Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (quant-ph) 2026-06-19

Electrical Noise Produced by Micron-Sized Particles above a Surface Paul Trap

arXiv:2606.19585v1 Announce Type: new Abstract: Electric field noise produced by the surface of ion trap electrodes reduces the fidelity of quantum computing operations. Despite decades of investigation its microscopic origins remain unclear. Here, we measure electric field noise at trapping locations along the symmetry axis of a linear surface Paul trap. We find that noise levels vary by three orders-of-magnitude in one 600$\,\mu$m section of the trap. Optical and scanning electron microscope images show micron-sized particles close to the trapping locations with the highest noise levels. We find that modeling the particles as a lossy dielectric with a effective loss tangent $\tan\theta=0.33(0.06)$ describes the magnitude of the noise, as well as its spatial and frequency dependence. Our observations may explain the large variation of reported noise levels in literature.

02.
arXiv (CS.CL) 2026-06-11

External Experience Serving in Production LLM Systems: A Deployment-Oriented Study of Quality-Cost Trade-offs

Production LLM systems accumulate reusable operational experience, but the practical deployment issue is not merely whether such experience can help. It is how different serving strategies trade off quality against online cost under realistic constraints. Injecting external experience can improve task quality, yet it also increases prompt burden, latency, and serving pressure. We study external experience serving as a deployment-oriented quality-cost trade-off problem. We evaluate this question in a real production moderation setting, with tool-use and GPQA as supporting contrast tasks that expose different output-cost regimes. We compare no-experience baselines, random experience controls, global prompt injection, and retrieval-based selective injection, and analyze both task quality and serving cost. The results show that, once experience becomes case-dependent, selective retrieval provides a stronger operating point than unconditional global injection. They further show that retrieval quality matters more than simply increasing Top-$K$, and that the same serving policy can exhibit substantially different cost-benefit profiles across short-output and decode-heavy regimes. These findings suggest that external experience is best treated as a selective, cost-aware serving decision rather than as a universal add-on. Overall, in the settings studied here, external experience pays off only when both the serving interface and the task-specific cost structure make its quality gains worth the online cost.

03.
arXiv (CS.CL) 2026-06-25

Measuring Research Difficulty of Academic Papers: A Case Study in Natural Language Processing

With the rapid growth of the number of academic papers, systematically evaluating the difficulty of research and its relationship to academic impact offers important significance for research topic selection and resource allocation. However, current studies lack quantitative assessments of research difficulty and its correlation with academic impact. This paper proposes a comprehensive evaluation system for research difficulty, incorporating factors such as academic collaboration, content, and references. Taking the field of Natural Language Processing (NLP) as a case study, we extract both internal and external features from academic papers, compute multiple research difficulty indicators. We assign their weights using the entropy weight method and perform a weighted sum to obtain the research difficulty score of academic papers. This paper uses the citation frequency of academic papers to measure academic impact. To validate our approach, NLP experts assessed the difficulty of a sample of papers, and correlation analyses confirmed the reliability of our measurement. Empirical results reveal that in NLP, factors such as the number of pages, reference count, and participation of high-level institutions are significantly associated with academic impact. Moreover, we identify an inverted U-shaped relationship between research difficulty and academic impact. It suggests that moderately difficult research tends to achieve greater academic impact.

04.
arXiv (CS.LG) 2026-06-19

Influence-Guided Concolic Testing of Transformer Robustness

arXiv:2509.23806v2 Announce Type: replace-cross Abstract: Concolic testing for neural networks alternates concrete execution with constraint solving to search for inputs that flip model decisions. We present a concolic tester for Transformer classifiers that uses SHAP estimates to rank pending path predicates by their impact on the current prediction. To support self-attention with multiple heads in execution backed by SMT solving, we implement attention semantics in pure Python that are compatible with the solver and make the softmax boundary explicit by concretizing exponentiation arguments. We evaluate our method on CIFAR-10 across three compact Transformer classifiers, ResNet18, and VGG16 under a one-pixel budget and a 900s horizon. Across the 500 model–input pairs in this matched comparison, our method achieves 60% success, compared with 15% for a differential evolution baseline that treats the model as a black box. In the primary two-layer Transformer branch-ordering study, SHAP-based predicate prioritization raises success from 56% to 60% and reduces median attack time by 51%. These results show that influence-guided path exploration can make concolic testing a practical way to find adversarial examples in Transformer models.

