Papers

Graph Neural Networks (GNNs) are highly vulnerable to adversarial perturbations in both topology and features, making the learning of robust representations a critical challenge. In this work, we bridge GNNs with control theory to introduce a novel defense framework grounded in integer- and fractional-order Lyapunov stability. Unlike conventional strategies that rely on resource-heavy adversarial training or data purification, our approach fundamentally constrains the underlying feature-update dynamics of the GNN. We propose an adaptive, learnable Lyapunov function paired with a novel projection mechanism that maps the network's state into a stable space, thereby offering theoretically provable stability guarantees. Notably, this mechanism is orthogonal to existing defenses, allowing for seamless integration with techniques like adversarial training to achieve cumulative robustness. Extensive experiments demonstrate that our Lyapunov-stable graph neural flows substantially outperform base neural flows and state-of-the-art baselines across standard benchmarks and various adversarial attack scenarios.

Recent world-model-based Vision-Language-Action (VLA) architectures have improved robotic manipulation through predictive visual foresight. However, dense future prediction introduces visual redundancy and accumulates errors, causing long-horizon plan drift. Meanwhile, recent sparse methods typically represent visual foresight using high-level semantic subtasks or implicit latent states. These representations often lack explicit kinematic grounding, weakening the alignment between planning and low-level execution. To address this, we propose StructVLA, which reformulates a generative world model into an explicit structured planner for reliable control. Instead of dense rollouts or semantic goals, StructVLA predicts sparse, physically meaningful structured frames. Derived from intrinsic kinematic cues (e.g., gripper transitions and kinematic turning points), these frames capture spatiotemporal milestones closely aligned with task progress. We implement this approach through a two-stage training paradigm with a unified discrete token vocabulary: the world model is first trained to predict structured frames and subsequently optimized to map the structured foresight into low-level actions. This approach provides clear physical guidance and bridges visual planning and motion control. In our experiments, StructVLA achieves strong average success rates of 75.0% on SimplerEnv-WidowX and 94.8% on LIBERO. Real-world deployments further demonstrate reliable task completion and robust generalization across both basic pick-and-place and complex long-horizon tasks.

Cross-view geo-localization (CVGL) aims to estimate the geographic location of a street image by matching it with a corresponding aerial image. This is critical for autonomous navigation and mapping in complex real-world scenarios. However, the task remains challenging due to significant viewpoint differences and the influence of confounding factors. To tackle these issues, we propose the Causal Learning and Geometric Topology (CLGT) framework, which integrates two key components: a Causal Feature Extractor (CFE) that mitigates the influence of confounding factors by leveraging causal intervention to encourage the model to focus on stable, task-relevant semantics; and a Geometric Topology Fusion (GT Fusion) module that injects Bird's Eye View (BEV) road topology into street features to alleviate cross-view inconsistencies caused by extreme perspective changes. Additionally, we introduce a Data-Adaptive Pooling (DA Pooling) module to enhance the representation of semantically rich regions. Extensive experiments on CVUSA, CVACT, and their robustness-enhanced variants (CVUSA-C-ALL and CVACT-C-ALL) demonstrate that CLGT achieves state-of-the-art performance, particularly under challenging real-world corruptions. Our codes are available at https://github.com/oyss-szu/CLGT.

Deep learning has achieved remarkable success in medical image segmentation, often reaching expert-level accuracy in delineating tumors and tissues. However, most existing approaches remain task-specific, showing strong performance on individual datasets but limited generalization across diverse imaging modalities. Moreover, many methods focus primarily on the encoder, relying on large pretrained backbones that increase computational complexity. In this paper, we propose a decoder-centric approach for generalized 2D medical image segmentation. The proposed Deco-Mamba follows a U-Net-like structure with a Transformer-CNN-Mamba design. The encoder combines a CNN block and Transformer backbone for efficient feature extraction, while the decoder integrates our novel Co-Attention Gate (CAG), Vision State Space Module (VSSM), and deformable convolutional refinement block to enhance multi-scale contextual representation. Additionally, a windowed distribution-aware KL-divergence loss is introduced for deep supervision across multiple decoding stages. Extensive experiments on diverse medical image segmentation benchmarks yield state-of-the-art performance and strong generalization capability while maintaining moderate model complexity. The source code will be released upon acceptance.

