arXiv:2607.13239v1 Announce Type: new
Abstract: Foundation models, including large language models (LLMs) and vision-language models (VLMs), are increasingly used for transportation management center (TMC) tasks such as anomaly detection, incident reporting, and traveler information. Deploying multiple such models across TMC functions raises a portfolio question: which model should serve each function, in which deployment mode, and under what shared hardware budget? We formulate this as the Foundation Model Deployment Portfolio (FMDP) problem, a mixed-integer program minimizing total cost of ownership (TCO) subject to per-function quality, latency, and safety constraints over shared GPU capacity. We prove the problem NP-hard by reduction from the 0-1 knapsack problem and propose a polynomial-time greedy heuristic. In an illustrative case study with five TMC functions and 19 candidate (model, mode) pairs, FMDP identifies a mixed portfolio costing $34/mo (97% below the cheapest feasible all-closed-API baseline) by routing four functions to open-source APIs and the one function whose quality floor no open-source model meets to a closed API. Break-even analysis shows that on-premise GPU investment becomes reasonable only above approximately 309 vision queries/hour or if API prices double.
Science Journals
arXiv:2607.13241v1 Announce Type: new
Abstract: Traffic forecasting is highly challenging due to complex and nonlinear spatial and temporal dependencies. Self-attention mechanisms have been widely adopted to model dynamic and long-range dependencies, achieving state-of-the-art performance, but suffer from limited scalability due to quadratic computational and memory complexity. To address this, we propose an Efficient Multi-Attention Graph Network (EMAGN) that linearises the spatial attention mechanism itself, inspired by the theory of fast high-dimensional Gaussian filtering. Two learned clustering matrices C_k and C_v adaptively group key and value vectors into M super-clusters, reducing complexity from O(N^2 d) to O(NMd) without sacrificing the flexibility of attention for dynamic dependency modelling. Experimental results on PEMS-BAY and METR-LA show that EMAGN achieves accuracy within 2.7-3.2% MAE of full-attention GMAN while reducing training time by 32%, inference time by 38%, and GPU memory by 58%. Critically, at K=16 attention heads, full-attention GMAN runs out of memory on a standard 11 GB GPU entirely while EMAGN continues to operate, demonstrating a categorical expansion of feasible model configurations. EMAGN also surpasses Linformer and Performer in both accuracy and efficiency within the same backbone, owing to its traffic-network-aware adaptive clustering.
arXiv:2607.13608v1 Announce Type: new
Abstract: Automatic scientific discovery has long been a goal of computational scholars - a machine that can discover nature's secrets on its own, moving computational systems beyond data-fitting tools toward the generation and refinement of mechanistic models of the universe. Recent advances in symbolic regression (SR) and large-language-model (LLM)-based agents suggest that such systems can recover equations from data, incorporate domain priors, and automate parts of the research workflow. However, most existing approaches either focus on narrow equation-discovery benchmarks or broad end-to-end automation pipelines, while biological systems remain comparatively underexplored. Here, we introduce the MEDA system, an LLM- and SR-powered agentic framework for discovering ordinary-differential-equation (ODE) models of biological and biologically inspired dynamical systems. MEDA retrieves background knowledge, defines admissible variables, generates mechanistic constraints, proposes candidate ODEs, and fits and evaluates them. We evaluate it across canonical model retrieval, reasoning-based extrapolation to unseen variants, and open-ended discovery, with and without experimental data. Across these settings, MEDA recovered the correct state variables, achieved strong structural recovery in retrieval and extrapolation tasks, and produced biologically plausible discovery-oriented models. Ablation and robustness analyses show that knowledge-guided formalization and mechanistic constraints are load-bearing components, whereas numerical fitting alone can preserve trajectory-compatible but biologically incorrect equations.