05.
arXiv (CS.CL) 2026-06-16

Compositional Reasoning Depth Predicts Clinical AI Failure: Empirical Evidence Consistent with Transformer Compositionality Limits in Electronic Health Record Question Answering

作者:

Aggregate accuracy benchmarks conceal a systematic structure in how large language models fail at electronic health record (EHR) question answering: questions requiring more inferential steps produce disproportionately more errors. Motivated by theoretical results on transformer compositionality limits, we introduce a pre-specified hop-count taxonomy – the number of distinct reasoning steps required to answer a clinical question from an EHR – as a principled predictor of model failure. We annotate 313 clinician-generated MedAlign EHR question-answer pairs across four hop levels and evaluate 301 questions in a within-model ablation (claude-sonnet-4-6, zero-shot vs. extended thinking) and cross-architecture replications (gpt-4o and gpt-5.4-2026-03-05, zero-shot). All three models, spanning two providers and two OpenAI generations (GPT-4 and GPT-5), show monotone accuracy decline with hop count: Claude Sonnet zero-shot falls from 30.6% (hop=1) to 17.6% (hop=4) (Cochran-Armitage z=-2.30, p=0.011; OR per hop 0.72, 95% CI [0.56,0.92], p=0.008); GPT-4o replicates this (37.8% to 14.7%; OR 0.58 [0.45,0.75], p

06.
arXiv (CS.AI) 2026-06-19

Evaluation of EEG Foundation Models for Event-Based Burst-Suppression Detection in ICU

arXiv:2606.20074v1 Announce Type: cross Abstract: Burst suppression (BS) is a clinically relevant electroencephalographic (EEG) pattern used to monitor sedation depth and brain activity in critically ill patients, particularly during induced coma in Intensive Care Units (ICUs). Automatic burst detection remains challenging because BS patterns vary substantially between patients and annotated datasets are scarce. Recently, EEG Foundation Models (FMs) have shown promise across several downstream EEG applications, but their usefulness for BS detection remains unexplored. We present the first study to evaluate EEG FMs for burst detection in reduced-montage ICU EEG without patient-specific calibration. We compare REVE-base, LUNA-large and LuMamba-Tiny with an adaptive thresholding baseline and a task-specific EEGNet baseline. Additionally, we complement conventional EEG window-based classification with event-based burst detection evaluation. This helps assessing clinically whether burst episodes are correctly detected, reducing the impact of expected annotation variability. The best model, REVE-base, achieved the highest event-based F1-score ($0.868 \pm 0.167$) and reduced burst-per-minute error by 52.1% and 36.2% compared to EEGNet and adaptive thresholding respectively, supporting FMs for scalable EEG monitoring in ICU. Ablation experiments showed that full fine-tuning was the most effective adaptation strategy with respect to frozen-backbone training, two-step fine-tuning, and LoRA-based adaptation, improving event-based F1-score over frozen-backbone training by up to $+0.102$ for LUNA-large. With reduced labeled datasets, pretrained REVE-base outperformed random initialization by $+0.723$ event-based F1 points at 25% of the cohort, demonstrating the benefit of pretraining FM representations when adapted to burst detection with limited labeled data.