Vision-language models (VLMs) have advanced rapidly, yet they still struggle with basic spatial reasoning. Despite strong performance on general benchmarks, modern VLMs remain brittle at understanding 2D spatial relationships such as relative position, layout, and counting. We argue that this failure is not merely a data problem, but is closely tied to dominant design choices in current VLM pipelines: reliance on CLIP-style image encoders and the flattening of images into 1D token sequences with 1D positional encoding. We present a controlled diagnostic study within the LLaVA framework to isolate how these choices affect spatial grounding. We evaluate frontier models and LLaVA variants on a suite of spatial benchmarks, comparing CLIP-based encoders against alternatives trained with denser or generative objectives, as well as variants augmented with 2D positional encoding. Our results show consistent spatial performance gaps across models, and indicate that encoder objectives and positional structure shape spatial behavior, but do not fully resolve it.

Referring image segmentation aims to produce a pixel-level mask for the image region described by a natural-language expression. Although pretrained vision-language models have improved semantic grounding, many existing methods still rely on uniform refinement strategies that do not fully match the diverse reasoning requirements of referring expressions. Because of this mismatch, predictions often contain fragmented regions, inaccurate boundaries, or even the wrong object, especially when pretrained backbones are frozen for computational efficiency. To address these limitations, we propose SERA, a Spatio-Semantic Expert Routing Architecture for referring image segmentation. SERA introduces lightweight, expression-aware expert refinement at two complementary stages within a vision-language framework. First, we design SERA-Adapter, which inserts an expression-conditioned adapter into selected backbone blocks to improve spatial coherence and boundary precision through expert-guided refinement and cross-modal attention. We then introduce SERA-Fusion, which strengthens intermediate visual representations by reshaping token features into spatial grids and applying geometry-preserving expert transformations before multimodal interaction. In addition, a lightweight routing mechanism adaptively weights expert contributions while remaining compatible with pretrained representations. To make this routing stable under frozen encoders, SERA uses a parameter-efficient tuning strategy that updates only normalization and bias terms, affecting less than 1% of the backbone parameters. Experiments on standard referring image segmentation benchmarks show that SERA consistently outperforms strong baselines, with especially clear gains on expressions that require accurate spatial localization and precise boundary delineation.

Understanding and answering questions based on a user's pointing gesture is essential for next-generation egocentric AI assistants. However, current Multimodal Large Language Models (MLLMs) struggle with such tasks due to the lack of gesture-rich data and their limited ability to infer fine-grained pointing intent from egocentric video. To address this, we introduce EgoPointVQA, a dataset and benchmark for gesture-grounded egocentric question answering, comprising 4000 synthetic and 400 real-world videos across multiple deictic reasoning tasks. Built upon it, we further propose Hand Intent Tokens (HINT), which encodes tokens derived from 3D hand keypoints using an off-the-shelf reconstruction model and interleaves them with the model input to provide explicit spatial and temporal context for interpreting pointing intent. We show that our model outperforms others in different backbones and model sizes. In particular, HINT-14B achieves 68.1% accuracy, on average over 6 tasks, surpassing the state-of-the-art, InternVL3-14B, by 6.6%. To further facilitate the open research, we will release the code, model, and dataset. Project page: https://yuuraa.github.io/papers/choi2026egovqa