arXiv:2607.13155v1 Announce Type: new
Abstract: Generative molecular models can support early drug discovery by proposing new candidate compounds de novo. In practice, useful candidates must balance target-relevant activity, synthetic accessibility, physicochemical properties, and other multiparameter design constraints. However, metrics commonly used to evaluate molecular generators only weakly reflect whether the generated compounds are medicinally plausible and suitable for downstream computation. This can produce false positives in model evaluation, incorrect assumptions, and inefficient use of computational resources. We introduce HEDGEHOG, a unified six-stage filtration benchmark that is inspired by industrial hit identification workflows: (i) preprocessing; (ii) physicochemical descriptor screening; (iii) structural alerts and graph-sanity checks; (iv) synthesis feasibility; (v) docking and binding affinity estimation; and (vi) three-dimensional pose and interaction checks. We evaluate 23 molecular generators across three model classes under a standardized protocol. Across 230,000 generated molecules, only 0.65% of initial molecules survive all stages. Our results expose a central limitation of current molecular generators: molecules that appear acceptable under isolated criteria rarely satisfy medicinal chemistry, synthesis, docking, and 3D pose filters simultaneously.
arXiv:2607.13048v1 Announce Type: new
Abstract: Streaming inference pipelines increasingly pair lightweight fast models with Large Language Models (LLMs) that provide rich semantic understanding at substantial cost. The central question of when to invoke the LLM has received limited formal treatment. We cast this as a risk-based sequential stopping problem, where a trigger policy fires when a risk functional over the observation history exceeds a threshold. Within this framework, we prove six results: a minimum inter-event time bound excluding trigger chattering; optimality of threshold policies via smooth pasting; approximate SPRT guarantees under estimated parameters; O(sqrt(T log T)) regret for stationary streams, extending to O(sqrt((C_T + 1) T log T)) under C_T changepoints; O(1/sqrt(T)) convergence of online gradient descent for adaptive thresholds; and a calibration-to-miss-rate transfer inequality. Several classical trigger families, including event-triggered, optimal stopping, SPRT, CUSUM, and Bayesian triggers, can be expressed as special cases of this framework. On turbofan degradation data (CMAPSS) with real LLM calls, we empirically verify the theoretical assumptions, ablate the risk function design, compare against six baselines including a RouteLLM-style router and contextual bandits, and analyze cost sensitivity and LLM failure modes. The results confirm sublinear regret, with alpha < 1 for all principled triggers; high diagnostic quality, with 92.9 percent of 1600 LLM diagnoses reaching grounding score >= 0.75 under our rubric; and that anomaly-score-driven risk functions dominate alternatives by roughly an order of magnitude on the Pareto AUC.
arXiv:2607.13245v1 Announce Type: new
Abstract: While 3D Scene Graphs (3DSGs) provide crucial structured representations for embodied agents, conventional Ahead-of-Time, build-everything-then-filter pipelines conflict with the real-time, low-latency demands of edge platforms, inducing a perceptual saturation effect via severe observation redundancy. To resolve this, we present JITOMA (Just-In-Time On-demand Memory Activation), a closed-loop framework that unifies task reasoning, perception, and memory into a just-in-time growth process. Instead of exhaustively mapping the entire environment, JITOMA leverages a top-down task heatmap at the frontend to filter continuous observations, routing minimal streams to maintain a global foundation of low-cost, dormant anchors. Upon a cognitive query, the backend Large Language Model (LLM) parses the robotic intent to dynamically awaken task-relevant anchors, triggering resource-intensive operations -- such as dense node captioning and functional inference -- exclusively within the activated local subgraph. To evaluate these dynamic capabilities and study perceptual saturation trade-offs, we introduce JITOMA-Bench, a comprehensive suite for long-horizon multi-tasking and complex multi-step reasoning. Extensive experiments demonstrate that JITOMA substantially reduces active graph size and captioning latency, while maintaining stable processing time under long-horizon task switching.