07.
arXiv (CS.AI) 2026-06-24

THEIA: Learning Complete Kleene Three-Valued Logic in a Pure-Neural Modular Architecture

arXiv:2604.11284v5 Announce Type: replace-cross Abstract: We present THEIA, a 2.75M-parameter modular neural architecture that learns the complete Kleene three-valued logic (K3) truth table from task data without external symbolic inference or hand-encoded K3 gate primitives. Across 5 seeds it passes all 39 K3 rules at >99% per-rule accuracy. K3 learnability is not the central finding: Transformer baselines also pass all 39 rules, and flat MLPs match THEIA on Phase-1 accuracy within 0.04pp. The contributions are two properties of the learned system. (1) Uncertainty-verdict asymmetric propagation. THEIA preserves Has-Unknown at every upstream boundary (80.0/91.1/90.8/99.7% across Arith/Order/Set/Logic vs. ~52% majority) while final-verdict decodability stays at or below a 73.4% U-vs-non-U oracle reference under linear and nonlinear probes. Activation patching on non-absorbent T->U cases flips 4,898/4,898 OR and 4,719/4,719 AND pairs across 5 seeds, ruling out residual shortcuts. (2) Reliability spectrum under discretized end-to-end training, on tasks decomposable along the engine boundaries. A mod-3 sequential composition task generalizes from 5- to 500-step evaluation at 99.96+-0.04% (5 seeds). Under identical Gumbel-softmax training, flat MLPs collapse to chance by 50 steps; a 2x2 ResMLP grid reaches >=99% on only 3/20 (config, seed) trials; a pre-LN Transformer reaches 99.24+-0.34%. Straight-through discretization prevents 0.999^500 compounding; the architectural separator is sustaining Phase-1 accuracy under Phase-3 training, where flat MLPs fail. Auxiliary: under per-architecture development defaults (not optimizer-controlled), THEIA reaches 12/12 Kleene coverage 6.5x faster than a parameter-comparable 8L Transformer; this narrows to ~3.6x with Transformer-standard tuning and 4.93x with the same recipe on both. Ratios are config-specific, not asymptotic.

08.
arXiv (math.PR) 2026-06-17

The Erdős-Hajnal High-Girth Subgraph Conjecture Holds in the Polynomial Chromatic-Sparsity Regime

作者:

arXiv:2606.17901v1 Announce Type: cross Abstract: For a graph $G$ put $h_r(G)=\max{\chi(H):H\subseteq G,\operatorname{girth}(H)\ge r}.$ Erdős and Hajnal asked whether $h_r(G)\to\infty$ as $\chi(G)\to\infty$, for every fixed $r\ge4$. We prove this in every fixed polynomial edge-density regime: for all $r\ge4$, $k\ge2$, $P,C>0$, there is $M=M_{r,k}(P,C)$ such that $\chi(G)\ge M,\ e(G)\le C\chi(G)^P\Longrightarrow h_r(G)\ge k.$ Quantitatively, after replacing $P$ by $P\vee2$ and $C$ by $C\vee2$, $M_{r,k}(P,C)\le \exp!\left(O_{r,k}\bigl((P+2+\log(C\vee2))^2\bigr)\right),$ and consequently the same conclusion holds throughout the quasi-polynomial range $e(G)\le \exp\bigl(C_0(\log\chi(G))^a\bigr),\ 1 < a < 3/2,$ for all sufficiently large $\chi(G)$. In each fixed polynomial-density regime we also obtain $f_{P,C}(k,r)\le k^{O_{r,P,C}(1)}.$ The proof combines a chromatic-defect random extraction lemma, compact and near-quadratic sparse-core bases, and a peeling/thinning bootstrap increasing the admissible edge exponent by $1/(r-1)$. We also prove structural saturation results for possible counterexamples, including Moore-strength exact-cycle packings and quadratic saturation in projected colour-pair space. Finally, writing $h_r^{\mathrm f}(G)=\max{\chi_{\mathrm f}(H):H\subseteq G,\operatorname{girth}(H)\ge r},$ we develop a fractional random-extraction framework based on Mohar-Wu preservation. We prove sufficient cheap-cycle-killing criteria and verify them for several structured families, including clique-organised families, line graphs of incidence graphs of equal-order generalized quadrangles and generalized hexagons, and the Bohman-Keevash tracking-time triangle-free-process graph. We also isolate a density-free obstruction that any proof using this fractional surgery route must overcome.

