arXiv:2606.19496v1 Announce Type: new Abstract: Generative models can produce individually plausible samples while deviating substantially from a target set in the distribution of key features. For example, a model pretrained on broad drug-like chemical space may generate molecules whose molecular features differ from those of a therapeutic class of interest, such as known antibiotics. Correcting such distributional miscalibration is challenging: direct finetuning on the target set can overfit and does not control which features are matched. To fill this gap, we introduce kernel Calibrating Generative Models (kCGM). kCGM minimizes a maximum mean discrepancy (MMD) between generated and target feature distributions using an unbiased score-function estimator, with KL regularization to remain close to the pretrained model. On a target set of 174 antibiotics, direct finetuning sacrifices chemical validity for feature-distribution matching, whereas kCGM improves target feature matching while increasing validity. We further demonstrate kCGM in protein and DNA generation tasks, showing it can adapt autoregressive, continuous-space diffusion, and discrete diffusion models using only feature-level supervision. Code is available at https://github.com/smithhenryd/cgm.
Science Journals
arXiv:2606.19499v1 Announce Type: new Abstract: Tropes are recurring narrative devices in television and film. We carry out a computational analysis of tropes in the sitcom Friends, using human-curated trope annotations from TVTropes, episode transcripts, and IMDb ratings. Because automatic trope detection remains challenging, we treat existing trope annotations as a curated analytical layer and focus on their downstream narrative and semantic functions. We first examine the relationship between episode-level trope frequency and audience reception. We find a statistically significant positive association between trope count and weighted IMDb ratings, although the modest explanatory power suggests that more than trope density alone explains audience evaluation. We then connect trope annotations to dialogue transcripts and represent trope-related dialogue using TF-IDF-based semantic features. Using PCA and k-means clustering, we group 1,954 distinct tropes into 15 semantically interpretable clusters. Chi-square analyses show that the six main characters are unevenly distributed across these clusters, with character-specific trope profiles that are broadly consistent with their established narrative identities. Finally, we project trope clusters into the ousiometric power-danger space to examine their semantic organization. The results show that "Physical and Sexual Comedy" occupies a region associated with relatively high danger, while "Revelation, Surprise, and Reaction" occupies a region associated with relatively high power. Overall, our work demonstrates a way to operationalize trope measurement and shows that identifiable trope clusters can provide holistic "distant reading" descriptions of characters and stories.
arXiv:2606.19949v1 Announce Type: new Abstract: Skeletonization of tubular structures from 3D imaging is essential for tasks such as morphometric analysis, transport or flow simulation, and procedural planning in domains including vascular networks, plant root systems, and neural connectomes. However, automatic skeleton extraction often introduces topological artifacts, such as erroneous connections between nearby branches and fragmented centerlines caused by noise or missing data. Correcting these artifacts manually can be time-consuming and error-prone, especially when precise interaction is required. We present a semi-automatic correction method that reconstructs a plausible centerline connection from minimal user input. Given a user-selected source and target point, our method traces a path by combining (i) component-wise minimum spanning trees for stable local propagation and (ii) a filtered 3D Delaunay edge graph for bridging gaps and handling ambiguous junctions. Candidate steps are ranked using a score that accounts for direction continuity, spatial proximity, component consistency, and target-directed progress. The output is an ordered polyline (or edge sequence) that can be used as a suggested correction and integrated into downstream skeleton post-processing workflows. We implement the system in C++ with an interactive viewer based on Libigl and demonstrate representative qualitative results on brain vessel datasets, including correction of typical "crossing" and "dotted" artifacts. While our current validation is qualitative, the method is lightweight and serves as a practical building block toward more comprehensive interactive correction pipelines in biomedical imaging and related domains.