arXiv:2607.13437v1 Announce Type: new
Abstract: Recent vision models such as CLIP and SAM enable training-free segmentation and semantic encoding for fine-grained classification. A common approach is to compare the representations of segmented image regions with the text prompt embeddings of the corresponding labels. However, it remains unclear how different local regions and CLIP-based scoring strategies affect the selection of discriminative evidence, especially when ground-truth labels are unavailable. In this paper, we propose a unified CLIP-guided label-free region scoring framework for fine-grained classification. The framework evaluates cosine similarity-based, margin-based, and entropy-based scoring strategies using both SAM-generated masks and random crops, and introduces two label-free pseudo-label variants based on global image embeddings and local region embeddings. We conduct experiments on five fine-grained classification datasets to systematically compare different region generation methods and scoring strategies. The results show that Soft Negative Margin scoring achieves the strongest performance, and pseudo-label scoring closely approximates true-label performance. Although SAM produces semantically meaningful masks, random-crop-based pseudo-label scoring consistently outperforms SAM-based scoring across all datasets, suggesting that random crops preserve surrounding information and provide more stable semantic context when pseudo-labels are noisy. In addition, SAM masks benefit from aggregating embeddings from all regions, whereas random crops tend to perform better with a smaller top-k subset. These findings provide new insights for fine-grained classification.
arXiv:2607.13068v1 Announce Type: new
Abstract: Every mainstream GPU is built compute-heavy and capacity-light: it pairs enormous arithmetic throughput with too little memory to hold a modern model. In contrast, large language model decoding requires little compute and a large amount of memory: a GPU's floating-point units run at single-digit-percent utilization during decoding, and the memory the workload does need is sold only bundled with yet more compute. The compute is recovered only at hyperscale, where Mixture-of-Experts (MoE) models are spread across 96--320-GPU expert-parallel clusters serving thousands of concurrent users, a scale available to a handful of operators. We formalize the inefficiency with two fixed per-chip constants. F/B, the roofline ridge point, determines whether the compute can be utilized; F/S, the compute bundled with each GB of memory, determines how much compute must be bought. We then argue for a rebalanced decode accelerator: less compute, far more commodity memory, and a deliberately lower and cheaper bandwidth. The Skymizer HTX-301, a purpose-built 28nm PCIe accelerator using commodity DDR5, occupies that design point. Its entry cost is low. A single eight-chip card holds DeepSeek-R1 671B for about \$19,000, and a 4U server of four four-chip cards serves two users at a deterministic 20.3 tokens per second each for about \$28,000. Either costs less than a single H100, while the minimum GPU deployment for the model is an eight-GPU node near \$350,000. Concurrency then scales out by adding hardware: eight 4U servers carry sixteen users for about \$224,000, two-thirds of the node's price, with the cost per token unchanged at about \$12 per million against the node's \$21. The HTX-301's decisive advantage is a supply chain free of every rationed input: it uses no high-bandwidth memory, no CoWoS, and no leading-edge logic.
arXiv:2607.13438v1 Announce Type: new
Abstract: Regional heterogeneity in social characteristics, temperature change, and vulnerability to climate impacts is likely to influence the magnitude of anthropogenic climate change, but has not been considered in coupled social-climate models, which seek to represent interactions between social and climate dynamics. Here, we examine how the projected mean global temperature anomaly and population support for mitigation respond to heterogeneity in these factors across five regions of the world, using a coupled social-climate model. We find that heterogeneity in climate impacts increases the temperature anomaly by 0.2$^\circ$C, while social heterogeneity increases it by an additional 0.1$^\circ$C. The projected temperature anomaly also increases with higher variability in climate impacts across regions, even for the same average global climate impact. Finally, we identify a social-climate tipping point, where low vulnerability to impacts under existing social conditions in one region can tip the system into an alternative slow-mitigation, high-temperature state. Our results show that heterogeneity in climate impacts leads to higher global mean temperatures and efforts to reduce global disparities could improve both social and climate outcomes.
arXiv:2607.13439v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly applied to vulnerability detection due to their strong code comprehension capabilities, but most existing approaches rely on isolated functions or context extracted by fixed program-analysis rules. These methods cannot adaptively explore repository-level dependencies to gather sufficient context when vulnerabilities span multiple functions or files, compromising detection reliability. We present DREA (Decoupled Reasoning and Exploration Agents), a hypothesis-driven framework for repository-level vulnerability detection. DREA decouples reasoning from exploration through two collaborating agents: a planning agent backed by an advanced LLM that forms vulnerability hypotheses and directs the investigation, and an explorer agent powered by a lightweight model that retrieves repository-level context on demand. Goal-directed context acquisition is the primary source of detection improvement in this design, while offloading token-heavy exploration to the local model keeps inference economically tractable. To support evaluation, we construct RepoPairBench, a repository-grounded benchmark of validated Python vulnerability-fix pairs from real-world projects. Beyond binary detection accuracy, we introduce a reasoning correctness evaluation to assess whether a model's rationale matches the documented vulnerability mechanism. Across three LLMs, DREA improves Pair-Correctness from 19-26% to 30-42% while offloading over 93% of tokens to the explorer, reducing estimated billable API cost by a factor of 16-48. Reasoning correctness analysis further reveals that 26-55% of true positives, for both DREA and the function-only baseline, are correct predictions supported by flawed rationales, identifying security reasoning quality as a shared bottleneck for current LLMs.