09.
arXiv (CS.CV) 2026-06-24

Geometry-Aware Style Transfer in 3D Gaussian Splatting

In this paper, we present a novel geometry-aware style transfer framework for 3D Gaussian splatting (3DGS) that simultaneously transfers appearance attributes and geometric structures. Unlike prior works that primarily focus on color-based stylization and often overlook structural adaptation, our method explicitly incorporates geometry adaptation through a decoupled optimization scheme that alternately updates color and geometry parameters. This strategy alleviates potential interference between color and geometry updates, leading to stable and consistent scene-level geometry transformation. The decoupled optimization is enabled by the proposed geometry-aware contrastive feature matching (GCFM). GCFM integrates RGB, depth, and edge cues into a contrastive objective and is employed in both optimization phases to effectively transfer structural characteristics from style images to Gaussian primitives. Extensive experiments show that our approach achieves superior performance in both qualitative fidelity and quantitative metrics, significantly outperforming existing 3DGS-based stylization methods. Our code is available at \href{https://github.com/oweixx/gast}{https://github.com/oweixx/gast}.

10.
arXiv (CS.LG) 2026-06-17

Data augmented bootstrap: Unifying confidence interval construction by approximate invariance

arXiv:2606.09049v2 Announce Type: replace-cross Abstract: We propose the data augmented bootstrap (DAB), a framework for constructing confidence intervals from approximately invariant transformations of the data. As special cases, DAB recovers popular methods that rely on exact group symmetries, such as conformal prediction, wild bootstrap for Maximum Mean Discrepancy U-statistics and the recently proposed SymmPI. Meanwhile, DAB also recovers the classical bootstrap method, which exploits the dataset's approximate invariance under uniform sampling of data indices as the dataset size grows. For all DAB methods, we establish theoretical coverage results that interpolate between finite-sample and asymptotic guarantees according to the strength of the invariance, and without assuming a group structure. The approximate invariance is measured in the Kolmogorov distance and, for statistics that satisfy Gaussian universality, reduces to conditional mean and variance matching. This allows us to incorporate data augmentation (DA), a widely used machine learning heuristic based on approximate invariances, into known statistical methods. We empirically test the performance of incorporating DA into bootstrap, wild bootstrap and conformal prediction for simulated settings as well as for image, language and scientific data.

11.
arXiv (math.PR) 2026-06-19

Extremal representations of functions of matrices and applications to multivariate prediction

arXiv:2606.19359v1 Announce Type: cross Abstract: Motivated by two seminal results of multivariate prediction theory by Helson and Lowdenslager and by Wiener and Masani we prove extremal representations of functions of matrices and derive their prediction-theoretic consequences. We also sketch a way to obtain matricial inequalities from our results. The main goal of the paper is the computation of the infimum of a set of values of the form $tr(A \Delta A^*)$, where $\Delta$ is a given non-negative Hermitian $n \times n$ matrix and the choices for $A$ exhauste a certain set of $n \times n$ matrices. In particular, we focus on norm-bounded unit spheres with certain types of properties of unitary invariance, what allows an application of the theory of majorization.

12.
arXiv (CS.AI) 2026-06-19

AI-enhanced tuning of quantum dot Hamiltonians toward Majorana modes

arXiv:2601.02149v4 Announce Type: replace-cross Abstract: We propose a neural network-based model capable of learning the broad landscape of working regimes in quantum dot simulators, and using this knowledge to autotune these devices - based on transport measurements - toward obtaining Majorana modes in the structure. The model is trained in an unsupervised manner on synthetic data in the form of conductance maps, using a physics-informed loss that incorporates key properties of Majorana zero modes. We show that, with appropriate training, a deep vision-transformer network can efficiently memorize relation between Hamiltonian parameters and structures on conductance maps and use it to propose parameters update for a quantum dot chain that drive the system toward topological phase. Starting from a broad range of initial detunings in parameter space, a single update step is sufficient to generate nontrivial zero modes. Moreover, by enabling an iterative tuning procedure - where the system acquires updated conductance maps at each step - we demonstrate that the method can address a much larger region of the parameter space.