arXiv:2606.20216v1 Announce Type: new Abstract: Machine learning algorithms deployed for evolving streaming environments must handle the non-stationary data distributions, commonly referred to as concept drift. The presence of concept drift poses a major challenge for many real-world applications because it can severely degrade their predictive performance, hindering their ability to support robust decision-making. Consequently, the timely and efficient detection of drift events is critical for sustaining high accuracy over time. This study examines theoretically the concept drift characteristics and numerous drift detection algorithms across several categories. Furthermore, we evaluate their performance on both synthetic and real-world datasets exhibiting diverse streaming scenarios and drift characteristics, such as abrupt and gradual changes. This study aims to enhance understanding of the complex notion of concept drift characteristics and behavior of drift detectors, along with their applicability to diverse contexts.
arXiv:2606.20223v1 Announce Type: new Abstract: Camera-trap monitoring in African tropical forests increasingly extends beyond closed-canopy interiors to riverbanks, clearings, and park edges. Among available open tools for African forest camera-trap classification, DeepForestVision is the only one providing a matched offline workflow for both photographs and videos, and previous work showed that it outperformed other available baselines on a comparable benchmark. However, it was designed for closed-canopy, ground-level forest interiors and uses a 35-class prediction space that becomes too coarse when deployments encounter arboreal primates, birds, semi-aquatic taxa, or human-associated confounders such as livestock. We present DeepForestVisionV2, an ecology-driven expansion from 35 to 64 prediction classes (61 animal classes plus human, vehicle, and blank) designed to address three recurrent deployment gradients: vertical stratification, scene openness, and anthropogenic interfaces. DeepForestVisionV2 retains the same offline workflow and is trained on 1,535,010 photographs and 243,354 videos from multi-country African tropical-forest projects. Evaluation combines a cross-country cropped-photo validation set, used to assess robustness across sites and camera-trap settings, with three held-out Uganda video benchmarks spanning the targeted gradients. On the validation set, DeepForestVisionV2 reaches 0.86 accuracy, 0.82 macro-F1, and 0.81 balanced accuracy. On the deployment benchmarks, it preserves or improves baseline accuracy despite its harder classification task, while increasing the number of identified taxa from 22 to 29 in forest-interior videos and from 4 to 9 at riverbanks. In the park-edge use case, it raises accuracy from 0.62 to 0.86 and reduces false alarms from 11 to 0. These results show that DeepForestVisionV2 materially improves field utility while preserving robustness across sites, habitats, and camera-trap settings.
arXiv:2606.20231v1 Announce Type: new Abstract: Can intelligence be measured? We propose that intelligence can be defined as the lawful amplification of rare but valid futures: a system increases the probability of outcomes that would be unlikely under passive dynamics but remain admissible under the constraints of the domain. We start with the premise that an intelligent system must model the world and its own place within it. Because the system is part of the world it models, this leads naturally to recursive self-simulation: the system represents futures in which its own actions are part of the trajectory. Our central results give a necessity statement and a conditional near-sufficiency statement connecting this architecture to a precise thermodynamic measure of lawful amplification of rare-valid futures: high rare-valid lift is impossible unless the internal simulation identifies rare-valid futures with high fidelity; conversely, when rare-valid fidelity is high and the simulation contains an effective policy, the achievable lift approaches the actuation-limited optimum. Thus recursive self-simulation is not merely a plausible feature of intelligence but, under the stated assumptions, is necessary and nearly sufficient for high thermodynamic intelligence. The resulting framework makes intelligence measurable on a universal scale, from passive matter and feedback controllers, large language models, and humans as text generators to Maxwell-demon-like information engines.
arXiv:2606.20232v1 Announce Type: new Abstract: This paper investigates mobile target search under imperfect perceptions caused by sensor limitations, malicious jamming, or communication noise. Searchers and targets operate in a grid-shaped area with bounded mobility, leading to a dynamic interplay between search and evasion. To capture this adversarial interaction under imperfect perceptions, we adopt the partially observable stochastic game (POSG) approach, which generalizes partially observable Markov decision processes (POMDPs) by incorporating target intelligence. To handle false alarms and missed detections caused by perceptual uncertainties, we propose a novel detectability concept to determine whether a search strategy guarantees eventual detection, and provide sufficient detectability criteria based on stochastic recurrence analysis. We further develop a server-assisted distributed algorithm that utilizes the aggregative potential game structure for searchers and a KL-divergence-based reduction for target prediction. Numerical simulations validate the effectiveness of the proposed algorithm and support the detectability analysis.
