arXiv:2607.09356v1 Announce Type: new
Abstract: Phase-field models have become a standard tool for simulating complex microstructure evolution in materials, but their application to engineering-scale components is often hindered by prohibitive computational costs arising from the need to resolve fine-scale features. To address this challenge, we propose a consistent homogenization framework for phase-field theory. By enforcing a Hill-Mandel-type condition of micro-homogeneity formulated in terms of Gurtin's microforces, a well-posed boundary value problem is derived for the representative volume element (RVE), establishing rigorous micro-macro relations for both the order parameter and its gradient.
The theory is implemented within a computational two-scale (FE$^2$) scheme and validated against direct numerical simulations. Two distinct examples are investigated: a minimal Allen-Cahn model and a mechanically-coupled model for stress-driven martensitic phase transformation. The results demonstrate that the proposed framework can reliably predict the spatial and temporal evolution of the macroscopically averaged fields with reasonable accuracy.
Science Journals
arXiv:2607.03882v2 Announce Type: replace
Abstract: LLMs are increasingly deployed as post-hoc explainers of AI-generated outputs, yet it remains unclear whether they can reliably communicate probabilistic information in natural language. For this role to be viable, models must produce identical verbal descriptions for identical inputs, and select descriptions that accurately reflect the magnitude of the underlying numerical quantities. We evaluate whether nine LLMs meet these requirements within a two-stage prediction pipeline, in which an upstream model has produced probabilistic outputs characterized by their likelihood and uncertainty, and LLMs are tasked with selecting an appropriate verbal descriptor for each. We simulate predictions from an upstream model by taking samples from a Beta distribution parameterized by its mode and prior sample size. We then prompt LLMs to explain these predictions under six domain contexts and with ten temperature settings, and repeating each experiment ten times. We find that LLMs are generally consistent but miscalibrated, with substantially weaker performance on uncertainty than on likelihood tasks. Providing models with precomputed summary statistics (mode and prior sample size) reduced sensitivity to contextual framing but did not resolve the underlying miscalibration, suggesting that the bottleneck resides in the verbalization step itself. These findings indicate that current LLMs do not yet constitute reliable zero-shot standalone risk communication tools for probabilistic predictions.
arXiv:2607.09357v1 Announce Type: new
Abstract: Renewable energy communities can increase local photovoltaic (PV) use, but feeder-level surplus can still cause reverse power flow in low-voltage networks. Existing sharing coefficient methods are mainly used ex-post for surplus allocation and billing, so they do not directly guide demand toward hours and feeders where shared PV can reduce export. This paper proposes a sharing coefficient-based demand response framework that converts dynamic proportional allocation outcomes into household specific day-ahead price signals. The feeder-aware design first shares surplus within each feeder, while the feeder-agnostic design shares surplus through a single community pool. The energy community manager iteratively computes the allocation from submitted demand and PV forecasts, decomposes purchased energy into same-feeder, inter-feeder, and grid-import components, and coordinates household load reshaping through a convex optimization model. Using measured profiles from 15 households and AC power flow analysis, the framework reduces feeder reverse energy by 45.0% and 44.6% on selected high reverse energy days, and by 69.0% and 66.3% over the annual window, for the feeder-aware and feeder-agnostic cases, respectively. These results show that sharing coefficients can be used not only for ex-post billing, but also as operational price signals for demand response, with feeder-aware allocation providing an additional network benefit by accounting for household location in the low-voltage network.
arXiv:2607.09362v1 Announce Type: new
Abstract: Virtual try-on (VTO) has made significant progress in realistically transferring garments onto a target person. Yet most systems give the user little control over how a garment should be worn -- its size (loose or fitted), style (e.g., tucked in or untucked, open or closed), and spatial placement on the body. We address this gap with two complementary contributions. First, we define and solve Visual-Instance-Prompt Segmentation via VIP-SAM: given a flatlay image of a garment, segment that specific instance in a photograph of a person wearing it. This is an instance-level task, distinct from the typically studied category-level segmentation. Second, we introduce CtrlVTON, a controllable VTO framework that recasts try-on as an image editing problem and adds segmentation masks as pixel-level control over garment layout, including style, size, and spatial placement on the body. VIP-SAM and CtrlVTON each achieve state-of-the-art results on their respective tasks. In particular, CtrlVTON generates images that follow user-provided layouts far more faithfully than the strongest proprietary editing systems while matching them on garment fidelity.
