is a minor-closed graph class. For every $\delta\in[0,\frac{3}{2})$, we prove that the set of $\subseteq$-minimal minor-closed graph classes with densities $>\delta$ is finite and we identify it completely. A consequence of our results is an algorithm that, given a finite set of graphs $\mathcal{Z}$, of total size $n$, either outputs the value of $\delta(\mathsf{excl}(\mathcal{Z}))$ or reports that $\delta(\mathsf{excl}(\mathcal{Z}))\geq \frac{3}{2}$, where $\mathsf{excl}(\mathcal{Z})$ is the class of graphs excluding the graphs in $\mathcal{Z}$ as minors. The algorithm runs in $2^{\mathsf{poly}(n)}$ time.
arXiv:2606.24932v1 Announce Type: cross Abstract: Recent advances in quantum computing and machine learning have motivated the development of quantum models for sequential data processing. In this paper, we propose a Recursive Quantum Long Short-Term Memory model, or Recursive QLSTM, which extends QLSTM through metacore-based recursive constructions. We numerically test the model under different input sequence lengths, metacore designs, and recursive rules, and identify the best-performing architecture among these variants. For this selected model, we further provide theoretical arguments explaining why its recursive structure improves temporal information propagation and enhances learning performance. Our results suggest that Recursive QLSTM offers a flexible and effective framework for quantum recurrent learning over input time series of various lengths.
arXiv:2606.25601v1 Announce Type: cross Abstract: Hyperparameter selection is a critical step in the deployment of modern artificial intelligence systems, given the need to tune degrees of freedom such as inference-time parameters, implementation-level settings, and thresholds driving decision rules. Despite its practical importance, hyperparameter selection is typically performed using best-effort empirical methods such as grid search or Bayesian optimization, which provide no formal statistical guarantees on reliability or safety. This monograph presents a unified statistical framework for reliable hyperparameter selection, centered on the learn-then-test (LTT) paradigm, which formulates the problem as multiple hypothesis testing over a candidate set of hyperparameters. The framework enables the selection of hyperparameters that provably satisfy application-specific reliability requirements -- such as bounds on average risk, quantile risk, or information-theoretic constraints -- with explicit, finite-sample control of error probabilities. The supporting statistical machinery, namely p-values, e-values, and concentration inequalities, is developed from first principles in a dedicated appendix.
arXiv:2603.03394v2 Announce Type: replace Abstract: Collaborative AI experimentation in industry-academia requires environments that support rapid trials while maintaining controlled access, organisational isolation, and traceable workflows. Although interest in AI sandboxes is increasing, practical guidance on designing and building governance-aware experimentation platforms remains limited. This work designs and operationalizes a governance-aware, multi-tenant AI sandbox that supports structured experimentation and produces reusable evaluation evidence across stakeholders. The sandbox was developed in an industry-academia ecosystem using iteratively validated requirements gathered from industrial partners. The solution adopts a layered reference architecture that separates a multi-tenant presentation layer from a backend control plane and isolates execution and data management concerns into dedicated layers. The sandbox supports governed onboarding, project-based collaboration, controlled access to AI services, and traceable experimentation through approval workflows and audit logging. By structuring experiment context and governance decisions as persistent records, the sandbox enables evaluation evidence to be reused and compared across projects and stakeholders. The development experience yields lessons learned and practical considerations that inform deployment and future evolution of governance-aware sandbox platforms.
arXiv:2603.09359v2 Announce Type: replace Abstract: Physics-informed neural networks (PINNs) have shown promise in addressing the ill-posed deconvolution problem in computed tomography perfusion (CTP) imaging for acute ischemic stroke assessment. However, existing PINN-based approaches remain deterministic and do not quantify uncertainty associated with violations of physics constraints, limiting reliability assessment. We propose Evidential Perfusion Physics-Informed Neural Networks (EPPINN), a framework that integrates evidential deep learning with physics-informed modeling to enable uncertainty-aware perfusion parameter estimation. EPPINN models arterial input, tissue concentration, and perfusion parameters using coordinate-based networks, and places a Normal--Inverse--Gamma distribution over the physics residual to characterize voxel-wise aleatoric and epistemic uncertainty in physics consistency without requiring Bayesian sampling or ensemble inference. The framework further incorporates physiologically constrained parameterization and stabilization strategies to promote robust per-case optimization. We evaluate EPPINN on digital phantom data, the ISLES 2018 benchmark, and a clinical cohort. On the evaluated datasets, EPPINN achieves lower normalized mean absolute error than classical deconvolution and PINN baselines, particularly under sparse temporal sampling and low signal-to-noise conditions, while providing conservative uncertainty estimates with high empirical coverage. On clinical data, EPPINN attains the highest voxel-level and case-level infarct-core detection sensitivity. These results suggest that evidential physics-informed learning can improve both accuracy and reliability of CTP analysis for time-critical stroke assessment. Source code is available at https://github.com/jhlee0619/EPPINN.