arXiv:2607.13440v1 Announce Type: new
Abstract: Moving Target Defense (MTD) assumes its pool of candidate configurations is safe to cycle among, i.e. latency and other observables do not trivially fingerprint the active choice, but this assumption has not been quantified at the pool level. We formalize this pool-safety problem as finding the largest $\varepsilon$-close subset of the Cartesian product of per-component implementation choices, reducing pairwise indistinguishability under an additive utility model to a densest-window query over a sum-set. We give four algorithms spanning the scalability spectrum -- full enumeration, meet-in-the-middle, FFT convolution, and Monte Carlo sampling -- covering configuration spaces from tens to $10^{38}$. We then measure the anonymity gap end-to-end on two production cloud case studies, and find that a component's latency differences do not survive deployment unchanged: a four-runtime serverless rotation, where nothing else masks the interpreter, collapses from four-way to three-way anonymity against a VPC-adjacent adversary, while a $27$-configuration three-tier stack, where the same interpreter differences are instead absorbed by a shared $8$~ms database round-trip, delivers nine-way effective anonymity. The framework and the two case studies together suggest a diagnostic for MTD design: rotating a component adds anonymity only if the latency differences among its variants are too small for the adversary to identify.
arXiv:2605.16281v2 Announce Type: replace
Abstract: Post-deployment accountability has become central to AI governance, yet little empirical evidence shows whether monitoring, incident reporting, and impact assessment obligations are visible when AI systems fail. This study analyzes real-world AI incidents from the AI Incident Database (2020--2026) and codes them against nine post-deployment provisions from the EU AI Act, the NIST AI Risk Management Framework, and the GDPR. The findings show substantial accountability gaps: 77.1\% of incidents lack evidence of EU AI Act post-market monitoring, and 99.6\% lack documented Data-Protection Impact Assessment evidence. Governance gaps are also systemic, with 9.8\% of incidents simultaneously non-compliant under two or more regimes. Incidents detected through internal monitoring show much higher compliance than externally detected incidents (87.5\% vs 5.3\% under the EU AI Act; 95.8\% vs 58.1\% under NIST), suggesting that monitoring capacity is a key condition for effective post-deployment governance. Building on these findings, the paper proposes the Proactive AI Governance Compliance Framework (PAGCF), a four-phase lifecycle for pre-deployment assessment, continuous monitoring, incident preparedness, and cross-framework verification.
arXiv:2607.13234v1 Announce Type: new
Abstract: Deepfake detectors that achieve near-perfect scores on academic benchmarks collapse on real-world content: recent in-the-wild evaluations report AUC drops of 45-50% for state-of-the-art open-source models. We argue this gap is structural: static detectors are trained once against a moving generative frontier. We present BitMind Forensics (BMF), trained through Bittensor SN34, an open adversarial competition that continually refreshes the training distribution. We evaluate one dated export comprising image, general-video, and human-video checkpoints across nineteen public datasets: the canonical face-swap suites (FaceForensics++, Celeb-DF v1/v2/++, DFDC, DFD, UADFV, DF40) and recent in-the-wild and AI-generated-media benchmarks (Sumsub, Deepfake-Eval-2024, WildRF, Community Forensics, AIGCDetectBench, GenImage, AI-GenBench, AIGIBench, RAID, GenVidBench, GenVideo-100K). BMF reaches 0.936 AUC on Sumsub's original images and 0.872 pooled AUC over its full four-condition manipulation battery (1.4M images), staying robust under perturbation (0.855 JPEG, 0.799 downscaled), while GPEN enhancement improves detection (0.996). On Deepfake-Eval-2024, it matches the best commercial detector on images (0.915 vs 0.90) and exceeds it on video (0.822 vs 0.79), far above the best open-source detectors (0.56 and 0.63). It reaches 0.991 AUC on a 21-generator AI-image panel and 0.918 on GenVidBench, and exceeds the FF++-trained frontier on DFDC (0.947 vs 0.843) and Celeb-DF v2 (0.9985 vs 0.956), both contamination-audited, with statistical parity on Celeb-DF++. In a temporal study, successive dated exports improve on held-out media from generators absent from the static baseline's training (image 0.842 to 0.902; video 0.864 to 0.936). Our evaluation harness is public, and at publication the production API serves the exact evaluated snapshot for independent verification.