Embodied task planning requires agents to execute long-horizon, goal-directed actions in complex 3D environments, where success depends on both immediate perception and accumulated experience across tasks. However, most existing LLM-based planners are stateless and reactive, operating without persistent memory and therefore repeating errors and struggling with spatial or temporal dependencies. We propose BrainMem(Brain-Inspired Evolving Memory), a training-free hierarchical memory system that equips embodied agents with working, episodic, and semantic memory inspired by human cognition. BrainMem continuously transforms interaction histories into structured knowledge graphs and distilled symbolic guidelines, enabling planners to retrieve, reason over, and adapt behaviors from past experience without any model fine-tuning or additional training. This plug-and-play design integrates seamlessly with arbitrary multi-modal LLMs and greatly reduces reliance on task-specific prompt engineering. Extensive experiments on four representative benchmarks, including EB-ALFRED, EB-Navigation, EB-Manipulation, and EB-Habitat, demonstrate that BrainMem significantly enhances task success rates across diverse models and difficulty subsets, with the largest gains observed on long-horizon and spatially complex tasks. These results highlight evolving memory as a promising and scalable mechanism for generalizable embodied intelligence.

Open-domain video platforms offer rich, personalized content that could support health, caregiving, and educational applications, but their engagement-optimized recommendation algorithms can expose vulnerable users to inappropriate or harmful material. These risks are especially acute in child-directed and care settings (e.g., dementia care), where content must satisfy individualized safety constraints before being shown. We introduce SafeScreen, a safety-first video screening framework that retrieves and presents personalized video while enforcing individualized safety constraints. Rather than ranking videos by relevance or popularity, SafeScreen treats safety as a prerequisite and performs sequential approval or rejection of candidate videos through an automated pipeline. SafeScreen integrates three key components: (i) profile-driven extraction of individualized safety criteria, (ii) evidence-grounded assessments via adaptive question generation and multimodal VideoRAG analysis, and (iii) LLM-based decision-making that verifies safety, appropriateness, and relevance before content exposure. This design enables explainable, real-time screening of uncurated video repositories without relying on precomputed safety labels. We evaluate SafeScreen in a dementia-care reminiscence case study using 30 synthetic patient profiles and 90 test queries. Results demonstrate that SafeScreen prioritizes safety over engagement, diverging from YouTube's engagement-optimized rankings in 80-93% of cases, while maintaining high levels of safety coverage, sensibleness, and groundedness, as validated by both LLM-based evaluation and domain experts.

Combining Chain-of-Thought (CoT) with Reinforcement Learning (RL) improves text-to-image (T2I) generation, yet the underlying interaction between CoT's exploration and RL's optimization remains unclear. We present a systematic entropy-based analysis that yields three key insights: (1) CoT expands the generative exploration space, while RL contracts it toward high-reward regions; (2) final reward is strongly negatively correlated with both the mean and variance of image-token entropy, highlighting the need to reduce uncertainty and instability; and (3) the entropy of the textual CoT directly governs downstream image quality, with lower-entropy CoTs leading to better generations. Motivated by these findings, we propose Entropy-Guided Group Relative Policy Optimization (EG-GRPO), a fine-tuning strategy that reallocates optimization budget by uncertainty: low-entropy tokens are excluded from reward-driven updates to preserve stability, while high-entropy tokens receive an entropy bonus that encourages structured exploration without collapse. Experiments on standard T2I benchmarks demonstrate that EG-GRPO achieves state-of-the-art performance.

While Vision Language Models (VLMs) have shown promise in Design-to-Code generation, they suffer from a "holistic bottleneck-failing to reconcile high-level structural hierarchy with fine-grained visual details, often resulting in layout distortions or generic placeholders. To bridge this gap, we propose DOne, an end-to-end framework that decouples structure understanding from element rendering. DOne introduces (1) a learned layout segmentation module to decompose complex designs, avoiding the limitations of heuristic cropping; (2) a specialized hybrid element retriever to handle the extreme aspect ratios and densities of UI components; and (3) a schema-guided generation paradigm that bridges layout and code. To rigorously assess performance, we introduce HiFi2Code, a benchmark featuring significantly higher layout complexity than existing datasets. Extensive evaluations on the HiFi2Code demonstrate that DOne outperforms exiting methods in both high-level visual similarity (e.g., over 10% in GPT Score) and fine-grained element alignment. Human evaluations confirm a 3 times productivity gain with higher visual fidelity.