arXiv:2606.23001v2 Announce Type: replace Abstract: On-device LLM inference is increasingly attractive for privacy-preserving, reliable, and cost-effective deployment, yet its energy and thermal costs remain a critical bottleneck. Existing systems primarily optimize for decoding speed, implicitly assuming that faster execution is always preferable. We show instead that on-device LLM inference often has exploitable configuration slack: modestly lowering NPU and memory frequencies preserves quality of experience (QoE) while substantially improving energy efficiency and reducing heat. Realizing this opportunity in production is challenging. The most energy-efficient NPU/DDR setting varies with the model, inference engine, platform, and runtime conditions, with no stable ranking across configurations. Commercial devices further lack component-level power sensing, and shell temperature evolves with request arrivals, response lengths, and thermal history. To address these challenges, we propose EnerInfer, the first on-device LLM inference framework that jointly manages energy efficiency, throughput, and thermal comfort for LLM workloads. EnerInfer replaces per-model profiling and sensor-heavy control with disaggregated, model-structure-aware prediction and ranking-driven online feedback. It predicts throughput and power for unseen LLMs across NPU/DDR frequency settings, selects QoE-satisfying efficient configurations under runtime interference, and uses lightweight limited-horizon thermal prediction to dynamically switch between energy-optimized and thermally constrained inference. Evaluations on real-world LLMs show that EnerInfer improves energy efficiency by up to 65%, 12%, and 24% on phones, a laptop, and a development board, respectively, without QoE violation.
arXiv:2510.01299v3 Announce Type: replace-cross Abstract: We present domain-adapted large language model agents designed to support Cherenkov Telescope Array operation and data analysis. The agents combine contextual knowledge with automated validation and iterative correction to produce more reliable outputs. This approach reduces manual effort, improves consistency, and helps accelerate operational and scientific workflows. The results demonstrate the potential of agentic systems as practical assistants in specialized research environments.
arXiv:2606.23947v1 Announce Type: new Abstract: The Wisconsin HTS Axisymmetric Mirror (WHAM) is a compact high-field magnetic mirror. In such magnetic mirrors, cross-field transport is dominated by the flute instability (Endrizzi et al., 2023). To investigate density fluctuations associated with the flute instability, we designed a Doppler backscattering (DBS) diagnostic for WHAM, to be installed at the midplane port window. The diagnostic uses a two-channel tunable Ka-band (26.5--40 GHz) source and X-mode polarization. The azimuthal launch angle is set mechanically by rotating the external quasioptical assembly. As such, the system is reconfigurable during dedicated setup periods. Using the \textit{Scotty} beam-tracing code (Hall-Chen et al., 2022), we show that the proposed DBS system can measure density fluctuations with perpendicular wavenumbers $1 \leq k_\perp \leq 3~\mathrm{cm}^{-1}$ over radial locations $0.7 \leq \rho \leq 0.9$, where $\rho$ is the normalized radial coordinate. This is achieved with probe frequencies between 28 and 38.5 GHz, an elevation launch angle of $0^\circ$, and azimuthal launch angles in the range $1^\circ$--$3^\circ$. The selected configurations have low mismatch angle at cutoff, $|\theta_{m,c}|<1^\circ$. The quasioptical system uses a Ka-band horn and a biconvex ultra-high molecular weight polyethylene lens, and satisfies the port-access constraints in WHAM. The planned microwave system has a monostatic, homodyne architecture based on two phase-coupled Ka-band microwave channels. These two channels will be for the transmitted signal and coherent local oscillator (LO) for IQ downconversion, respectively. As the two phase-coupled channels can be independently tuned or swept with a controlled frequency offset, the same microwave chain can also support profile-reflectometry measurements using cutoff-delay information.