arXiv:2607.13749v1 Announce Type: new
Abstract: Neural networks trained on modular arithmetic exhibit grokking, a delayed transition from memorisation to generalisation known to depend on model capacity: too little and the network memorises slowly or not at all, too much and it generalises almost immediately. What happens at the extreme of this spectrum, when the architecture's expressible function class collapses to a finite-dimensional algebraic variety? We study two-layer networks with a holomorphic monomial activation sigma(z)=z^k, trained on modular tasks encoded via roots of unity. Here the network output, regardless of hidden width, is confined to a (k+1)-dimensional subspace of characters of (Z_p)^2, an O(k/p^2) slice of the full function space. We give a complete algebraic characterisation of this subspace: a task is representable if and only if its discrete Fourier support lies on the diagonal u+v = k (mod p), which for linear-phase targets reduces to the arithmetic criterion m+n=k. This is not merely a constraint on eventual generalisation but on memorisation itself: because the outputs are algebraically confined, a non-representable target cannot be fit even on the training set, and we prove a positive lower bound on the training loss, independent of width. Across 585 runs the algebraic prediction matches the observed outcome with 99.8% accuracy, with no memorisation regime and no grokking; outcomes split cleanly into instant success and outright failure. This binary behaviour is the limiting case of the capacity-grokking relationship: when the expressible class shrinks to a fixed algebraic object, the question of when a network will grok dissolves into whether it can represent the target at all. A bottleneck ablation connects this extreme to standard networks, tracing a continuous path from representational failure, through memorisation without generalisation, to grokking with a shrinking gap as capacity grows.
arXiv:2607.13248v1 Announce Type: new
Abstract: The evaluation of mathematical reasoning in large language models (LLMs) has predominantly focused on high-resource languages like English. This has created a significant barrier to the equitable development and deployment of AI in linguistically diverse regions such as Bangladesh, where over 230 million people speak Bengali. Despite this global significance, there has been minimal prior work on mathematical reasoning in Bengali and no existing research that systematically benchmarks a perturbated Bengali mathematical dataset, leaving a critical void in assessing model robustness and true comprehension beyond pattern recognition. This study addresses this gap by introducing GSM-Plus-BN, a novel perturbated Bengali mathematical dataset derived from the English GSM-Plus benchmark and verified by human translators. We evaluate six open-source LLMs Qwen3-32B, Llama-3.1-8B-Instant, Llama-3.3-70B-Versatile, Llama-4-Scout-17B-16E-Instruct, GPT-OSS-120B, and GPT-OSS-20B using a benchmark of 9,000 evaluation samples comprising 1,000 seed questions and 8,000 perturbed variants under both Standard Prompting and Chain-of-Thought (CoT) Prompting. Experimental results show that GPT-OSS-20B achieves the highest seed question accuracy of 96.08% under Standard Prompting, while larger models such as Llama-3.3-70B and GPT-OSS-120B demonstrate superior robustness across perturbation types. Furthermore, CoT prompting substantially improves reasoning for most models compared to Standard Prompting, yet a notable performance gap persists across all models relative to their English benchmarks, underscoring the inherent difficulty of perturbed Bengali text. This research makes a foundational contribution by providing GSM-PLUS-BN as a new resource and baseline for future Bengali mathematical reasoning research.