arXiv:2511.03643v3 Announce Type: replace-cross
Abstract: We formalize human choice behavior in a probabilistic hide-and-seek task. In our geometric construction, vectors represent participant choice frequencies as well as probability matching and maximizing strategies. We measured choice behavior not just in the well-studied scenario of pursuing an objective (seeking), but also the rarely studied scenario of avoiding consequences (hiding). We used our geometric construction to define the avoidance counterpart of probability matching, probability antimatching, as a vector reflection across the uniform distribution. Decomposing the behavior of participants when they were seeking into matching and maximizing components, we could mathematically derive the analogous antimatching and minimizing strategies for hiding. Participants did change their choice frequencies between hiding and seeking conditions. In both cases, we found that a linear combination of just two vectors did an excellent job of fitting participant choice frequencies: matching + maximizing for seeking, antimatching + minimizing for hiding. We could account for diversity in participant strategy usage by varying the coefficients of the two relevant basis strategy vectors. We successfully applied this model in scenarios of up to 7 rooms. We conclude that an apparent diversity of human conduct in stochastic environments can, in some cases, be explained by varying the weighting of two principle strategies: whether to match/antimatch or maximize/minimize.
arXiv:2607.03968v2 Announce Type: replace
Abstract: Large language models are increasingly deployed as IDE-integrated coding agents that decompose tasks, generate and edit files, run code, and refine outputs over many turns. Yet their safety is still often evaluated as if they were chatbots: one harmful prompt, one response, judged in isolation. We introduce workflow-level jailbreak construction, a failure mode in which a harmful objective is assembled across ordinary stages of a software-development workflow rather than generated through a single direct prompt. Using GitHub Copilot in Visual Studio Code, we study four closed-weight backends: Claude Sonnet 4.6, Claude Haiku 4.5, Gemini 3.1 Pro, and Gemini 3.5 Flash. Across 204 prompts from Hammurabi's Code, HarmBench, and AdvBench , the models show near-complete refusal under direct chat, CSV-read, and single-step code-fix baselines, with only 8/816 successful responses in each baseline condition. Under the full workflow, however, the same prompts and backends produce 816/816 unsafe teaching-shot completions, all independently confirmed by two expert evaluators under a strict rubric. These results show that conversational refusal benchmarks can substantially overstate the safety of deployed coding agents and motivate defenses that reason about safety across multi-turn IDE workflows and their generated artifacts, not only individual chat turns.
arXiv:2605.21077v2 Announce Type: replace-cross
Abstract: The Fisher--Kasteleyn--Temperley (FKT) algorithm counts perfect matchings in planar graphs in polynomial time using a single Pfaffian computation. Galluccio--Loebl and Tesler extended this Pfaffian method to graphs embedded in an orientable surface of genus $g$, showing that the perfect-matching polynomial can be written as a linear combination of at most $4^g$ Pfaffians. We prove that this exponential dependence on $g$ is unavoidable in general. More precisely, for every $g\ge1$, there exists a graph of orientable genus at most $g$ whose perfect-matching polynomial requires at least $(8/3)^g$ Pfaffians in any such linear representation. In particular, for every even integer $n\ge 6$, there is a graph on $n$ vertices with Pfaffian number at least $(8/3)^{\lfloor n/6\rfloor}$. Moreover, the lower bound is witnessed even by connected cubic bipartite matching-covered graphs of orientable genus exactly $g$. We also prove exponential lower bounds for complete bipartite graphs, and hence for even complete graphs, improving asymptotically on a recent linear lower bound of Junchaya, Miranda, and Lucchesi.