arXiv:2606.19416v2 Announce Type: replace Abstract: Loan origination is the process by which a lender creates a new loan, from application and underwriting through approval and funding. This process serves a critical role in evaluating the eligibility and level of risk posed by an applicant. Recently, firms have begun using mortgage loan agents to augment human loan officers, despite a lack of any public benchmark. To fill this gap, we present MortarBench, a loan origination agent benchmark. MortarBench uses a financial data synthesis and mutation pipeline to generate examples with broad edge case coverage that match real-world distributions and questions. We find that state-of-the-art large language models (LLMs) perform poorly, with closed-source models achieving at most 77.1\% exact match accuracy. We also discover systematic biases in LLM perception of foreignness related to non-English names. Noting these weaknesses, we introduce CRIT, a confidence calibration framework. Our method increases accuracy to 80.5\% while improving risk management steering and reducing bias.
arXiv:2606.23759v1 Announce Type: new Abstract: Verilog debugging remains one of the most time-consuming stages in digital circuit design. Recent advances in Large Language Models (LLMs) have enabled automated debugging; however, most existing approaches rely solely on test outputs and compiler feedback in an end-to-end manner, limiting their effectiveness on complex bugs. A key challenge is that the root cause of an error may be far removed from its observable outputs, making it difficult for LLMs to trace long dependency chains in code. This challenge is further exacerbated in large codebases, where long context lengths hinder efficient reasoning. To address these limitations, we propose VeriPilot, an LLM-powered debugging framework that leverages golden reference models to enable fine-grained bug localization and repair. VeriPilot goes beyond output-level comparison by aligning internal variable semantics between the Verilog design and its corresponding golden model through LLM-based analysis. It then performs step-by-step signal tracing using Control-Data-Flow Graphs (CDFGs) derived from static analysis, identifying a minimal set of suspicious code regions along with their correct counterparts from the golden model. These structured insights are subsequently provided to the LLM to guide reasoning and automated code repair. Experimental results on the Comprehensive Verilog Design Problems (CVDP) benchmark from NVIDIA demonstrate that VeriPilot improves the repair success rate of GPT-4o from 54.3\% to 85.71\%, significantly enhancing both bug localization accuracy and repair effectiveness for complex Verilog designs. The source code and benchmark are publicly available at Github https://github.com/YihanWn/VeriPilot.git.
arXiv:2606.23830v1 Announce Type: new Abstract: Molecular surfaces encode the geometric and physicochemical patterns that determine antibody-antigen recognition, central to epitope prediction. However, existing methods rely on sequences or backbone structures and struggle to capture discontinuous, surface-driven epitopes. This study presents SurfBind, a surface-centric learning framework for epitope prediction that operates directly on molecular surface representations. SurfBind integrates geometric and physicochemical cues through a Transformer-based architecture with patch-level surface modeling, binder-aware cross-attention, and a hierarchical coarse-to-fine prediction paradigm. Experiments on challenging epitope identification benchmarks, including SAbDab and DB5.5, demonstrate that SurfBind achieves state-of-the-art performance and strong generalization across unseen antibodies and conformational states, highlighting the value of interaction-aware surface modeling for understanding the crucial mechanisms of protein-protein interactions.
arXiv:2606.23832v1 Announce Type: new Abstract: The use of dedicated corridors for Advanced Air Mobility (AAM) traffic is one of the most commonly proposed pathways to integrating them into existing airspace operations. Most prior research has focused on the design of networks of AAM corridors and conflict resolution for aircraft within corridors. It is also generally believed that while attractive from an implementation perspective, corridor-based operations may be inefficient, especially in the absence of centralized traffic management. In this paper, we show that contrary to this belief, it is possible for autonomous aircraft to learn to self-organize into corridor flows in decentralized settings. We illustrate our approach using scenarios in which fixed-wing aircraft need to safely and efficiently traverse (1) a single corridor with metering after the exit, (2) a sequence of two consecutive corridors, and (3) a corridor that splits into two. We find that in decentralized settings with only local information, the aircraft are able to conform to the corridor boundaries more than 94% of the time and reach their goal in a relatively efficient manner. Furthermore, tactical interventions to handle violations of the separation minimum are needed only infrequently in low- and medium-density settings. However, such tactical interventions become more frequently necessary only when traffic density is high.