arXiv:2509.10033v2 Announce Type: replace
Abstract: Sparse dictionary coding represents signals as linear combinations of a few dictionary atoms. It has been applied to images, time series, graph signals and multi-way spatio-temporal data by jointly employing temporal and spatial dictionaries. Data-agnostic analytical dictionaries, such as the discrete Fourier transform, wavelets and graph Fourier, have seen wide adoption due to efficient implementations and good practical performance. On the other hand, dictionaries learned from data offer sparser and more accurate solutions but require learning of both the dictionaries and the coding coefficients. This becomes especially challenging for multi-dictionary scenarios since encoding coefficients correspond to all atom combinations from the dictionaries. To address this challenge, we propose a low-rank coding model for 2-dictionary scenarios and study its data complexity. Namely, we establish upper and lower bounds on the number of samples needed to learn dictionaries that generalize to unseen samples from the same distribution. We propose an alternating convex optimization solution, called AODL, which employs alternating optimization between the sparse coding matrices and the learned dictionaries. We demonstrate its quality for data reconstruction and missing value imputation in both synthetic and real-world datasets. For a fixed reconstruction quality, AODL learns up to $90\%$ sparser solutions compared to non-low-rank and analytical (fixed) dictionary baselines. In addition, the learned dictionaries reveal interpretable insights into patterns from training samples.
arXiv:2603.16515v2 Announce Type: replace
Abstract: The transition between the ELMy H-mode and the EDA H-mode is studied on Alcator C-Mod using an experimental database and predictive pedestal models. High-resolution Thomson scattering measurements are used to compare the pedestal density, $n_{e}^\mathrm{ped}$, and the separatrix density, $n_{e}^\mathrm{sep}$ with main chamber neutral measurements. $n_{e}^\mathrm{ped}$ is sensitive to neutral sources only in the ELMy H-mode regime and not in the EDA H-mode regime. Density fluctuation spectra reveal that quasi-coherent structures become stronger at higher densities and more coherent in the EDA relative to the inter-ELM phases of ELMy H-modes, before weakening again at the highest values of $n_{e}^\mathrm{ped}$. The Saarelma-Connor pedestal density prediction model is validated for ELMy H-modes up to $n_{e}^\mathrm{ped} = 2.0 \times 10^{20}$ m$^{-3}$. An additional transport channel driven by resistive ballooning modes (RBM), $D_\mathrm{RBM}$, scaling directly with $\alpha_{t}$ and inversely with $k_\mathrm{RBM}^{2}\hat{q}_\mathrm{cyl}$ is shown to improve the prediction for EDA H-modes, finding good model agreement up to $n_{e}^\mathrm{ped} = 3.0 \times 10^{20}$ m$^{-3}$. EPED scans in $n_{e}^\mathrm{ped}$ are then performed at three values of $n_{e}^\mathrm{sep}/n_{e}^\mathrm{ped}$. Increasing this ratio moves the peeling-ballooning branch transition to lower $n_{e}^\mathrm{ped}$, increasing $p^\mathrm{ped}$ in the peeling branch and decreasing it in the ballooning branch. Agreement is found for large ELM H-modes. SPARC pedestal density predictions for an ELMy and an EDA/QCE-like H-mode are performed and found consistent with assumptions used in previous EPED modeling. Inclusion of $D_\mathrm{RBM}$ significantly weakens the density gradient near the separatrix, lowering $n_{e}^\mathrm{ped}$ by approximately 20%.
arXiv:2607.13614v1 Announce Type: new
Abstract: Group conversations are a fundamental yet complex form of social interaction central to human cognition and telecommunication technology. While understanding and facilitating these interactions has been a long-standing goal, findings are often isolated within specific in-person or videoconferencing settings due to a scarcity of datasets that bridge the two. We introduce VIP-MINGLE, a multimodal dataset comprising 59 hours of recordings (32 groups, 105 participants), featuring paired within-subject sessions in both settings. The dataset includes raw audio/video, psychometric data, processed multimodal features (e.g., diarized speech, facial expressions, transcriptions), and time-resolved human annotations. Our analysis reveals significant behavioral distribution shifts across multiple modalities between settings, reinforcing the need for a cross-setting corpus. VIP-MINGLE serves as a critical resource for developing robust models of group conversations across settings.