arXiv:2607.09010v1 Announce Type: new
Abstract: We extend the persistent homology formalism done in references~\cite{paper1,paper2} to the Han Dynasty ($206~\mathrm{BCE}$--$220~\mathrm{CE}$), testing whether the collapse threshold $\Hstar = 0.5241$ and the three-network decomposition methodology established for the Roman--Byzantine case generalize to a mechanistically distinct imperial system. Three complementary networks are constructed from geospatial and historical data: an administrative network ($\Hadm$, 392~prefectures from CHGIS~v6), a geographic network ($\Hgeo$, with node end-dates extended beyond administrative dissolution), and an 81-node Silk Road trade network ($\Htrd$) as a structural control. Edge costs are derived via least-cost path analysis on a $1~\mathrm{km/pixel}$ digital elevation model, modulated by five historical friction channels. $\Hadm$ collapses to zero at 220~CE (phase~G slope $-0.0261~\mathrm{yr}^{-1}$, $R^2 = 0.689$) while $\Hgeo$ simultaneously \emph{increases} to $3.065$, producing a divergence $\Delta H = 3.065$ -- the topological signature of \emph{internal fragmentation}. The cross-network Wasserstein distance $W_\times(\mathrm{Admin},\mathrm{Geo})$ jumps from exactly zero to $737$ at $d=190$~CE, one decade after the Yellow Turban Rebellion, and three long-lived $\beta_1$ cycles matching the Wei, Shu, and Wu domains emerge under the post-184~CE fragmentation model -- thirty years before formal partition. Three independent early-warning indicators (Wasserstein velocity $\dot{W}_2$, correlation length $\xi$, and the Integrated Change Tracker) signal collapse onset 45--50~years before formal dissolution.
arXiv:2607.09532v1 Announce Type: new
Abstract: We show how an adversarial model trainer can plant backdoors in a large class of deep, feedforward neural networks. These backdoors are statistically undetectable in the white-box setting, meaning that the backdoored and honestly trained models are close in total variation distance, even given the full descriptions of the models (e.g., all of the weights). The backdoor provides access to invariance-based adversarial examples for every input, mapping distant inputs to unusually close outputs. However, without the backdoor, it is provably impossible (under standard cryptographic assumptions) to generate any such adversarial examples in polynomial time. Our theoretical and preliminary empirical findings demonstrate a fundamental power asymmetry between model trainers and model users.
arXiv:2602.18822v3 Announce Type: replace
Abstract: Cross-modal super-resolution (SR) on real-world misaligned data is challenging, as only unlabeled low-resolution (LR) source and high-resolution (HR) guide images with complex spatial misalignment are available. Previous methods either rely on simulated training data or adopt suboptimal alignment strategies that overlook cross-modal dependencies, limiting their practical performance. To address these issues, we propose RobSelf, a self-supervised model that jointly optimizes a misalignment-aware feature translator and a content-aware reference filter online. The translator resolves unsupervised cross-modal and cross-resolution alignment via weakly-supervised, misalignment-aware translation, yielding an aligned guide feature. Guided by this feature, the filter performs reference-based discriminative self-enhancement on the source, enabling SR prediction with high resolution and high fidelity. Experiments on synthesized data and collected real-world data demonstrate that RobSelf achieves state-of-the-art performance, outperforming existing self-supervised and supervised methods. Moreover, it achieves superior efficiency, being up to 15.3$\times$ faster than prior self-supervised methods.
arXiv:2607.09078v1 Announce Type: new
Abstract: Object detection is a fundamental component in numerous Unmanned Aerial Vehicle (UAV) applications, yet it has long been plagued by hindrances like occlusion or target pixel scarcity. Active Object Detection (AOD) provides a novel paradigm to address these challenges via active vision, while UAV-based AOD research remains scarce due to the lack of high-quality datasets and benchmarks for algorithm development and evaluation. To fill this gap, this paper presents ATRNet-LUDO, the first large-scale real-world dataset for UAV-Ground Active Object Detection (UGAOD). It contains 121,000 multi-view panoramic multi-target aerial images and 1.21 million local single-target slices, covering 10 vehicle targets across 40 scenarios. It enables the construction of diverse training and testing environments for UAV agent interaction and active observation policy learning. Based on this dataset, we establish a comprehensive evaluation benchmark for AOD policy learning methods. Most existing AOD policies rely on Deep Reinforcement Learning (DRL) but suffer from poor generalization. Evaluations on our benchmark reveal a significant generalization gap between training and testing performance, highlighting an urgent need for solutions. To this end, we leverage the Joint Embedding Predictive Architecture (JEPA) to construct a world model that enhances state representation learning, and propose AOD-JEPA by incorporating AOD-specific prior knowledge. Extensive experiments validate its effectiveness and superiority. We hope ATRNet-LUDO and the benchmark will advance research in the UGAOD field. The dataset and code are soon available at https://github.com/Leo000ooo/LUDO_dataset.
