Computer Science & Mathematics

Modern surgical AI increasingly requires multimodal understanding rather than isolated perception, yet existing datasets remain largely limited to single-task supervision such as recognition or templated question answering. Current medical VQA resources often rely on synthetic templates, lack natural descriptive supervision, or provide sparse annotations that hinder reasoning about spatial relations and procedural context. We introduce MIS-VQA, a surgical vision-language dataset generated through an annotation-driven pipeline that converts heterogeneous surgical annotations into structured textual scene representations. Using EndoVis18 and CholecT50, we transform phase, instrument, interaction, triplet, and segmentation labels into unified contexts that guide large language models to generate clinically coherent question-answer pairs. MIS-VQA supports four response types-concise and detailed descriptions, perception-level grounding, and multi-step reasoning-enabling hierarchical evaluation from atomic visual grounding to compositional reasoning. To address annotation sparsity, we incorporate confidence-filtered segmentation and cross-annotation consistency checks, enabling spatially grounded supervision at scale. MIS-VQA establishes a unified benchmark for surgical vision-language understanding with spatial supervision and multi-tier reasoning evaluation. To demonstrate its utility, we introduce LLaVA-SurgXpert, a family of surgical vision-language models with visual encoders to analyze how visual representations influence multimodal reasoning. Experiments show consistent improvements over existing surgical VQA baselines while revealing capability gaps between perceptual grounding and higher-level reasoning hidden in prior benchmarks.

Inferring directional handover traffic from aggregate mobility observations remains challenging because population density evolution does not fully describe the movement of individual users across cell boundaries. This paper develops a kinetic framework for geometry-induced serving-region crossings in nonstationary cellular networks, connecting boundary-normal velocity statistics with fluid spatiotemporal stochastic geometry (F-STSG) for density-based transport reconstruction. The framework distinguishes signed net population transfer from gross bidirectional crossings and reveals two information limits: density evolution does not uniquely determine physical pairwise transport, while first-order transport alone cannot resolve counter-moving crossings. We establish exact recovery conditions for potential transport and develop coherent and Gaussian mobility closures for estimating directional crossing rates. Sharp moment-constrained bounds further characterize the remaining uncertainty under bounded normal velocities without assuming a specific velocity distribution. An error bound links directional estimation accuracy to transport reconstruction and unresolved mobility information. The formulation also recovers the classical stationary Poisson--Voronoi handover law. Numerical experiments examine the information limits, reconstruction sensitivity, and closure accuracy. On 47 held-out vehicular simulation windows, F-STSG achieves lower net-flow reconstruction errors than three baselines, while auxiliary velocity statistics improve directional estimates under matched Gaussian mobility.

Large language models are increasingly being used in different stages of scientific knowledge production, including the suggestion of references. These models can operate at different levels of reasoning effort and provide summaries of their reasoning processes. This raises two questions: Does increasing reasoning effort improve the verifiability of generated references, and do the resulting reasoning summaries reveal how the model's behavior changes with greater effort? We examined summaries from GPT-5.5 as it generated scholarly references for 1,000 papers under paired low- and high-effort settings. Generated references were matched to external bibliographic records, and reasoning summaries were scored for four predefined features and explored for additional features discovered by GPT-5.5. High effort increased the mean paper-level resolver-verified reference rate from 78.3% to 84.8%, and produced longer reasoning summaries. All four predefined features received higher scores under high effort. The discovered features changed less uniformly: some appeared equally often in both conditions, whereas others were more common under high effort. Within each effort condition, neither feature set showed a consistent increase in verification rates with higher scores.

Cross-modality medical image translation is challenged by anatomically inconsistent synthesis that can compromise clinical diagnosis. We propose StaX-Net, a Stackelberg interleaved asymmetric network that incorporates a partially coupled encoder-discriminator structure, and a 2D feature aggregation and spatial transformation (FAST) layer that distills target-supervised priors under the learning using privileged information paradigm. The shared blocks between the generator and discriminator enhance architectural feature alignment. To stabilize the leader-follower training, a Stackelberg-inspired asymmetric optimization scheme employs explicit shared-parameter freezing and asymmetric learning rates. The FAST layer distills target-domain priors during training and applies a source-only transform at inference. This constrains the synthesis process through target-informed feature representations relative to standard generative architectures. The code is available on GitHub for public use.