Implicit Neural Representations (INRs) provide compact models of geometry, but it is unclear when their learned shapes can be edited without retraining. We show that the Gram operator induced by the INR's penultimate features admits deformation eigenmodes that parameterize a family of realizable edits of the SDF zero level set. A key finding is that these modes are not intrinsic to the geometry alone: they are reliably recoverable only when the Gram operator is estimated from sufficiently rich sampling distributions. We derive a single closed-form update that performs geometric edits to the INR without optimization by leveraging the deformation modes. We characterize theoretically the precise set of deformations that are feasible under this one-shot update, and show that editing is well-posed exactly within the span of these deformation modes.

Recent advances in computer vision facilitate fully automatic extraction of object-centric relational representations from visual-inertial data. These state representations, dubbed 3D scene graphs, are a hierarchical decomposition of real-world scenes with a dense multiplex graph structure. While 3D scene graphs claim to promote efficient task planning for robot systems, they contain numerous objects and relations when only small subsets are required for a given task. This magnifies the state space that task planners must operate over and prohibits deployment in resource constrained settings. This thesis tests the suitability of existing embodied AI environments for research at the intersection of robot task planning and 3D scene graphs and constructs a benchmark for empirical comparison of state-of-the-art classical planners. Furthermore, we explore the use of graph neural networks to harness invariances in the relational structure of planning domains and learn representations that afford faster planning.

Low-dose CT (LDCT) reduces radiation exposure but introduces protocol-dependent noise and artifacts that vary across institutions. While federated learning enables collaborative training without centralizing patient data, existing methods personalize in image space, making it difficult to separate scanner noise from patient anatomy. We propose ProFed (Projection Guided Personalized Federated Learning), a framework that complements the image space approach by performing dual-level personalization in the projection space, where noise originates during CT measurements before reconstruction combines protocol and anatomy effects. ProFed introduces: (i) anatomy-aware and protocol-aware networks that personalize CT reconstruction to patient and scanner-specific features, (ii) multi-constraint projection losses that enforce consistency with CT measurements, and (iii) uncertainty-guided selective aggregation that weights clients by prediction confidence. Extensive experiments on the Mayo Clinic 2016 dataset demonstrate that ProFed achieves 42.56 dB PSNR with CNN backbones and 44.83 dB with Transformers, outperforming 11 federated learning baselines, including the physics-informed SCAN-PhysFed by +1.42 dB.

Generative models based on the Flow Matching objective, particularly Rectified Flow, have emerged as a dominant paradigm for efficient, high-fidelity image synthesis. However, while existing research heavily prioritizes generation quality and architectural scaling, the underlying dynamics of how RF models memorize training data remain largely underexplored. In this paper, we systematically investigate the memorization behaviors of RF through the test statistics of Membership Inference Attacks (MIA). We progressively formulate three test statistics, culminating in a complexity-calibrated metric that successfully decouples intrinsic image spatial complexity from genuine memorization signals. This calibration yields a significant performance surge – boosting attack AUC by up to 15% and the privacy-critical TPR@1%FPR metric by up to 45% – establishing the first non-trivial MIA specifically tailored for RF. Leveraging these refined metrics, we uncover a distinct temporal pattern: under standard uniform temporal training, a model's susceptibility to MIA strictly peaks at the integration midpoint, a phenomenon we justify via the network's forced deviation from linear approximations. Finally, we demonstrate that substituting uniform timestep sampling with a Symmetric Exponential (U-shaped) distribution effectively minimizes exposure to vulnerable intermediate timesteps. Extensive evaluations across three datasets confirm that this temporal regularization suppresses memorization while preserving generative fidelity.