arXiv:2606.20198v1 Announce Type: new Abstract: We present an algorithm for pitch spelling and key estimation. Given an input in MIDI-like format, containing information on note pitches (expressed in semitones relative to the lowest reference note) and bar boundaries, it estimates the appropriate note names, a global Key Signature, and a local scale for each bar. This related information elements are evaluated jointly during two stages of optimisation. During an initial 'modal' stage, a probable scale is proposed for each bar, minimising the number of accidentals to be printed in the printed score with a shortest-path search. Then, during a second stage called 'tonal', these local scales are used to estimate the Key Signature and note names that would result in the best musical notation for the entire piece. We present evaluations conducted on datasets comprising a variety of digital musical scores: jazz lead sheets taken from the Real Book, transcriptions of recordings of jazz soli and bass lines, traditional tunes, as well as classical scores for piano and monophonic instruments. Our procedure was originally designed for use in music transcription, specifically for building digital collections of jazz solos transcribed from audio recordings, for the purposes of music analysis, teaching and the preservation of cultural heritage. This method should also prove useful for other tasks related to the processing of musical notation. Furthermore, to this end, we have defined new distances between various common jazz scales, which may be of some interest to musicological studies.
arXiv:2606.19521v1 Announce Type: new Abstract: In multi-task learning, handling an increasing number of objectives can quickly become challenging, both in terms of the computational resources and the decision maker's capacity to choose appropriate trade-offs. A widely used approach is thus to aggregate the individual losses in a single loss function by a weighted sum. This often fails to capture either the decision maker's preferences as a result of the shape of the Pareto front, or requires multiple adjustments and computations which becomes prohibitively expensive in deep learning applications. To address these issues, we introduce a novel framework, Preference Pareto Exploration (PPE), which enforces the decision maker's preferences while accounting for the geometry of the Pareto set in an interactive exploration process. PPE is based on a predictor-corrector method that performs predictor steps tangential to the manifold of Pareto-optimal solutions, following the decision maker's preference. The subsequent corrector step results in a new trade-off reflecting this preference. To avoid explicit Hessian computations when characterizing the tangent space of the manifold, we employ a Krylov subspace method that relies solely on matrix-vector products. These products can be efficiently obtained via automatic differentiation, ensuring both efficiency and robustness throughout the optimization process. The method's functionality and performance are demonstrated using both toy problems and examples from deep learning.
arXiv:2604.05435v2 Announce Type: replace Abstract: Incomplete or inconsistent discharge documentation drives care fragmentation and avoidable readmissions. Despite its critical role in patient safety, auditing discharge summaries relies on manual review and does not scale. We propose an automated framework for auditing discharge summaries using large language models (LLMs). Our approach operationalizes the DISCHARGED framework into a checklist of 46 questions. Using 50 summaries from the MIMIC-IV database, with clinician ground-truth labels, we benchmark 11 LLMs. Model-assessed mean documentation completeness ranges from 54.9% to 74.2%, and the best-performing models achieve a Cohen's kappa values around 0.5 against clinician labels, indicating moderate agreement. All models struggle to identify ambiguous documentation (Unclear), highlighting a key gap in current automated auditing. This work provides a clinician-validated benchmark and zero-shot baselines for systematic quality improvement in clinical documentation.