arXiv:2607.13892v1 Announce Type: new
Abstract: Computed tomography (CT) vision-language pretraining from paired volumes and radiology reports is a scalable yet challenging task. Existing methods commonly adopt global scan-report contrast, which is scalable but obscures heterogeneous organ evidence. Meanwhile, direct organ-level alignment remains coarse, since the same anatomy can exhibit multiple distinct radiological appearances. Therefore, pretraining requires a finer alignment unit: the organ-conditioned radiological pattern. In this work, we propose OCP-CT, an organ-conditioned pattern-token alignment framework for CT vision-language pretraining. Specifically, OCP-CT preserves a stable global CT-report contrastive branch and introduces an organ pattern interface: sparse Mixture-of-Experts (MoE) routes image and text tokens according to latent radiological patterns, learnable slots query the routed tokens into continuous pattern tokens, and paired token contrast aligns image-text pattern tokens with structured soft targets built from report-derived clinical similarity. On the publicly available CT-RATE and RAD-ChestCT benchmarks, OCP-CT achieves average AUROCs of 84.5% and 69.9% for zero-shot abnormality diagnosis, respectively. Compared with the strongest prior reported results, these results yield absolute AUROC gains of 6.7 and 0.8 percentage points.
arXiv:2607.13899v1 Announce Type: new
Abstract: We propose the AIMO Interpretability Challenge, a competition on distinguishing robust from spurious reasoning in frontier mathematical language models based on the models' internal mechanisms. The challenge is motivated by a central limitation of standard reasoning benchmarks: strong final-answer accuracy does not reveal whether a model relies on stable reasoning mechanisms or exploits brittle reasoning shortcuts. Building on AI Mathematical Olympiad (AIMO) problems and submissions, together with resources from the Fields Model Initiative, the competition will provide (1) newly-published olympiad-level math reasoning problems and their symbolic representations, allowing generation of novel functional variants, (2) access to frontier reasoning models, and (3) assessments of models' adversarial robustness on these problems. Participants will use these resources, along with our computing infrastructure support, to develop methods for identifying which models solve problems robustly. Our competition will also create a new, open robustness benchmark and baseline systems, aiming to provide a lasting foundation for standard benchmarking in mathematical reasoning and interpretability. Scientifically, the competition connects interpretability and generalization research around a central question in AI research: can we determine if, and to what extent, the decision-making of frontier AI models is generalizable and thus, reliable?
arXiv:2607.13901v1 Announce Type: new
Abstract: Critical thinking is a fundamental skill that helps learners move beyond simple memorization. One way to develop this skill is through high-order questioning. However, crafting such questions remains a challenge for educators, and classroom practices tend to rely on low-order questions. Large Language Models have demonstrated strong capabilities in generating high-order questions, especially when guided by prompts based on Bloom's Taxonomy. Yet, existing research has largely centered on this framework and focused only on English. This study addresses these gaps by introducing prompts grounded in two alternative frameworks: Claim-Evidence-Reasoning and Divergent Questioning within a multilingual context using Basque, Spanish, and English. Results indicate that while both an open-source and a proprietary model rather effectively generate questions in all three languages, only about half of the answerable questions are recognized by teachers as high-order. A positive finding is that the alternative frameworks produce structurally and conceptually varied questions, suggesting they could complement each other and provide viable alternatives to Bloom's Taxonomy.
arXiv:2607.13172v1 Announce Type: new
Abstract: We address the problem of safely training an agent policy and deploying a good and safe policy, in settings where the environment dynamics are unknown and no suitable reward function is available. In the context of safety-critical environments, we consider traditional reinforcement learning impractical and resort to the resource of human input. We introduce DROPJ, a human-centred method for both safe training and deployment. We first learn a world model (a learned simulator) from a dataset of prior real-world trajectories. A human then plays the game in this learned simulator to extract several informative simulated trajectories. From these, we sample pairs of simulated trajectory segments and elicit from a human their preference over these segments, as well as a reason (justification) for their choice. We then train a reward model from these justified preferences and use it, together with the world model, to directly deploy the agent using model predictive control. Running real-user experiments, we find that generating informative simulated trajectories from a user significantly reduces the computational cost during training compared to other strategies, and can also improve the performance during deployment. In the context of training within a learned simulator, we show that the use of preferences rather than other types of feedback substantially improves the performance during deployment. We further demonstrate that safety justifications accompanying preferences can significantly enhance safety or prioritise user-prescribed aspects of safety associated with them during deployment.