Valence-arousal (VA) estimation is crucial for capturing the nuanced nature of human emotions in naturalistic environments. While pre-trained vision-language models such as CLIP have demonstrated remarkable semantic alignment capabilities, their application to continuous regression tasks is often limited by the discrete nature of text prompts. In this paper, we propose a novel multimodal framework for VA estimation that introduces Distance-aware Soft Prompt Guidance to bridge the gap between semantic representations and continuous affective dimensions. Specifically, we partition the VA space into multiple discrete regions, each associated with distinct textual descriptions. Rather than relying on hard categorization, we employ a Gaussian kernel to compute soft labels based on the Euclidean distance between the ground-truth coordinates and the region centers, allowing the model to learn fine-grained emotional transitions. For multimodal integration, our architecture utilizes a CLIP image encoder and an Audio Spectrogram Transformer to extract robust visual and acoustic features. These features are temporally modeled using Gated Recurrent Units and integrated through a hierarchical fusion scheme that sequentially combines cross-modal attention for alignment and gated fusion for adaptive refinement. Experimental results on the Aff-Wild2 dataset show that the proposed semantic-guided approach outperforms the official baseline and demonstrates robust performance on in-the-wild data.

Multimodal Large Language Models (MLLMs) have shown strong capabilities in image understanding, motivating recent efforts to extend them to video reasoning. However, existing Video LLMs struggle in online streaming scenarios, where long temporal context must be preserved under strict memory constraints. We propose WAT (Watching Before Thinking), a two-stage framework for online video reasoning. WAT separates processing into a query-independent watching stage and a query-triggered thinking stage. The watching stage builds a hierarchical memory system with a Short-Term Memory (STM) that buffers recent frames and a fixed-capacity Long-Term Memory (LTM) that maintains a diverse summary of historical content using a redundancy-aware eviction policy. In the thinking stage, a context-aware retrieval mechanism combines the query with the current STM context to retrieve relevant historical frames from the LTM for cross-temporal reasoning. To support training for online video tasks, we introduce WAT-85K, a dataset containing streaming-style annotations emphasizing real-time perception, backward tracing, and forecasting. Experiments show that WAT achieves state-of-the-art performance on online video benchmarks, including 77.7% accuracy on StreamingBench and 55.2% on OVO-Bench, outperforming existing open-source online Video LLMs while operating at real-time frame rates.

Vision-based fall analysis has advanced rapidly, but a key bottleneck remains: visually similarmotions can correspond to very different physical outcomes because small differences in contactmechanics and protective responses are hard to infer from appearance alone. Most existingapproaches handle this by supervised injury prediction, which depends on reliable injury labels.In practice, such labels are difficult to obtain: video evidence is often ambiguous (occlusion,viewpoint limits), and true injury events are rare and cannot be safely staged, leading to noisysupervision. We address this problem with PHARL (PHysics-aware Alignment RepresentationLearning), which learns physically meaningful fall representations without requiring clinicaloutcome labels. PHARL regularizes motion embeddings with two complementary constraints:(1) trajectory-level temporal consistency for stable representation learning, and (2) multi-classphysics alignment, where simulation-derived contact outcomes shape embedding geometry. Bypairing video windows with temporally aligned simulation descriptors, PHARL captures localimpact-relevant dynamics while keeping inference purely feed-forward. Experiments on fourpublic datasets show that PHARL consistently improves risk-aligned representation quality overvisual-only baselines while maintaining strong fall-detection performance. Notably, PHARL alsoexhibits zero-shot ordinality: an interpretable severity structure (Head > Trunk > Supported)emerges without explicit ordinal supervision.

Emotion recognition in videos is a pivotal task in affective computing, where identifying subtle psychological states such as Ambivalence and Hesitancy holds significant value for behavioral intervention and digital health. Ambivalence and Hesitancy states often manifest through cross-modal inconsistencies such as discrepancies between facial expressions, vocal tones, and textual semantics, posing a substantial challenge for automated recognition. This paper proposes a recognition framework that integrates temporal segment modeling with Multimodal Large Language Models. To address computational efficiency and token constraints in long video processing, we employ a segment-based strategy, partitioning videos into short clips with a maximum duration of 5 seconds. We leverage the Qwen3-Omni-30B-A3B model, fine-tuned on the BAH dataset using LoRA and full-parameter strategies via the MS-Swift framework, enabling the model to synergistically analyze visual and auditory signals. Experimental results demonstrate that the proposed method achieves an accuracy of 85.1% on the test set, significantly outperforming existing benchmarks and validating the superior capability of Multimodal Large Language Models in capturing complex and nuanced emotional conflicts. The code is released at https://github.com/dlnn123/A-H-Detection-with-Qwen-Omni.git.