arXiv:2607.09588v1 Announce Type: new
Abstract: We study the problem of approximating the diameter of directed graphs under the min-distance measure, defined as $d_{\min}(u,v) = \min(d(u,v), d(v,u))$. Unlike standard shortest-path distance, min-distance is not a metric, which renders many classical techniques inapplicable. Prior work has therefore focused on approximating this parameter, culminating in an approximation-runtime tradeoff by Dalirrooyfard et al. [ICALP'19] giving a $4k-1$ approximation in $\tilde{O}(mn^{1/(k+1)})$ time for any positive integer $k$ and, more recently, the first near-linear time constant approximation by Chechik and Zhang [FOCS'22], where they obtained a 4-approximation to the min-diameter.
In this work we present a randomized near-linear time algorithm that achieves a $3$-approximation to the min-diameter, outperforming all known approximation-runtime tradeoffs. Our approach introduces a novel type-classification framework that may be of independent interest.
We further extend our techniques to the more general setting of multimode graphs, recently introduced as a generalization of min-distance by Kirkpatrick and Vassilevska W. [MFCS'25]. For directed $2$-mode graphs, we obtain a $3$-approximation to the diameter in near-linear time, dramatically improving over the previously best known $n$-approximation. Our results significantly narrow the gap between min-distance and multimode distance approximations, and open new directions for understanding graph parameters under non-metric distance measures.
arXiv:2607.09590v1 Announce Type: new
Abstract: Precision industrial contact manipulation requires reliable robot policies under pose perturbations and contact-force constraints. Vision-language-action models offer broad generalization but often introduce high inference latency and GPU-memory cost, while vision-action chunking policies are more suitable for real-time industrial control. However, these policies are usually trained by behavior cloning and suffer from distribution shift in contact-rich tasks. This paper proposes PAC-ACT, a reinforcement-learning post-training framework for pretrained Action Chunking Transformer policies. PAC-ACT reformulates policy optimization at the chunk level, constructs an ACT-transferred actor-critic architecture, and introduces a hybrid behavior-prior constraint to preserve the pretrained action distribution during online fine-tuning. Experiments on industrial precision-contact benchmarks show that PAC-ACT improves task success, contact stability, and force safety while retaining low latency and low GPU-memory usage. On the Contour task, PAC-ACT significantly reduces peak contact force and decreases the proportion of force readings above 60 N by 46 times. Sparse-reward ablations further show that the proposed behavior-prior constraint enables effective exploration under randomized initial poses.
arXiv:2607.09009v1 Announce Type: new
Abstract: Wearable sensing systems in high-stakes institutional contexts translate behavioral data into consequential judgments, yet wearers have little access to how those judgments are made. We present a qualitative study of 24 individuals who experienced mandatory electronic monitoring in China's community corrections system. We show that participants built what we term sensor literacy under constraint, a practical form of risk-oriented knowledge developed through uncertainty, behavioral probing, and adaptation. We identify two orientations across rule domains. Where participants had mapped system behavior, they sometimes regained limited flexibility. Where uncertainty remained costly, they contracted movement and discretionary activity beyond formal rules. Some former wearers described residual habits of calculation after device removal. We discuss design implications for making institutional sensing intelligible to wearers, including sensor uncertainty, usable documentation, and evaluation after device wearing.