Synthetic linguistic data have become increasingly prominent in linguistic research, from the analysis of large collections of AI-generated texts to controlled experiments investigating grammatical and lexical preferences. This glossary entry extends Bresnan's (2016) distinction between the linguistic garden and the bush by introducing an artificial plant nursery as a metaphor for synthetic data. We distinguish between synthetic bush data, consisting primarily of corpus-like collections of model-generated language, and synthetic garden data, in which model outputs are elicited under controlled conditions to investigate specific linguistic hypotheses. We argue that synthetic data constitute a distinct object of linguistic investigation: comparing them with natural language data can reveal both shared patterns and divergences, but their interpretation requires careful consideration of the conditions under which they are generated.

Spatial localization of surgical instruments through bounding-box annotations is a foundational component of surgical scene understanding, supporting detection, tracking, interaction analysis, and grounded vision-language applications. Yet existing annotations often cover selected frames, video subsets, or dataset variants rather than complete procedures. Manual extension is time-consuming and expert-dependent. Additionally, weakly supervised and detector-based pseudo-labels require task-specific training and may be imprecise. In this study, we design a scalable MLLM-based annotation framework that converts surgical frames and existing instrument-presence labels into accurate bounding boxes, and systematically analyze its reliability, cost, and design sensitivity. We instantiate the framework using six MLLMs and a multi-call configuration on laparoscopic cholecystectomy in Cholec80. Using a carefully designed stratification scheme, two annotators independently label 1,000 frames, and an expert surgeon reviews one annotation set to establish the reference. Results show that instrument-set F1 ranges from 88.2 to 92.7,whereas mIoU ranges from 0.180 to 0.735, showing that comparable recognition can conceal large localization differences. A structured multi-stage annotation pipeline achieves 0.784 mIoU, compared with 0.735 for the best single-call baseline and 0.814 human-human agreement. Using the best accuracy pipeline, we create and publicly release a spatial extension of Cholec80 comprising 285,977 bounding boxes across all 184,498 annotated frames from the 80 videos. The annotations and human references are made available on GitHub

Speech-based Alzheimer's Disease (AD) detection holds significant promise due to its scalable and non-invasive nature. However, current Speech Large Language Models (SpeechLLMs), similar to conventional AD detection systems, still treat AD detection primarily as a simple categorical classification task. This paradigm provides limited interpretability of predictions, while leaving the inherent expressiveness and reasoning capabilities of SpeechLLMs largely underutilized. In this work, we take the first step toward reframing AD detection as a deep reasoning task via Reinforcement Learning (RL). We propose ADReasoner, an RL-enhanced SpeechLLM designed to generate accurate diagnostic predictions with interpretable rationales grounded in fine-grained acoustic-linguistic cues. To achieve this, we first systematically construct ADCoT-7K, an AD speech reasoning dataset combining real data and our quality-filtered synthetic data, where each instance features Chain-of-Thought (CoT) annotations and detailed captions. Second, observing the weak perception of AD-related acoustic-linguistic cues in current SpeechLLMs, we leverage Supervised Fine-Tuning (SFT) to develop a cue-perception-enhanced foundation model, ADReasoner-Base, thereby improving AD speech understanding. Third, alongside two standard rule-based rewards, we introduce a reasoning reward and a reasoning-label consistency reward into Group Relative Policy Optimization (GRPO) for RL. Furthermore, a shrinkage mechanism is employed to suppress misleading reasoning signals, while a gating mechanism progressively introduces both rewards to stabilize policy training. Experiments on three AD datasets (Pitt, PROCESS-1, PROCESS-2) demonstrate that ADReasoner outperforms previous state-of-the-art models in both detection performance and explanation quality, advancing AD detection toward interpretable reasoning from speech.

Aesthetic image cropping (AIC) aims to improve image composition and aesthetic quality by removing extraneous or distracting content. Over the past decade, AIC methods have evolved from specialized vision models to multimodal large language models(MLLMs). Reliable evaluation that reflects human preferences is essential for benchmarking AIC methods and guiding model optimization. However, the current AIC evaluation relies largely on geometry-based metrics such as intersection over union (IoU), as well as image aesthetic assessment(IAA) models. Although inconsistencies with human preferences have long been observed, the robustness and reliability of these evaluation criteria remain underexplored. To bridge this gap, we first construct CropBench, a large-scale benchmark comprising 106,995 human preference pairs for 1,019 images, with each candidate set containing the original image and candidate crops produced by 14 AIC methods. A systematic study on CropBench reveals that the existing criteria, ranging from IoU to off-the-shelf IAA assessment models, only exhibit limited alignment with human preferences. Building upon CropBench, we further propose CropReward, an MLLM-based reward model that evaluates a candidate crop by comparing with the original uncropped image and outputs a score indicating explicit aesthetic improvement or degradation relative to the original. In addition, we employ CropReward to construct CropReward-300k, the largest AIC dataset to date, and further develop CropAlign, a preference-aligned AIC model trained solely with CropReward supervision, without human-annotated crop boxes. Experimental results demonstrate that CropReward aligns with human preferences significantly better than existing criteria. CropAlign also surpasses the state-of-the-art AIC methods, demonstrating the advantage of CropReward in guiding AIC model optimization.