arXiv:2604.07328v3 Announce Type: replace Abstract: How does the choice of training data influence an AI model? This broad question is of central importance to interpretability, privacy, and basic science. At its technical core is the data deletion problem: after a reasonable amount of precomputation, quickly predict how the model would behave in a given situation if a given subset of training data had been excluded from the learning algorithm. We present a data deletion scheme capable of predicting model outputs with vanishing error $\varepsilon$ and failure probability $\delta$ in the deep learning setting. Our precomputation and prediction algorithms are only $\tilde{O}(\log(1/\delta)/\varepsilon^2)$ factors slower than regular training and inference, respectively. The storage requirements are those of $\tilde{O}(\log(1/\delta)/\varepsilon^2)$ models. Our proof is based on an assumption that we call stability. In contrast to the assumptions made by prior work, stability appears to be fully compatible with learning powerful AI models. In support of this, we show that stability is satisfied in a minimal set of experiments with microgpt. Our code is available at https://github.com/SamSpo1/microgpt-sketch. At a technical level, our work is based on a new method for locally sketching an arithmetic circuit by computing higher-order derivatives in random complex directions. Forward-mode automatic differentiation allows cheap computation of these derivatives.
arXiv:2605.09383v2 Announce Type: replace Abstract: In safety-critical scenarios, the protection level of the autonomous navigation system is crucial for enabling mobile robots to perform safe tasks. However, existing studies on probabilistic navigation systems for robots usually perform offline accuracy evaluations using limited datasets and assume that the results can be applied to unknown real-world environments. As a result, current autonomous mobile robots often lack protection levels for online safety assessment. To fill this gap, we propose a safety-critical LiDAR-inertial odometry (LIO) that provides deterministic protection levels based on on-manifold deterministic state estimation. By adopting the unknown but bounded assumption, we derive a neat closed-form relationship between point cloud noise and the uncertainty of the estimation from the iterated closest point algorithm. Using this relationship, we design an on-manifold ellipsoidal set-membership filter and implement it within the LIO system. Leveraging the properties of the set-membership filter, our system offers the feasible sets of the estimated locations as the deterministic protection levels, serving as safety references for the robots' downstream autonomous operations. The experimental results show that our system can provide effective deterministic online safety references for diverse robots in various environments.
arXiv:2605.09550v2 Announce Type: replace Abstract: Artificial intelligence is increasingly entering digital games through diverse functions. While prior work has shown that player attitudes toward game AI are strongly context-dependent, less is known about how these attitudes are structurally combined within different groups of players. This study addresses this gap by modeling players' cross-context AI acceptance as interpretable attitude profiles. Based on questionnaire data from 771 digital game players, we apply Archetypal Analysis (AA) to centered acceptance ratings across eight representative AI application contexts in games. The analysis identifies seven distinctive profiles: AI-Skeptics, Broad AI-Supporters, Creative-Play Explorers, Experience-Oriented Supporters, Systemic Order Advocates, Emotion-Centered Supporters, and Governance-Skeptics. Exploratory one-vs-rest (OvR) logistic regressions further suggest that profile membership is associated with players' perceived AI literacy, gaming habits, disciplinary background, personality traits, and application-specific priorities. By shifting attention from isolated acceptance judgments to patterned preference structures, this study provides an exploratory empirical vocabulary for segmenting game AI audiences and offers preliminary design implications for more context-sensitive and player-sensitive AI integration in digital games.
arXiv:2606.20068v1 Announce Type: new Abstract: While reinforcement learning from verifiable rewards (RLVR) typically has relied on a single binary verification signal, symbolic proof assistants in formal reasoning offer rich, fine-grained structured feedback. This gap between structured processes and unstructured rewards highlights the importance of feedback that is both dense and sound. In this work, we demonstrate that the Lean proof assistant itself can serve as a symbolic process oracle, supplying both outcome-level and fine-grained tactic-level verified feedback during training. Proof attempts are parsed into tactic sequences, and Lean's elaboration marks both locally sound steps and the earliest failing step, yielding dense, verifier-grounded credit signals rooted in type theory. We incorporate these structured rewards into a GRPO-style reinforcement learning objective with first-error propagation and first-token credit methods that balances outcome- and process-level advantages. Experiments with STP-Lean and DeepSeek-Prover-V1.5 show that tactic-level supervision outperforms outcome-only baselines in most settings, delivering improvements on benchmarks such as MiniF2F and ProofNet. Beyond empirical gains, our study highlights a broader perspective: symbolic proof assistants are not only verifiers at evaluation time, but can also act as process-level reward oracles during training. This opens a path toward reinforcement learning frameworks that combine the scalability of language models with the reliability of symbolic verification for formal reasoning.