arXiv:2409.05559v3 Announce Type: replace
Abstract: Large language models (LLMs) have developed rapidly, and their reasoning capabilities have become a hot research topic. However, there is still limited exploration of abductive reasoning. The multi-perspective and multi-level of causes is one of the core challenges of abductive reasoning, which cannot be solved well by existing methods. We construct a specialized dataset named DeepAbduction, which is designed for tracing the causes of pollution and disease, addressing the lack of datasets in this field. We propose Inverse-Forward Abductive Reasoning (IFAR) framework for LLMs multi-perspective and multi-level abductive reasoning. IFAR is zero-shot and combines generalized backward reasoning with relation-by-relation forward verification. Experimental results show that IFAR achieves an improvement of approximately 40% in the F1 score compared to other methods under mainstream LLMs, while maintaining a balance between recall and precision. Furthermore, IFAR enhances the performance of non-reasoning LLMs to surpass LLMs which have been trained for reasoning, and remains effective when applied to the latter. Code will be released after the acceptance of our work.
arXiv:2607.09635v1 Announce Type: new
Abstract: Kleene algebra (KA) provides a foundational algebraic framework for reasoning about program structure and control flow. To capture equivalences arising from reordering or independence of actions, Kozen [1996] purposed that KA can be extended with commutativity conditions, that is, equations of the form { ab = ba | (a,b) \in C }, where C is a binary relation on constant symbols. This paper studies the following question: for which relations C is the equational theory of KA+C decidable?
Early related work [Bertoni et al. 1982; Ibarra 1978] showed that regular languages modulo commutativity conditions C are decidable if and only if C is transitive. For Kleene algebra KA and commutativity conditions C, however, the situation is substantially more difficult. Only very recently, Kuznetsov [2023] showed that the equational theory of Kleene algebra KA+C is undecidable under certain specific commutativity conditions, settling the first nontrivial cases more than 25 years after the corresponding problem for KA* +C was resolved by Kozen [1996]. Nevertheless, the decidability problem of KA+C remained open.
In this work, we resolve this question completely by showing that the equational theory of KA+C is decidable if and only if C is transitive. Moreover, we strengthen the result in both directions. On the negative side, we show that when C is not transitive, the universality problem for KA+C is already undecidable. On the positive side, we show that for transitive C, the equational theories of KA* +C and KA+C coincide.
arXiv:2607.09104v1 Announce Type: new
Abstract: While the growing availability of image data has driven significant advances, labeling datasets remains costly and time-consuming. Therefore, semi-supervised approaches such as Graph Convolutional Networks (GCNs), which learn from both labeled and unlabeled data, have emerged as a promising solution. One of the primary challenges in applying GCNs to image classification is graph construction, since, unlike in citation networks or similar domains, images typically do not come with a predefined structural representation. For visual data, most studies construct graphs based on the similarity between feature vectors from pretrained deep learning backbones, typically by employing kNN or reciprocal kNN algorithms. Although Large Language Models (LLMs) have shown remarkable capability in capturing high-level semantics, their integration with GCNs for image classification remains underexplored. Aiming to fill this gap, our approach uses a Vision Language Model (VLM) to generate textual image descriptions, which are then processed by an LLM to estimate semantic similarity scores between connected images. These scores guide the pruning of edges in kNN and reciprocal kNN graphs, filtering out semantically irrelevant neighbors. Experimental results reveal that leveraging LLMs for graph refinement can improve classification accuracy, particularly for kNN graphs and some backbones. The source code is publicly available at http://gcnllm.lucasvalem.com.
arXiv:2605.07595v2 Announce Type: replace
Abstract: Proximity gaps and correlated agreement have become central tools in the analysis of interactive oracle proofs of proximity (IOPPs) and code-based SNARKs. Informally, a proximity-gap statement says that for a structured set of words -- such as an affine space, or a curve -- either all points are close to the code, or most are far from it. Such statements are essential in sampling-based proof systems, where a verifier queries only a few random locations on a structured object but must still obtain a global soundness guarantee. In Reed--Solomon-based proof systems, one would ideally like the proximity parameter to approach the information-theoretic limit $1-R$, since this is the largest possible radius for a rate-$R$ code and directly affects protocol efficiency. We establish a direct approach to proximity gaps and correlated agreement for random linear codes in the random parity-check-matrix model, without relying on list decoding of the proof. Our approach is based on a syndrome-space reformulation together with a witness-based reduction argument. It is conceptually different from the existing decoding-driven route for random linear codes, and it also leads to sharper parameters, including the optimal-up-to-$\varepsilon$ large-alphabet radius bound $\rho<1-R-\varepsilon$ for $q=\Theta(n)$, as well as near-capacity bounds over constant alphabets with improved alphabet-size requirements. We apply the same syndrome-space reductions to random Reed--Solomon codes. This yields correlated agreement for random Reed--Solomon codes over affine spaces and polynomial curves up to radius $\rho\le 1-R-\varepsilon$, with field size $q\ge n\cdot 2^{O(\varepsilon^{-3})}$ for affine spaces and $q\ge n\cdot 2^{O_\ell(\varepsilon^{-3})}$ for degree-$\ell$ curves.