arXiv:2606.23840v1 Announce Type: new Abstract: Explainability is often framed as a property of an AI model, with explanations extracted from its internals and shown to users. In this argument paper, we instead provide an embodied account of explainability based on Dourish and enactivist cognition: understanding is created in use as people act on affordances in shared practice. Using demonstrations and conceptual analysis, we reveal ontological obstacles when "looking inside" large language models: surrogates import external abstractions that can be mistaken for the model's, and focusing on internal reasoning misses that explainers participate in their own understanding. We discuss these obstacles in XAI practice, arguing that many explanations are misnamed, which skews their purpose and can increase overreliance. Finally, we highlight how embodied explanations reorganize sense-making by making what matters publicly available for action, and argue that explainability claims should be reserved for designs that provide affordances to probe, coordinate, and repair behaviour in situated practice.
arXiv:2606.23860v1 Announce Type: new Abstract: Who sets the rules for artificial intelligence, and on what terms, has become a defining question of global governance. For several years that contest ran through principles and ethics codes; it now runs through institutions. China's proposed World Artificial Intelligence Cooperation Organization (WAICO) is the most consequential recent entrant and the least examined. We place WAICO within the emerging regime complex for AI and argue that its importance lies not in any single commitment but in the position it is designed to hold. Coding a cross-section of fifteen international AI governance instruments and institutions on how they admit members, how they are organized, and what they prioritize, we find that WAICO's proposed design joins three features that no constituted multilateral body currently combines: membership open to any sovereign state, no values or regime-type test for entry, and an agenda built around development and the global capability divide. The incumbent Western-led bodies gate membership by shared values and concentrate on rights and safety; the universal United Nations bodies are open but anchored in human rights; a development-first agenda is otherwise carried by the regional strategies of the Global South. Among constituted institutions, the only occupant of WAICO's intended position is China's own 2023 precursor initiative. We read this as the formation of a second, still-proposed pole in global AI governance, organized around sovereignty and development rather than rights and safety, and argue that WAICO would be the first standing organization built to anchor it. We report the full coding, state testable expectations against which the claim can be judged as the organization takes shape, and release the dataset for replication.
arXiv:2606.23877v1 Announce Type: new Abstract: Jupyter Notebooks are an increasingly popular coding environment used across many domains, especially in Python-based data science and scientific computing. Originally used for prototyping and interactive exploration, notebooks are increasingly used to develop more complex programs, leading to a rapid rise in buggy notebooks on platforms like GitHub. To address this trend, we present JupOtter, a bug detection system designed specifically for Jupyter Notebooks. JupOtter features three novel contributions: (1) a notebook-specific tokenization strategy that preserves cell structure, (2) a cell-level bug prediction technique, and (3) a new labeled dataset, OtterDataset, containing over 21,000 notebooks annotated for fine-grained cell-level bug detection. JupOtter achieves cell-level bug detection F1 scores that surpass static analyzers and large language models in two out of three evaluation datasets.
arXiv:2606.23829v1 Announce Type: new Abstract: The representation of cloud microphysics and its nonlinear character and scale-dependence is a remaining source of uncertainty in Earth system models (ESMs). Here, we develop and couple online a machine learning (ML)-based cloud microphysics parameterization with the Icosahedral non-hydrostatic modeling framework (ICON). The primary challenge is achieving numerically stable, long-term online coupling when transitioning from training with km-scale data to application in coarse-scale simulations, where the coupled system encounters atmospheric states and feedbacks not seen during training. The training data is obtained from a global convection-permitting ICON simulation at 5 km resolution. The ML microphysics scheme uses a two-stage design: a classifier to identify active grid cells and a regressor to predict cloud microphysical tendencies. Physical constraints such as enforcing mass positivity and overshoot prevention prove essential for numerical stability in the coupled system. We demonstrate that achieving stable online coupling requires enforcing physical constraints and careful dataset curation, and that strong offline performance alone is insufficient. The coupled model maintains numerical stability over decade-long simulations with a performance in reproducing the observed climate comparable to the classical graupel scheme. The ML-based scheme eliminates two microphysics-specific tuning parameters of the classical graupel scheme, though systematic improvements in long-term mean-state biases are not yet realized. This study demonstrates that stable, decade-long climate simulations with an ML-based cloud microphysics scheme trained on convection-permitting data are feasible, providing a foundation for future hybrid ESMs.