13.
arXiv (CS.CV) 2026-06-25

TACO: Towards Task-Consistent Open-Vocabulary Adaptation in Video Recognition

Adapting CLIP for open-vocabulary video recognition necessitates a delicate balance between newly acquired video knowledge and the pretrained generalization. While existing studies pursue this generalization-specialization trade-off with additional regularizations or constraints, we argue that they overlook the deviation of representations beyond the fine-tuning data distribution, resulting in suboptimal adaptation effects. We believe such deviation is inherited from the inconsistency between the fine-tuning and evaluation objectives, where model optimization is restricted to the known training distribution but evaluated on unseen ones. In this paper, we introduce TACO, a simple yet effective framework to mitigate the potential negative effects induced by this inconsistency. Our key insight is that adaptation should preserve OOD-relevant alignment beyond the training distribution. To this end, we propose Relative Structure Distillation, which regularizes the relative geometry of the representation space and suppresses harmful alignment shift during training. We further decouple the representation space from the optimization space with a lightweight specialization projection, allowing task-specific adaptation without directly overspecializing the representations used at test time. TACO establishes state-of-the-art performance on diverse benchmarks under cross-dataset and base-to-novel settings. Code will be released at https://github.com/ZMHH-H/TACO.

14.
arXiv (CS.CV) 2026-06-16

Fusing Transferred Priors and Physics-based Decomposition for Underwater Image Enhancement

The underwater images are captured within diverse water-medium conditions, leading to complex degradation, including color bias, low contrast, and blur effect. Recently, learning-based methods have demonstrated their potential for underwater image enhancement (UIE). However, most of the previous work focus on the training strategy or network design to make the enhanced result aligned well with the labels in datasets, ignoring that the labels are selected from the enhanced results of previous UIE methods and these pseudo-labels are noisy. Consequently, the performance of their models is not satisfactory to a certain extent. However, collecting the true labels of the underwater images is challenging. In this work, we propose a transfer learning-based UIE that does not require underwater images to have paired noisy or true labels for learning. Instead, the UIE task is first divided into global color correction, haze removal, and background noise suppression following the underwater physics. Then multiple types of prior from other vision tasks are leveraged as cross-domain supervision in each step. In this way, a novel UIE is available via transfer learning, and the physics-aligned UIE decomposition provides theoretical soundness. Qualitative and quantitative experiments demonstrate that our proposal based on physics and priors fusion achieves SOTA performance in the UIE task and effectively boosts downstream vision tasks, significantly outperforming benchmark methods. Project repo: https://github.com/Haru2022/P2-UIE.

15.
arXiv (CS.AI) 2026-06-17

From Paper to Program: Knowledge Externalization for AI-Assisted Quantum Many-Body Code Generation

作者:

arXiv:2604.04089v3 Announce Type: replace-cross Abstract: Large language models can write scientific code, but direct paper-to-program translation remains fragile when correctness depends on tacit conventions in the literature. We identify this bottleneck as knowledge externalization: converting implicit computational assumptions – index conventions, gauge choices, fermionic signs, contraction order, and memory constraints – into an explicit technical specification before implementation. We evaluate a multi-stage, human-in-the-loop workflow that inserts such a specification, with validation and stop gates, between theory extraction and code generation. The workflow is tested on two algorithmically distinct quantum many-body tasks: variational sweep-based Density-Matrix Renormalization Group (DMRG) from a pedagogical review and constructive Pfaffian conversion of Hartree–Fock–Bogoliubov states to matrix product states from the five-page Letter by Jin et al., Phys. Rev. B 105, L081101 (2022), for which no public code is available. For DMRG, all 16 specification-guided model pairings in a $4\times4$ grid satisfy physics-validation criteria, compared with 6/13 direct attempts. A prose-specification ablation indicates that externalized content, not \LaTeX{} formatting, is the essential ingredient. For Pfaffian-MPS, the workflow succeeds in 11/26 archived attempts, whereas direct prompting yields zero audited passes. Cross-specification transfer is asymmetric: non-GPT specifications implemented by GPT~5.5 pass 4/4, while GPT~5.5 specifications implemented by weaker models fail 4/4, indicating a residual implementation-model bottleneck. The resulting Paper-to-Program Many-Body skill provides an auditable protocol for AI-assisted implementation of many-body algorithms and for diagnosing where externalization succeeds or fails.