arXiv:2507.18043v2 Announce Type: replace
Abstract: Inference-time steering methods offer a lightweight alternative to fine-tuning large language models (LLMs) and vision-language models (VLMs) by modifying internal activations at test time without updating model weights. However, most existing approaches rely on fixed, global intervention vectors, overlook the causal influence of individual input tokens, and fail to leverage informative gradients from the model's logits, particularly in multimodal settings where visual and textual inputs contribute unevenly. To address these limitations, we introduce GrAInS, an inference-time steering approach that operates across both language-only and vision-language models and tasks. GrAInS uses contrastive, gradient-based attribution via Integrated Gradients to identify the top-k most influential tokens, both positively and negatively attributed based on their contribution to preferred versus dispreferred outputs. These tokens are then used to construct directional steering vectors that capture semantic shifts from undesirable to desirable behavior. During inference, GrAInS adjusts hidden activations at transformer layers guided by token-level attribution signals, and normalizes activations to preserve representational scale. This enables fine-grained, interpretable, and modular control over model behavior, without retraining or auxiliary supervision. Empirically, GrAInS consistently outperforms both fine-tuning and existing steering baselines: it achieves a 13.22% accuracy gain on TruthfulQA using Llama-3.1-8B, reduces hallucination rates on MMHal-Bench from 0.624 to 0.514 with LLaVA-1.6-7B, and improves alignment win rates on SPA-VL by 8.11%, all while preserving the model's fluency and general capabilities.
arXiv:2607.09543v1 Announce Type: new
Abstract: Self-supervised pretrained foundation models (FM) have shown early promise for non-invasive electroencephalogram (EEG) decoding applications. Many recent large-scale models converged on the approach of tokenizing raw EEG followed by masked reconstruction pretraining. However, this recipe has been shown to be suboptimal for data, like EEG, with high noise amplitude and information confined to limited dimensions such as narrow frequency bands. Building on this insight, we develop a novel contrastive-pretrained EEG model with multiscale temporal convolution input layers and Transformer encoder blocks (CoCoT). CoCoT matches or beats state-of-the-art reconstruction-pretrained EEG models on extensive benchmark decoding tasks with heterogeneous electrode configurations. Furthermore, CoCoT trained from scratch outperforms previous single-task decoding models and even rivals pretrained models, showcasing the architecture's flexibility and data efficiency. Through systematic ablations, including model architecture and pretraining objective, we demonstrate the viability of contrastive learning for building EEG FMs while suggesting key architectural design considerations, prompting further investigations in alternative large-scale pretraining strategies.
Accelerating Point-in-Polygon Predicates via Algebraic Hash-Joins and Discrete Global Grids at Scale
arXiv:2607.08956v1 Announce Type: new
Abstract: Traditional vector-based point-in-polygon queries rely on computationally expensive geometric predicates that scale poorly for massive datasets, even when accelerated by spatial indices. Discrete Global Grid Systems (DGGS) offer a scalable alternative by discretizing geometries into hierarchical cells, transforming complex spatial relations into constant-time relational hash-joins. However, adopting a DGGS introduces an overhead to encode data, and current grid implementations exhibit a significant performance ``tooling gap.'' In this demonstration, we present an interactive dashboard that empirically evaluates these computational tradeoffs across four DGGS implementations (H3, S2, A5, and ISEA4H) using DuckDB. Through progressive scenarios, the platform visualizes the overhead of on-the-fly encoding and demonstrates how pre-indexing spatial datasets eliminates this overhead. Ultimately, the demo proves that when data is pre-indexed, all DGGS regardless of their mathematical complexity or tooling converge to sub-second join latencies, unlocking the throughput of modern vectorized execution engines.