Interactive long video generation requires prompt switching to introduce new subjects or events, while maintaining perceptual fidelity and coherent motion over extended horizons. Recent distilled streaming video diffusion models reuse a rolling KV cache for long-range generation, enabling prompt-switch interaction through re-cache at each switch. However, existing streaming methods still exhibit progressive quality degradation and weakened motion dynamics. We identify two failure modes specific to interactive streaming generation: (i) at each prompt switch, current cache maintenance cannot simultaneously retain KV-based semantic context and recent latent cues, resulting in weak boundary conditioning and reduced perceptual quality; and (ii) during distillation, unbounded time indexing induces a positional distribution shift from the pretrained backbone's bounded RoPE regime, weakening pretrained motion priors and long-horizon motion retention. To address these issues, we propose Anchor Forcing, a cache-centric framework with two designs. First, an anchor-guided re-cache mechanism stores KV states in anchor caches and warm-starts re-cache from these anchors at each prompt switch, reducing post-switch evidence loss and stabilizing perceptual quality. Second, a tri-region RoPE with region-specific reference origins, together with RoPE re-alignment distillation, reconciles unbounded streaming indices with the pretrained RoPE regime to better retain motion priors. Experiments on long videos show that our method improves perceptual quality and motion metrics over prior streaming baselines in interactive settings. Project page: https://github.com/vivoCameraResearch/Anchor-Forcing

Diabetic retinopathy (DR) is a leading cause of preventable blindness, and automated fundus image grading can play an important role in large-scale screening. In this work, we investigate three CLIP-based approaches for five-class DR severity grading: (1) a zero-shot baseline using prompt engineering, (2) a hybrid FCN-CLIP model augmented with CBAM attention, and (3) a ranking-aware prompting model that encodes the ordinal structure of DR progression. We train and evaluate on a combined dataset of APTOS 2019 and Messidor-2 (n=5,406), addressing class imbalance through resampling and class-specific optimal thresholding. Our experiments show that the ranking-aware model achieves the highest overall accuracy (93.42%, AUROC 0.9845) and strong recall on clinically critical severe cases, while the hybrid FCN-CLIP model (92.49%, AUROC 0.99) excels at detecting proliferative DR. Both substantially outperform the zero-shot baseline (55.17%, AUROC 0.75). We analyze the complementary strengths of each approach and discuss their practical implications for screening contexts.

Current text-to-video models can make individual frames look convincing while still getting simple interactions wrong: objects move before contact, an intended action is skipped, a placed object keeps drifting, or a support relation breaks. Our starting point is that standard frame-first denoising updates every latent region at every step, even when the prompt implies that only a local interaction should be active. We introduce Event-Driven Video Generation (EVD), a small DiT-compatible intervention that gives the sampler an explicit event signal. A lightweight head predicts token-level event activity; training losses tie that activity to latent state change; and event-gated sampling, with hysteresis and an early-step schedule, applies the update field mainly where an interaction is forming. On EVD-Bench, EVD improves human preference and VBench dynamics for state persistence, spatial accuracy, support relations, and contact stability, while keeping appearance quality comparable to the base model. The results suggest that a modest amount of event structure can correct several interaction failures that otherwise remain hidden behind good frame-level appearance.

Timestep sampling $p(t)$ is a central design choice in Flow Matching models, yet common practice increasingly favors static middle-biased distributions (e.g., Logit-Normal). We show that this choice induces a speed–quality trade-off: middle-biased sampling accelerates early convergence but yields worse asymptotic fidelity than Uniform sampling. By analyzing per-timestep training losses, we identify a U-shaped difficulty profile with persistent errors near the boundary regimes, implying that under-sampling the endpoints leaves fine details unresolved. Guided by this insight, we propose Curriculum Sampling, a two-phase schedule that begins with middle-biased sampling for rapid structure learning and then switches to Uniform sampling for boundary refinement. On CIFAR-10, Curriculum Sampling improves the best FID from $3.85$ (Uniform) to $3.22$ while reaching peak performance at $100$k rather than $150$k training steps. Our results highlight that timestep sampling should be treated as an evolving curriculum rather than a fixed hyperparameter.