arXiv:2606.20081v1 Announce Type: new Abstract: We present a theoretical study of Cherenkov and transition radiation for swift electron beams penetrating dielectric nanospheres using material models of different sophistication. Specifically, we perform a combined time-domain (numerically, via the discontinuous Galerkin time-domain method) and frequency-domain (numerically and analytically, via Mie-based theory) study, including the induced-field distribution, cathodoluminescence (CL) multipole/directional decomposition, as well as the time-dependent angular power flow. For low velocities below the Cherenkov threshold, we show that transition radiation is dominant in the far-field CL, and the near-fields at the transition points are primarily responsible for the main features observed in the far-field. For higher velocities far beyond the Cherenkov threshold, we identify the fingerprints of the observable Cherenkov front. Specifically, a constant-permittivity model allows us to isolate the respective contributions of CR and TR to the far-field radiation, thereby facilitating the interpretation of the results for a more realistic material model that includes material resonances. Our combined time- and frequency-domain framework provides a direct view of radiative excitation channels for swift electron beams penetrating dielectric nanoparticles, thereby revealing their interplay beyond the conventional frequency-domain analyses.
arXiv:2606.19544v1 Announce Type: new Abstract: LLM-as-a-Judge has become the dominant evaluation paradigm for language models, but judge validation in practice relies on exact-match agreement, a metric that does not correct for chance and systematically overstates discriminative ability. We present the largest systematic evaluation of LLM-as-a-Judge to date: 21 judges from nine providers across MT-Bench, JudgeBench, and RewardBench, evaluated under three protocols (agreement, consistency, bias audit) over 118 runs and approximately 541,000 individual judgments. Four findings emerge, consistent across the full cohort, including the April 2026 frontier: kappa deflation between exact match and Cohen's kappa is universal (33--41 pp on MT-Bench), judge rankings shift by up to 14 positions across benchmarks, high test--retest reliability (>0.95) coexists with severe position bias (>0.10) in two production-deployed judges (instantiating a consistency--bias paradox), and verbosity bias is small (<0.011) across our cohort under a single pairwise rubric. We distill these into a Minimum Viable Validation Protocol.
arXiv:2606.19547v1 Announce Type: new Abstract: The electronegativity equalization principle provides a simple framework to describe charge redistribution, yet its conventional formulation is limited to a first-order description based on chemical potential equalization. In this work, we introduce 'multi-equalization,' a generalized framework that extends this concept by incorporating higher-order responses within Conceptual Density Functional Theory. This approach represents molecules as sets of flexible electron density partitions, allowing different electronic descriptions (e.g., atomic densities or localized orbitals) to be treated within a unified formalism. We demonstrate that correlations between energy and density derivatives with respect to the number of electrons lead to the simultaneous equalization of multiple descriptors, including chemical hardness and Fukui indices. A constructive algorithm is introduced to determine the optimal density partitions satisfying these multi-equalization conditions. This scheme provides a consistent description of both global charge transfer and local reactivity, overcoming the intrinsic limitations of traditional electronegativity equalization models. Notably, the inclusion of density response functions enables local hardness equalization, introducing spatial resolution into reactivity descriptions. Under multi-equalization, local reactivity descriptors become constrained functionals of the global electron density. This framework establishes a deeper connection between charge equalization models and formal density functional theory, offering a theoretically grounded route toward improved predictions of molecular reactivity.