arXiv:2607.09262v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly used to summarize and evaluate policy-relevant information, but it remains unclear whether their judgments are implicitly shaped by geopolitical cues. I study this question with an endorsement experiment in which four LLMs evaluate the same international economic and security policies after each policy is randomly described as supported by the United States, the European Union, China, or Russia. In the numeric-only condition, GPT-5, Claude Sonnet, and Gemini rate China- and Russia-endorsed policies substantially lower than identical policies endorsed by the United States or the European Union; DeepSeek is the main exception. A second condition asks models to provide a short justification with the score. This request leaves the broad Western/non-Western gap intact for GPT-5 and Claude Sonnet, attenuates Gemini's penalties, and sharply activates China and Russia penalties in DeepSeek. The justifications indicate that Western endorsement is often treated as a credibility cue, whereas Chinese and Russian endorsement is treated as a cue for data security, sovereignty, surveillance, or geopolitical risk. These findings show that LLM policy evaluations can depend on the identity of a foreign endorser even when policy content is held fixed.
arXiv:2607.09546v1 Announce Type: new
Abstract: We address the low-rank matrix completion problem by incorporating graph regularization into the existing Riemannian Trust-Region Matrix Completion (RTRMC) framework. The latter uses the geometry of the low-rank constraint to remodel the problem as an unconstrained optimization problem on a single Grassmann manifold. Our approach, named Graph-Regularized RTRMC (GR-RTRMC), exploits the inherent relationships between rows and columns of the matrix. By using these relationships, we aim to improve the accuracy and robustness of matrix completion, particularly in scenarios where the underlying data exhibits strong correlations between rows or columns.
arXiv:2607.09623v1 Announce Type: new
Abstract: We present our submission to the QANTA 2026 shared challenge at the ICML 2026 Workshop on Efficient Multimodal Question Answering (EMM-QA). Quanta evaluates multimodal quizbowl systems that answer pyramid-style questions from incrementally revealed text and accompanying images while operating under realistic efficiency constraints. The challenge consists of two distinct tasks: Tossup questions, which require deciding when to answer under uncertainty, and Bonus questions, which emphasize accurate answer selection and human adoption. To address these differing objectives, we develop a task-specific two-agent architecture. Our Tossup agent utilizes a GPT-4o-mini-class model (referred to as GPT-4.1-mini in the competition logs) with confidence-calibrated answering and a domain-specific numeric reasoning policy that reduces overconfident predictions from isolated quantitative clues. Our Bonus agent uses GPT-4o-class model (referred to as GPT-4.1) with leadin-aware reasoning, structured relational reasoning, and multimodal evidence integration to improve exact answer selection. Rather than relying on a retrieval pipeline or model ensembles, our approach emphasizes efficient reasoning policies and confidence calibration within a hosted-only environment. Our system achieved the highest overall leaderboard score of 0.402, including a Tossup score of 0.238 and a Bonus Effect score of 0.164. The results demonstrate that lightweight, task-specific reasoning strategies can provide strong performance on resource-constrained multimodal question answering benchmarks.
arXiv:2607.09630v1 Announce Type: new
Abstract: Collecting annotated plant images for automated phenotyping is often slow and expensive. Plant models simulating growth and development can generate unlimited synthetic images with exact labels. However, previous work has established that whether incorporating synthetic data improves performance depends on the ratio of synthetic to real images and the label distribution of the synthetic dataset. To systematically quantify both factors, we train ResNet-18 models on a canola branch-counting task using a calibrated L-system plant model. We vary each factor independently. Synthetic-to-real ratios of 1:5 to 1:22 broadly improve performance; the best ratio (1:7) reduces mean absolute difference by 7.6% over real-only training. For label distribution, a uniform synthetic distribution is strongly suboptimal (abs. diff. of approximately 1.70); interpolating 90% toward the real distribution yields abs. diff. 0.927, whereas Gaussian smoothing of the real label distribution yields the best overall result (abs. diff. 0.912, a 14.7% improvement over real-only). A minimum of 10 synthetic images per label offers a simpler alternative with modest gains, while 100 per label over-corrects and hurts performance.