Accurate detection and localization of traumatic injuries in abdominal CT remain challenging because voxel-level annotations are limited and expensive to obtain. We present a label-efficient framework for 3D abdominal trauma detection that combines self-supervised pretraining with semi-supervised transformer-based detection. First, we use Masked Image Modeling (MIM) on 1098 CT volumes to pretrain a 3D U-Net encoder for anatomical representation learning. Next, we adapt V-DETR to dense volumetric CT through a feature adapter that converts the encoder feature grid into a compact token sequence for transformer decoding. The pretrained encoder is then integrated with V-DETR and 3D Vertex Relative Position Encoding (3D V-RPE) to improve the localization of irregularly shaped injuries. Finally, semi-supervised teacher-student consistency regularization leverages 2,000 additional unlabeled volumes during detector training. To the best of our knowledge, this is the first application of a 3D DETR-style detector to the RSNA abdominal trauma detection task. On this benchmark, the proposed method achieves 31.33% test mAP@0.50 using only 78 labeled training volumes, corresponding to a 1.53x improvement over supervised-only training. These results show that combining medical-domain pretraining with semi-supervised learning is an effective strategy for label-scarce 3D medical detection.

Autoregressive diffusion enables real-time frame streaming, yet existing sliding-window caches discard past context, causing fidelity degradation, identity drift, and motion stagnation over long horizons. Current approaches preserve a fixed set of early tokens as attention sinks, but this static anchor cannot reflect the evolving content of a growing video. We introduce MemRoPE, a training-free framework with two co-designed components. Memory Tokens continuously compress all past keys into dual long-term and short-term streams via exponential moving averages, maintaining both global identity and recent dynamics within a fixed-size cache. Online RoPE Indexing caches unrotated keys and applies positional embeddings dynamically at attention time, ensuring the aggregation is free of conflicting positional phases. These two mechanisms are mutually enabling: positional decoupling makes temporal aggregation well-defined, while aggregation makes fixed-size caching viable for unbounded generation. Extensive experiments validate that MemRoPE outperforms existing methods in temporal coherence, visual fidelity, and subject consistency across minute- to hour-scale generation.

Text-to-Image (T2I) generation is primarily driven by Diffusion Models (DM) which rely on random Gaussian noise. Thus, like playing the slots at a casino, a DM will produce different results given the same user-defined inputs. This imposes a gambler's burden: To perform multiple generation cycles to obtain a satisfactory result. However, even though DMs use stochastic sampling to seed generation, the distribution of generated content quality highly depends on the prompt and the generative ability of a DM with respect to it. To account for this, we propose Naïve PAINE for improving the generative quality of Diffusion Models by leveraging T2I preference benchmarks. We directly predict the numerical quality of an image from the initial noise and given prompt. Naïve PAINE then selects a handful of quality noises and forwards them to the DM for generation. Further, Naïve PAINE provides feedback on the DM generative quality given the prompt and is lightweight enough to seamlessly fit into existing DM pipelines. Experimental results demonstrate that Naïve PAINE outperforms existing approaches on several prompt corpus benchmarks.

Digital zoom on smartphones relies on learning-based super-resolution (SR) models that operate on RAW sensor images, but obtaining sensor-specific training data is challenging due to the lack of ground-truth images. Synthetic data generation via “unprocessing” pipelines offers a potential solution by simulating the degradations that transform high-resolution (HR) images into their low-resolution (LR) counterparts. However, these pipelines can introduce domain gaps due to incomplete or unrealistic degradation modeling. In this paper, we demonstrate that principled and carefully designed degradation modeling can enhance SR performance in real-world conditions. Instead of relying on generic priors for camera blur and noise, we model device-specific degradations through calibration and unprocess publicly available rendered images into the RAW domain of different smartphones. Using these image pairs, we train a single-image RAW-to-RGB SR model and evaluate it on real data from a held-out device. Our experiments show that accurate degradation modeling leads to noticeable improvements, with our SR model outperforming baselines trained on large pools of arbitrarily chosen degradations.