arXiv:2606.23920v1 Announce Type: new Abstract: The task of compositional generation involves using a conditional generative model, trained only on a subset of the possible conditions, to produce samples from compositionally-defined target distributions such as a geometric combination of the source distributions. In this work, we argue that this task is often infeasible for vanilla conditional diffusion models: we conjecture that no inference-time technique can efficiently produce samples from the target distribution in certain well-motivated settings. This idea is supported by theory-guided generalization arguments and carefully-designed experiments on both synthetic and realistic data. In particular, while recent methods such as Feynman-Kac correction reduce inference-time approximation error, our results show that score estimation error has a more catastrophic effect on performance when the target distribution is out-of-distribution with respect to the sources, highlighting the need for a different approach to this task.
arXiv:2606.23825v1 Announce Type: new Abstract: Efficient small object detection is bottlenecked by the inherent feature scarcity of tiny targets, which is further aggravated by operations of spatial-domain detectors that indiscriminately discard critical high-frequency details. Recovering these fragile cues within the spatial domain is notoriously difficult, as it often requires computationally expensive architectural upscaling that inadvertently amplifies background noise. To bridge this gap, we propose a paradigm \textbf{shift from spatial to spectral} feature processing, introducing a holistic solution with the following novelty: (1) A versatile \textbf{Frequency-Guided Feature Representation framework} that generalizes across diverse detector architectures (both CNN and Transformer-based), offering a robust alternative to spatial-only feature extraction; (2) The unified \textbf{Decompose--Enhance--Reconstruct (DER)} operator, instantiated via three \textbf{lightweight, plug-and-play} modules -- Wavelet-Difference Gate (WDG), Log-Gabor Enhancer (LGE), and Frequency-Driven Head (FDHead) -- to systematically inject frequency-aware modulation into the backbone, neck, and head. This mechanism decouples feature modeling from resolution reduction, capturing discriminative high-frequency components to enable accurate localization with significantly reduced parameter redundancy; (3) Extensive validation on multi-domain benchmarks (VisDrone2019, UAVDT, TinyPerson, DOTAv1) demonstrating consistent gains. Notably, our proposed \textbf{DERNet} series outperforms YOLOv11 models under the same scale while requiring \textbf{only 1/6 of the parameters}, backed by rigorous spectral diagnostics and error decomposition analysis.
arXiv:2606.23839v1 Announce Type: new Abstract: We present a systematic study of the diagonal Born-Oppenheimer correction (DBOC) for atoms and molecules embedded in optical cavities and interacting with a quantized electromagnetic field. By explicitly evaluating the nuclear kinetic energy operator, we analyze cavity-induced modifications of DBOC within a quantum electrodynamics configuration-interaction (QED-CI) framework built on quantum electrodynamics Hartree-Fock (QED-HF) and strong-coupling quantum electrodynamics Hartree-Fock (SC-QED-HF) reference states. The analysis covers a diverse set of atomic and molecular systems, including He, H-, Be, H2, LiH, HF, ammonia (NH3), and formaldehyde (CH2O). We show that the presence of the cavity leads to shifts in molecular dissociation energies on the order of a few inverse centimeters. For several atomic systems, the inclusion of the DBOC yields a pronounced effect, with the correction magnitude reaching the experimental resolution. These findings reveal finite nuclear mass effects as an essential component of nuclear dynamics in cavity QED and suggest their relevance for precision analysis in strongly coupled light-matter systems.