arXiv:2607.13253v1 Announce Type: new
Abstract: In this paper, we develop a linear, fully decoupled, and unconditionally energy-stable fully discrete finite element scheme for the Cahn--Hilliard--Navier--Stokes (CHNS) system by employing the scalar auxiliary variable (SAV) approach. Unlike existing SAV-based formulations that typically introduce multiple auxiliary variables or additional techniques to handle different nonlinearities, we introduce only one scalar auxiliary variable together with a novel update of the auxiliary variable to reformulate all nonlinear terms arising from the Cahn--Hilliard and Navier--Stokes equations, yielding an equivalent reformulation of the original CHNS system. An implicit--explicit (IMEX) Euler scheme is applied for temporal discretization, where the linear terms are treated implicitly and the nonlinear terms explicitly, while a finite element method is adopted for spatial discretization. The resulting fully discrete scheme can be efficiently decomposed into two linear subproblems and one scalar quadratic algebraic equation, which significantly simplifies the implementation. Furthermore, we prove that the proposed scheme satisfies an unconditional discrete energy dissipation law and establish its stability with respect to several relevant norms. Optimal-order $L^2$ error estimates are also derived for the fully discrete finite element approximation. Finally, a series of numerical experiments are presented to verify the theoretical results and demonstrate the efficiency of the proposed method.
arXiv:2607.13764v1 Announce Type: new
Abstract: Multimodal optimization aims to locate multiple globally optimal or near-optimal solutions in a single run. This paper presents \emph{S-CARD-CMSA}, a score-aware candidate-archive and density-filtered reporting framework built on the covariance matrix self-adaptation evolution strategy with repelling subpopulations (RS-CMSA-ESII). The method is developed for the IEEE CEC 2026 Competition on Benchmarking Niching Methods for Multimodal Optimization. Rather than modifying the core search dynamics of RS-CMSA-ESII, S-CARD-CMSA preserves its sampling, covariance adaptation, taboo-region update, restart, and termination mechanisms. Two conservative extensions are introduced. First, a passive secondary candidate archive records the restart-level best candidates without influencing the search trajectory. Second, a score-aware density-filtered reporting rule constructs the final solution set by balancing robust peak ratio and precision-driven F1-score. Development experiments show that the density-filtered rule preserves the peak coverage obtained by a medium score-aware rule while reducing redundant reports. On a broader validation subset, it maintains the same mean RPR while improving mean precision, F1-score, and the official-score-oriented average. The method does not use true global-minimum locations during optimization; such information is used only for offline development analysis and post-run scoring. The source code of S-CARD-CMSA is available at https://github.com/ChauhanDikshit.
arXiv:2607.13767v1 Announce Type: new
Abstract: Multivariate graphs unite two distinct data perspectives: a topological structure defined by nodes and edges, and attribute data associated with each node. Analyzing such graphs therefore requires reasoning across two complementary spaces. However, existing systems typically emphasize the analysis of one space at a time, focusing either on topology or on attributes. As a result, exploration, analysis, and pattern discovery that depend on their interaction remain difficult. In this paper, we present ZipLine, a system designed to support integrative analysis of multivariate graphs by bridging both topology and attribute spaces. ZipLine introduces a predicate language that enables analysts to express patterns involving topology, node attributes, and neighborhood relations with a unified formalism. The system further provides a predicate-learning algorithm that maps analyst interactions across both topology (e.g., subgraph selection) and attribute views (e.g., value brushing), into the predicate language, enabling learned expressions that bridge the two spaces. This integrative approach supports iterative analysis by enabling analysts to refine patterns through coordinated reasoning over topology and attributes. We demonstrate ZipLine through three case studies in energy infrastructure, cybersecurity, and drug discovery analysis. The results show that ZipLine enables expressive multivariate graph analysis through unified reasoning across topology and attributes.