A common view regarding even-order nonlinear optical (NLO) effects, such as second-harmonic generation and the bulk photovoltaic effect, is that the breaking of inversion symmetry is necessary. However, this view relies on the assumption that light interacts with materials solely through its electric field. Here, we demonstrate that interference between the electric and magnetic components of the same light field can induce even-order NLO effects even in centrosymmetric materials. Using charge and spin photocurrents as prototypical examples, we present a unified microscopic theory showing how such electric-magnetic dipole interference (EMDI) drives NLO responses in inversion-symmetric crystals. Bilinear in both electric and magnetic fields, these responses scale linearly with light intensity. Combining symmetry analysis with first-principles calculations, we decompose the EMDI-induced photocurrents into shift-like and injection-like contributions and classify their dependence on crystal symmetry, time-reversal symmetry, and light polarization. Notably, we reveal that these EMDI-induced responses in centrosymmetric materials are weaker than conventional responses in non-centrosymmetric materials roughly by a factor of fine structure constant $α\approx [10^{-3}\text{--}10^{-2}] $. Nevertheless, they remain well within experimental reach. Our work thus unlocks NLO functionalities in centrosymmetric materials for e.g., materials characterization, photonics and optoelectronics. This is vital given that many functionally importantly materials, such as cubic Si and cubic $\mathrm{SrTiO_3}$, possess inversion symmetry.

Reconstructing and rendering high-fidelity, animatable, and try-on-ready virtual avatars using only a single camera is crucial for many applications. However, existing try-on-ready human avatar reconstruction methods struggle to achieve this from monocular videos or single images, primarily due to two issues: (1) existing approaches typically use a single layer to represent both garments and the underlying body, ignoring their distinct motion characteristics; and (2) input captured from a single camera often fails to fully cover both the garments and the underlying body. To address these challenges, we propose a physically integrated, double-layer GS-Mesh representation for reconstructing decoupled 3D clothed human avatars from monocular videos or even single images by incorporating 2D generated data. While 2D generation methods can produce video sequences from given pose sequences, they struggle to maintain inter-frame consistency. We decompose the inter-frame inconsistencies in the generated data into geometric and appearance inconsistencies. Based on our 3D representation, we ensure that the final reconstruction and rendering results are both 3D-consistent and temporally coherent. Compared with existing 3D garment transfer methods that rely on multi-view reconstruction and most 2D approaches that use diffusion models for image-based garment editing without novel-view or novel-pose synthesis, our method requires only a monocular video or even a single image while producing visually plausible cloth dynamics during virtual try-on and novel-pose animation. This represents a significant advancement toward virtual try-on with visually plausible cloth dynamics using only a single camera.

Frequency-selective Rayleigh fading in distributed acoustic sensing (DAS) based on phase-sensitive optical time-domain reflectometry (phi-OTDR) calls for effective diversity, but existing rotated-vector-sum (RVS) combining of nonoverlapping subbands incurs a spatial-resolution penalty. In this letter, we propose frequency-domain replica reconstruction and noise-whitened principal component combining (PCC) that combines a full-band replica and overlapping subbands using fixed weights estimated across probing periods after whitening correlated replica noise. This approach retains wider subbands and reduces reliance on noisy first-period alignment, balancing diversity with nominal spatial resolution. Experiments over approximately 7.1 km of fiber with 16-microsecond linear frequency-modulated (LFM) pulses recover 500-Hz vibrations with reduced background phase excursions and a 2.46-2.76-dB improvement in the vibration-to-background ratio (VBR) over the first-period RVS.