arXiv:2606.23917v1 Announce Type: new Abstract: Advances in generative AI have made image falsification highly realistic, demanding trustworthy authentication systems. Existing forensic detectors can target certain forgery types but lack interpretability, while vision-language models (VLMs) provide explanations but cannot exploit forensic traces for reliable detection. We propose Forensic Knowledge Graphs (FKGs), a unified framework that integrates forensic evidence extraction, structured reasoning, and human-interpretable explanation. Our FKG structure encodes forensic traces along with their causal dependencies and links to scene content. To generate accurate FKGs, we introduce a novel forensic authentication network and an Iterative Context Refinement strategy that guides VLMs to produce faithful, grounded explanations. We also present FKG-50K, a dataset of 50,000 realistic forgeries with ground-truth FKGs. Experiments demonstrate that FKG outperforms both forensic detectors and VLMs in detection, forgery identification and localization, and forensic justification.
arXiv:2606.24066v1 Announce Type: new Abstract: Speaker recognition has advanced rapidly with large-scale training datasets, yet Vietnamese remains under-resourced, with existing corpora limited in scale and acoustic diversity. Most large-scale datasets rely on facial cues to link speech with speaker identities, restricting data collection to recordings where speakers appear on camera. We propose a face-independent dataset construction pipeline and introduce VieSpeaker, a large-scale Vietnamese speaker recognition dataset. Our approach leverages textual metadata and large language model reasoning to infer speaker identities from transcripts and contextual information. VieSpeaker contains approximately 902 hours of speech from 4,715 speakers. Experiments show that models trained on VieSpeaker achieve improved robustness and generalization compared to existing Vietnamese datasets. This work demonstrates the feasibility of face-independent dataset construction and provides a new direction for building large-scale speech resources.
arXiv:2606.24105v1 Announce Type: new Abstract: As personal data privacy becomes increasingly critical in Internet of Things (IoT) environments, secure DNS protocols such as DNS over HTTPS (DoH) and DNS over TLS (DoT) have been widely adopted to protect device communications. However, without effective obfuscation, these protocols remain vulnerable to Website Fingerprinting (WF) attacks that can reveal user activity. With the ongoing deployment of DNS over HTTP/3 (DoH/3) in modern networked systems, padding strategies have been increasingly applied in practice. It is therefore essential to investigate whether DoH/3 can effectively mitigate WF attacks in realistic IoT and edge-network scenarios. To address this, we first collect and publicly release the first real-world benchmark dataset of DoH/3 traffic, generated from domain resolution processes in practical network environments. We further propose DoHFuse, a dual-branch learning framework that integrates inter-arrival time sequences and refined statistical features through an improved DMAG-LSTM, specifically designed to capture burst-aligned temporal patterns. Experimental results show that DoHFuse achieves an accuracy of 88.05% (precision 88.56, recall 87.96, F1 87.83) in a closed-world setting of 449 classes, and an AUPRC of 0.975 with an F1 score of 0.951 (precision 0.906, recall 1.0) in open-world detection. These findings demonstrate that DoH/3 traffic remains susceptible to WF attacks, suggesting that commonly deployed padding mechanisms alone are insufficient to ensure privacy protection in IoT-scale encrypted DNS communications.
arXiv:2606.24014v1 Announce Type: new Abstract: As AI systems are deployed across increasingly diverse and high-stakes settings, model alignment must generalize beyond the tasks and domains seen during training. This is especially important for reinforcement learning (RL), which can introduce unexpected misalignment through reward hacking, deception, or other unintended strategies. We study whether RL on beneficial behavior, instantiated in realistic domains, can produce broad and persistent alignment generalization beyond the training distribution. We construct a dataset of realistic situations designed to measure and train beneficial traits, such as truthfulness, fairness, risk awareness, and corrigibility, spanning varied domains, including health, science, and education. We then train models with RL on this dataset and evaluate them on more than 50 independent benchmarks of alignment and beneficial behavior. Compared to a compute-matched baseline, beneficial trait RL improves performance on over 80% of these out-of-distribution benchmarks. We observe substantial out-of-distribution alignment transfer: a beneficial-behavior RL intervention entirely limited to one domain, health, produces broad improvements on non-health alignment evaluations, including reduced reward hacking, deception, and general misalignment. Finally, we study alignment persistence: whether behavior remains robustly aligned under attempts to steer models towards misalignment. Models trained with beneficial trait RL show improved persistence, including greater resistance to adversarial prompting and harmful finetuning; further work is required to isolate the sources of these effects. These results suggest that RL to reinforce beneficial behavior in realistic domains can produce models that are more robustly aligned with human flourishing.