arXiv:2606.20410v1 Announce Type: new Abstract: MaRDI Open Interfaces is a software package that aims to improve interoperability in scientific computing, particularly, for nonlinear optimization. To this end, this package holds two main characteristics. First, it provides unified interfaces for typical numerical problems to help switching between solvers for the same problem type. Second, it automates data marshalling between programming languages. Hence, computational scientists can conduct experiments faster by using the package, with fewer code-modification and testing efforts. In this work we describe the general structure of the software package and show examples with the interface for nonlinear optimization.
arXiv:2606.20411v1 Announce Type: new Abstract: Direct Advantage Estimation (DAE) has been shown to improve the sample efficiency of deep reinforcement learning algorithms. However, its reliance on full environment observability limits its applicability in realistic settings, and its requirement to model transition probabilities incurs substantial computational overhead for high-dimensional observations. In the present work, we address both limitations. First, we extend the theoretical framework of DAE to partially observable domains with minimal modifications. Second, we reduce its computational complexity by introducing discrete latent dynamics models that efficiently approximate transition probabilities. We evaluate our approach on the Arcade Learning Environment and find that DAE scales effectively with function approximator capacity while retaining high sample efficiency.
arXiv:2606.20412v1 Announce Type: new Abstract: Diamond color-center qubits integrated with photonic circuits can be initialized, manipulated, entangled, and read individually with high fidelity, making them attractive for large-scale modular quantum computers, quantum networks, and distributed quantum sensing. However, the limited size of heteroepitaxially grown single-crystal diamond (SCD) and photonic-grade diamond-on-insulator (DOI) substrates remains a challenge for integration with existing manufacturing processes. Here, we develop a plasma etch recipe to thin direct-bonded (100) SCD membranes (<50 $\mu$m) into large-area, thin-film DOI substrates, and demonstrate free-standing photonic chiplets fabricated from the resulting DOI. The ICP-RIE recipe preserves diamond bonding, provides sufficient micromasking and surface-quality control, and enables thin-film DOI manufacture. We thin a 10 $\mu$m diamond plate bonded to SiO$_2$/Si and obtain a photonic-grade DOI substrate with diamond thickness $\leq$300 nm. The DOI film is around 300 nm thick over 0.5 $\times$ 0.5 mm$^2$, with surface roughness < 0.5 nm, while the bonding interface remains intact. Diamond photonic chiplets are fabricated on this DOI substrate using a standard two-step lithography process, without complex thin-film transfer, under-etching, or pedestal formation. We also present a colorimetric study of diamond visibility on SiO$_2$ and quantify color differences across thicknesses in common colorimetric spaces. This analysis enables automatic diamond-thickness extrapolation from standard optical microscope images with 5 nm resolution, in good agreement with white-light interferometry (WLI) measurements. The DOI substrate and colorimetric thickness-evaluation method provide an effective fabrication platform and reliable validation route for scalable manufacturing of diamond nanophotonic devices, opening a path toward large-scale integrated quantum systems.
arXiv:2606.20234v1 Announce Type: new Abstract: We propose a conservative adaptive rank method for the 1D1V Wigner-Poisson system. The method targets a central challenge in deterministic quantum kinetic simulations: reducing the cost of phase-space evolution while preserving the macroscopic invariants needed for physical fidelity. The scheme combines a sampling-based adaptive rank Wigner-Poisson update [7] with a conservative macroscopic correction. A conservative density-momentum solve provides local macroscopic updates, a Fermi-Dirac-type reconstruction transfers them to the kinetic solution, and a global quadratic moment correction enforces the discrete total energy constraint at the kinetic level. Unlike Maxwell-Boltzmann-type corrections commonly used in classical kinetic settings, the reconstruction uses a Fermi-Dirac-type form motivated by the model's quantum-statistical structure. The corrected state is incorporated into an ACA SVD representation, allowing the numerical rank to adapt to the phase-space complexity generated by the nonlocal Wigner operator and self-consistent Poisson field. Numerical experiments for the two-stream instability, strong Landau damping, and bump-on tail instability show that the method captures benchmark Wigner-Poisson dynamics for several values of the quantum parameter H, maintains bounded adaptive ranks, and preserves the specified global discrete invariants with conservation errors near machine precision. We also compare this formulation, which uses local density-momentum correction plus global total energy correction, with a related globally conservative formulation for mass, momentum, and energy [8]. The two approaches produce nearly identical phase-space and diagnostic results for the periodic benchmark test considered here, indicating that both correction strategies are compatible with adaptive rank compression for Wigner-Poisson dynamics in the tested 1D1V periodic setting.