16.
arXiv (CS.CV) 2026-06-11

Towards Fully Automated Exam Grading: Fairness-Aware Recognition of Handwritten Answers with Foundation Models

Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while fully digital exams tend to force a didactic narrowing towards closed question formats. A practical middle ground keeps paper-based, problem-oriented tasks but records the assessment-relevant answers as single capital letters in a table that a machine can read. The open question is whether this reading can be made accurate and, above all, fair enough for unsupervised grading. Earlier automated approaches reached only about 88%–91% recognition – too low – and failed on the cases that matter most: answers placed outside the cell, crossed out, or written in cursive. We show that general-purpose vision-language foundation models (VLMs), which interpret the page rather than match pixel templates, close this gap. On a benchmark of 61 anonymised exams (3141 answer positions) the best model reaches 98.4% accuracy, well above the previous baseline. Crucially, we centre the evaluation on fairness: we distinguish false negatives (a correct answer marked wrong, which disadvantages the student) from false positives, and a lightweight prompt that supplies the reference solution as context lowers the false-negative rate to 0.58%. Under an exemplary grading scheme only three of the 61 exams would be graded worse, all caught by a student self-review step. Fully automated, fairness-aware exam grading at scale is therefore defensible; we release the anonymised benchmark to support reproducibility.

17.
arXiv (CS.CV) 2026-06-19

Training-Free Metrics for Synthetic Object Detection Data: A Proxy for Detector Performance

With the recent advent of image generative models, synthetic data are increasingly being used to supplement limited real datasets for training computer vision models. However, not all synthetic datasets improve performance equally, and their effectiveness can only be assessed by training a downstream model, which is computationally expensive and time-consuming. This problem is pronounced in the task of object detection, where the required annotations are much more dense due to bounding boxes. In this paper, we propose a pre-computable metric family, dubbed Conditional-Composition Domain Match (CCDM), which serves as a proxy for the relative utility of candidate synthetic training sets for downstream detection. Experiments on the VisDrone-DET dataset show that the CCDM metric families achieve a Spearman correlation of 1.0 with the downstream performance of YOLOv8, clearly outperforming existing metrics for synthetic image evaluation.

18.
arXiv (CS.CL) 2026-06-25

Real-Time Voice AI Hears but Does Not Listen

Speech conveys information through both words and vocal delivery. We evaluate four leading production realtime voice systems-OpenAI's GPT Realtime 2, Google's Gemini 3.1 Flash Live, and Alibaba's Qwen3.5 Omni Plus and Omni Flash-on tasks where the words and the delivery patterns both convey meaningful information. Across three consequential scenarios, all four systems act on the words rather than the voice. They end calls with crying callers who insist nothing is wrong, approve wire transfers authorized in frightened voices, and enroll callers whose agreement is clearly sarcastic. Surprisingly, this is often not a failure of perception. When asked directly, three of the four systems reliably identify the distress, fear, or sarcasm they later ignore when making decisions. We observe a similar pattern when these realtime voice systems estimate accent and age, as their responses frequently follow the biases of the words rather than the acoustic properties of the speaker. We term this disconnect between perception and action the emotional intelligence gap of voice AI. Prompting systems to explicitly attend to vocal delivery improves performance only partially and inconsistently. Our findings show that current realtime voice AI systems often behave as if speech had been reduced to a transcript, suggesting that they should be used with caution in settings where the tone and emotion of delivery convey important information.