Deep graph generators (DGGs) largely lack interpretability, which is a significant limitation for one of their most crucial applications, drug discovery, where interpretable and transparent generation could provide greater control over the desirability of generated molecules, thereby reducing trial-and-error in candidate selection. Common approaches to interpretability, such as disentanglement, cannot guarantee that the learned representations are human-interpretable. To address this, we propose incorporating concept bottlenecks into DGGs, allowing users to interpret, debug, and steer the generation process. We show that this consistently improves distributional fidelity (lower FCD), with only modest, non-systematic effects on validity and uniqueness. We further explore Concept Controller, an optimization-based steering approach that has proven to be effective for steerable image generation. We evaluate our approach on a one-shot molecular graph generation framework, though the underlying ideas can be transferred to sequential graph generation as well.

High-resolution 3D Gaussian Splatting training increases pixel and Gaussian-tile intersection counts, expanding rasterization caches that coexist with projection and gradient state at peak memory. A complete-view update aggregates gradients across the view; each region's forward and backward caches can be released after its contribution is collected. BoundedGS separates rasterization and projection backward passes and bounds the active local rasterization working set according to its measured load. Candidate-indexed projection and compact gradient storage address the view-level peaks exposed by this change, while CPU-assisted structure rebuilding and bounded image supply manage resources outside rendering. In a resolution sweep at a fixed count of about 6M Gaussians, the measured rendering working set stays within 1.073-1.077 GiB from 1K to 16K; at 11K, a warm pass takes 19.79 seconds versus 21.76 for full-frame execution. Across five native 5K outdoor scenes, full training reduces the mean rendering working set from 9.97 to 1.76 GiB (82.3%), retains similar reconstruction quality, and shortens training time from 36.7 to 33.9 minutes. Project page: https://xiaobin2001.github.io/BoundedGS-Web/

Scalable 3D Gaussian Splatting allows one scene representation to serve different Gaussian-count budgets, yet its quality-budget curve may lag behind models optimized separately for each budget. Reducing the number of Gaussians changes both the selected members and the reconstruction tasks carried by the survivors; parameters inherited from a high-budget model may fit a low-budget representation poorly. We propose ParetoGS, which describes member changes and parameter evolution with a small set of adapted budget anchors and decodes intermediate budgets without separate optimization at every requested count. ParetoGS alternates pruning and fixed-member adaptation to build nested anchors with persistent identities, and encodes their parameter changes relative to the full model in a compact residual atlas. At decoding, it interpolates the states of shared members and gradually introduces the rendering contributions of new members, connecting sparse anchors into a continuous-budget representation. Across 13 scenes, ParetoGS attains better average PSNR, SSIM, and LPIPS with fewer Gaussians than the compared prefix methods at 10% of each method's own full model. From a shared full model, it also outperforms FlexGS-imp at its trained 1%, 5%, and 10% budgets and NanoGS-imp at the 1% and 0.1% budgets. On Garden, its non-anchor quality approaches an empirical reference from independently optimized GNS models at higher retention ratios and exceeds it at some lower-budget points. Across 13 scenes, encoding reduces mean additional state from 390.0 to 85.7 MB with an average PSNR change of 0.01 dB. Project page: https://xiaobin2001.github.io/ParetoGS-Web/

Generative AI interfaces can insert sponsored content at multiple points in an answer, where click-through rates (CTRs) depend on the advertiser-position pair. Under the cascade model, users read ads sequentially and stop after their first click, so both ad assignment and display order affect welfare. For arbitrary pair-specific CTRs, Balseiro et al. (WWW 2026) gave a $4/(1-\varepsilon)$-approximation whose monotonicity has not been established, and a monotone $O(\log m)$-approximation for $m$ candidate positions. Earlier truthful constant-factor mechanisms require factorized CTR models and do not apply. Whether monotonicity is compatible with a constant-factor approximation in this general setting was left open by that work. We give a polynomial-time mechanism for arbitrary pair-specific CTRs and any limit on displayed ads that is truthful and individually rational in expectation and achieves a 2-approximation to optimal cascade welfare. Our mechanism selects an advertiser-position matching and a designated last ad, then randomly thins earlier ads so that each selected ad's expected cascade CTR is the same fraction of its standalone CTR. This yields an exact closed-form expression for expected welfare, which we maximize over a bid-independent lottery range using linear programming. Exact optimization ensures monotonicity, while a random-prefix decomposition proves the factor-two guarantee, which is tight for this range. We also show that exactly optimizing the previous welfare surrogate can violate monotonicity, and evaluate welfare and running time on synthetic instances.