arXiv:2606.20404v1 Announce Type: new Abstract: Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/
arXiv:2606.20244v1 Announce Type: new Abstract: Vision-language models (VLMs) often underperform on evidence intensive tasks because decisive visual evidence are small, localized, and easy to overlook, leading to failures in evidence readout even when high-level reasoning is intact. Prior inference-time visual interventions can improve grounding without retraining, but they are largely open-loop and lack a mechanism to verify whether highlighted evidence is actually used. We study answer-span prediction entropy as a model-internal feedback signal and show that naive entropy minimization is ambiguous, since low entropy may arise from evidence-grounded confidence or shortcut collapse. To resolve this ambiguity, we introduce low-entropy anchors and an entropy-shaping objective that reduces answer uncertainty while preserving baseline high-confidence tokens. We instantiate this principle in SPOT-E, a plug-and-play test-time method that produces question-conditioned spotlights, optimized per instance via light-weight tuning based on Group Relative Policy Optimization (GRPO). Across all benchmarks and different VLM families, SPOT-E yields consistent gains and improved robustness under visual corruptions. Code is publicly available at: \url{https://github.com/YinBo0927/SPOT-E}
arXiv:2606.20414v1 Announce Type: new Abstract: Spiking neural networks (SNNs) promise energy-efficient computing due to their sparse spatio-temporal activity. However, effectively translating such irregular sparsity into practical performance and energy gains remains challenging, as full-event computing architectures are still underexplored. This paper proposes ExSpike, a general full-event neuromorphic architecture that fully exploits irregular sparsity in SNNs. To realize pure event-driven execution, we first propose a set of dataflow optimizations to ensure that the inputs to each SNN layer remain spike-based, thereby enabling full-event execution throughout the network. We then design a hardware-efficient full-event architecture, named ExSpike, which supports the optimized pure event-driven dataflow and an additional Attention Core for spike-driven self-attention. To further improve computing efficiency, we introduce adjacent-position event compression to reduce redundant accumulations across spatially adjacent spike sequences. ExSpike is implemented on an AMD Xilinx Virtex-7 FPGA and evaluated on both classification and segmentation workloads. Experimental results show that ExSpike achieves high normalized energy efficiency across diverse SNN models while maintaining competitive accuracy, delivering up to 479.15 GOPS, 281.85 GOPS/W, and 0.80 GOPS/W/PE. In particular, ExSpike achieves up to 10$\times$ higher PE-normalized energy efficiency than the SOTA FPGA-based SNN accelerator (FireFly-T). The code for ExSpike is available at \url{https://github.com/xiaoyuehai/ExSpike}.
arXiv:2606.20477v1 Announce Type: new Abstract: We study how to train visually grounded vision-language models (VLMs) for radiology without manual spatial annotations. We introduce RefRad2D, a large-scale bilingual (German/English) dataset of 1.2M CT and MR image-text pairs derived from clinical practice, with task-specific VQA and spatial grounding subsets generated automatically via LLM-based curation and automated segmentation. Trained on this data, our model RadGrounder jointly performs report generation, visual question answering, and spatial grounding via bounding-box detection or segmentation. On external VQA benchmarks (Slake, VQA-RAD), RadGrounder achieves competitive results with specialized medical VLMs. Adding our clinical data to the training mixture improves open-ended VQA over fine-tuning on the downstream datasets alone, showing the transferability of our dataset. Crucially, adding grounding supervision does not degrade language quality, enabling spatially verifiable outputs at no cost to VQA performance.