19.
arXiv (CS.CL) 2026-06-18

TW-LegalBench: Measuring Taiwanese Legal Understanding

Large language models (LLMs) have shown impressive capabilities across diverse tasks, yet their performance on jurisdiction-specific legal reasoning remains underexplored. We present TW-LegalBench that utilizes Taiwanese legal system's rich official corpus open to the public to fill the gap in evaluating LLMs on Taiwanese law, among common-law benchmarks that focus on English sources and civil-law benchmarks focusing on sources of Simplified Chinese. TW-LegalBench comprises three task types: (1) over 16,000 multiple-choice questions (MCQs) across five years of official examinations in 18 professional domains; (2) 117 open-ended essay questions (OEQs) from examinations for legal professionals with official scoring rubrics; and (3) more than 14,000 legal judgment prediction (LJP) instances covering hundreds of crime categories. We evaluate 13 LLMs using accuracy for MCQs, a decomposed LLM-as-Judge framework based on the scoring rubric points for OEQs, and metrics for sentencing accuracy and statute citation for LJP. Our results reveal that top-performing models exceed the passing threshold for qualified lawyers (passing rate: 11%) but fall short of that for judges and prosecutors (passing rate: 1~2%). For LJP, while models demonstrate reasonable verdict type accuracy and sentence prediction capability, they struggle to cite exact legal articles. These findings highlight that reliable legal text generation remains challenging for LLMs, even though their performance on qualification examinations approaches human level.

20.
arXiv (CS.AI) 2026-06-12

Boosting Direct Preference Optimization with Penalization

作者:

arXiv:2606.12505v1 Announce Type: cross Abstract: Offline preference optimization has become a practical substitute for reinforcement learning from human feedback, but pairwise objectives such as Direct Preference Optimization (DPO) and its variants use only the chosen and rejected responses stored in a static dataset. This leaves a useful signal unused: the response that the reference model itself would generate for the same prompt. We propose Direct Preference Optimization with Penalization (DPOP), a simple extension of DPO that augments the base preference loss with a gated penalty on reference-greedy responses. DPOP activates this penalty only when the current policy still assigns a lower likelihood to the preferred response than to the rejected response. On AlpacaEval 2.0, DPOP improves length-controlled win rate over DPO, SimPO, and AlphaDPO on both Llama-3-8b-it and Gemma-2-9b-it, achieving relative gains of 5.3\% and 4.4\% over baselines on the two models, respectively. Ablations further show that a SimNPO-style length-normalized penalty is stronger than NPO and token-level unlikelihood in this setting.

21.
arXiv (CS.AI) 2026-06-18

Recursive Joint Simulation in Games

arXiv:2402.08128v3 Announce Type: replace Abstract: Game-theoretic dynamics between AI agents could differ from traditional human-human interactions in various ways. One such difference is that it may be possible to accurately simulate an AI agent, for example because its source code is known. Such an agent would then be fundamentally uncertain whether it is in the real world or in a simulation. Our aim is to explore ways of leveraging this possibility to achieve more cooperative outcomes in strategic settings. In this paper, we study an interaction between AI agents where the agents run a recursive joint simulation. That is, the agents first jointly observe a simulation of the situation they face. This simulation in turn recursively includes additional simulations (with a small chance of failure, to avoid infinite recursion), and the results of all these nested simulations are observed before an action is chosen. We show that the resulting interaction is strategically equivalent to an infinitely repeated version of the original game, allowing a direct transfer of existing results such as the various folk theorems. As evidence that the equivalence is robust, we show that it holds even when we relax some of the assumptions and that it also holds ``from the inside'' – meaning, for an agent that finds itself inside the game and has self-locating uncertainty.

22.
arXiv (CS.CV) 2026-06-25

Homogeneity Bias in Open-Weight LLMs Is Robust to Decoding Hyperparameters

Large language models (LLMs) reproduce homogeneity bias – the tendency to portray marginalized groups as more internally similar than dominant groups – but whether this bias is stable or an artifact of inference settings has only been studied in single proprietary models. We map homogeneity bias across a 5x5 temperature-by-top-p grid in seven open-weight instruction-tuned LLMs (7-20B parameters). Hispanic and Asian Americans are portrayed as more homogeneous than White Americans in at least 18 of 20 hyperparameter configurations across six of seven models, including at extreme sampling settings. African American and gender bias show model-specific variation in direction. A conservative cell-level re-analysis confirms Hispanic and Asian homogeneity as robust, while weaker African American and gender signals largely do not survive, establishing group-specific robustness. We also apply the same grid to a names-based paradigm in which group identity is signaled via racially distinctive surnames rather than explicit labels. The names paradigm corroborates Hispanic and Asian homogeneity bias, but Black-coded surnames elicit robustly less homogeneous outputs than White-coded names in every model tested – a reversal absent from the label paradigm – showing that how group identity is operationalized shapes which biases surface and in which direction.