Page-level calligraphy synthesis requires balancing glyph precision with layout composition. Existing character models lack spatial context, while page-level methods often compromise brushwork detail. In this paper, we present CalliMaster, a unified framework for controllable generation and editing that resolves this conflict by decoupling spatial planning from content synthesis. Inspired by the human cognitive process of “planning before writing”, we introduce a coarse-to-fine pipeline \textbf{(Text $\rightarrow$ Layout $\rightarrow$ Image)} to tackle the combinatorial complexity of page-scale synthesis. Operating within a single Multimodal Diffusion Transformer, a spatial planning stage first predicts character bounding boxes to establish the global spatial arrangement. This intermediate layout then serves as a geometric prompt for the content synthesis stage, where the same network utilizes flow-matching to render high-fidelity brushwork. Beyond achieving state-of-the-art generation quality, this disentanglement supports versatile downstream capabilities. By treating the layout as a modifiable constraint, CalliMaster enables controllable semantic re-planning: users can resize or reposition characters while the model automatically harmonizes the surrounding void space and brush momentum. Furthermore, we demonstrate the framework's extensibility to artifact restoration and forensic analysis, providing a comprehensive tool for digital cultural heritage.

Multimodal instruction tuning is often compute-inefficient because training budgets are spread across large mixed image-video pools whose utility is highly uneven. We present Goal-Driven Data Optimization (GDO), a framework that computes six sample descriptors for each candidate and constructs optimized 1$\times$ training subsets for different goals. Under a fixed one-epoch Qwen3-VL-8B-Instruct training and evaluation recipe on 8 H20 GPUs, GDO uses far fewer training samples than the Uni-10x baseline while converging faster and achieving higher accuracy. Relative to the fixed 512k-sample Uni-10x baseline, GDO reaches the Uni-10x reference after 35.4k samples on MVBench, 26.6k on VideoMME, 27.3k on MLVU, and 34.7k on LVBench, while improving Accuracy by +1.38, +1.67, +3.08, and +0.84 percentage points, respectively. The gains are largest on MVBench and MLVU, while LVBench improves more modestly, consistent with its ultra-long-video setting and the mismatch between that benchmark and the short-video/image-dominant training pool. Across MinLoss, Diverse, Temp, and Temp+, stronger temporal emphasis yields steadily better long-video understanding behavior. Overall, GDO provides a goal-driven data optimization framework that enables faster convergence with fewer training samples under a fixed training protocol. Code is available at https://github.com/rujiewu/GDO.

Generating precise diagnostic reports from High-Resolution Computed Tomography (HRCT) is critical for clinical workflow, yet it remains a formidable challenge due to the high pathological diversity and spatial sparsity within 3D volumes. While Video Language Models (VideoLMs) have demonstrated remarkable spatio-temporal reasoning in general domains, their adaptability to domain-specific, high-volume medical interpretation remains underexplored. In this work, we present AbSteering, an abnormality-centric framework that steers VideoLMs toward precise HRCT report generation. Specifically, AbSteering introduces: (i) an abnormality-centric Chain-of-Thought scheme that enforces abnormality reasoning, and (ii) a Direct Preference Optimization objective that utilizes clinically confusable abnormalities as hard negatives to enhance fine-grained discrimination. Our results demonstrate that general-purpose VideoLMs possess strong transferability to high-volume medical imaging when guided by this paradigm. Notably, AbSteering outperforms state-of-the-art domain-specific CT foundation models, which are pretrained with large-scale CTs, achieving superior detection sensitivity while simultaneously mitigating hallucinations.