23.
arXiv (CS.CV) 2026-06-12

Augmentation techniques for video surveillance in the visible and thermal spectral range

In intelligent video surveillance, cameras record image sequences during day and night. Commonly, this demands different sensors. To achieve a better performance it is not unusual to combine them. We focus on the case that a long-wave infrared camera records continuously and in addition to this, another camera records in the visible spectral range during daytime and an intelligent algorithm supervises the picked up imagery. More accurate, our task is multispectral CNN-based object detection. At first glance, images originating from the visible spectral range differ between thermal infrared ones in the presence of color and distinct texture information on the one hand and in not containing information about thermal radiation that emits from objects on the other hand. Although color can provide valuable information for classification tasks, effects such as varying illumination and specialties of different sensors still represent significant problems. Anyway, obtaining sufficient and practical thermal infrared datasets for training a deep neural network poses still a challenge. That is the reason why training with the help of data from the visible spectral range could be advantageous, particularly if the data, which has to be evaluated contains both visible and infrared data. However, there is no clear evidence of how strongly variations in thermal radiation, shape, or color information influence classification accuracy. To gain deeper insight into how Convolutional Neural Networks make decisions and what they learn from different sensor input data, we investigate the suitability and robustness of different augmentation techniques...

24.
arXiv (CS.LG) 2026-06-25

Leaking Circuit Secrets: Gradient Leakage Attacks on Graph Neural Networks

arXiv:2606.25589v1 Announce Type: new Abstract: As graph neural networks (GNNs) become standard tools for critical tasks in circuit design and analysis, their security and privacy risks require careful attention. Here, we present the first comprehensive evaluation of gradient leakage attacks (GLAs) on GNNs in circuit-design and hardware-security tasks, a practical threat that has been largely overlooked. We assess state-of-the-art (SOTA) GNNs, including GraphSAGE, GCN, GIN, and GAT, trained on standard netlist benchmarks (ISCAS'85, EPFL, and TrustHub), for their fundamental vulnerability to GLAs. We find that GLAs can expose sensitive information, such as gate types and distinctive properties of hardware Trojans, which may assist adversaries in analyzing logic locking schemes or evading Trojan detection mechanisms. Our analysis shows that these risks are influenced by architectural features, with attention mechanisms (GAT) exacerbating leakage, while injective aggregation (GIN) provides comparatively stronger resilience. We further evaluate several SOTA defense techniques, including differential privacy, gradient clipping, secure aggregation, model compression with quantization, and adversarial training. We find that these techniques improve resilience only in specific settings and can also compromise model performance. Overall, our work provides key insights toward privacy-preserving GNNs and highlights the need for more robust and efficient defenses. We release our full methodology and artifacts.

25.
arXiv (CS.CL) 2026-06-12

ChiKhaPo: A Large-Scale Multilingual Benchmark for Evaluating Lexical Comprehension and Generation in Large Language Models

Existing benchmarks for large language models (LLMs) are largely restricted to high- or mid-resource languages, and often evaluate performance on higher-order tasks in reasoning and generation. However, plenty of evidence points to the fact that LLMs lack basic linguistic competence in the vast majority of the world's 3800+ written languages. We introduce ChiKhaPo, consisting of 8 subtasks of varying difficulty designed to evaluate the lexical comprehension and generation abilities of generative models. ChiKhaPo draws on existing lexicons, monolingual data, and bitext, and provides coverage for 2700+ languages for 2 subtasks, surpassing any existing benchmark in terms of language coverage. We further show that 6 SOTA models struggle on our benchmark, and discuss the factors contributing to performance scores, including language family, language resourcedness, task, and comprehension versus generation directions. With ChiKhaPo, we hope to enable and encourage the massively multilingual benchmarking of LLMs.