Fresh from the feed

arXiv:2506.18588v2 Announce Type: replace-cross Abstract: Lipschitz continuity characterizes the worst-case sensitivity of neural networks to small input perturbations; yet its dynamics (i.e. temporal evolution) during training remains under-explored. We present a rigorous mathematical framework to model the temporal evolution of Lipschitz continuity during training with stochastic gradient descent (SGD). This framework leverages a system of stochastic differential equations (SDEs) to capture both deterministic and stochastic forces. Our theoretical analysis identifies three principal factors driving the evolution: (i) the projection of gradient flows, induced by the optimization dynamics, onto the operator-norm Jacobian of parameter matrices; (ii) the projection of gradient noise, arising from the randomness in mini-batch sampling, onto the operator-norm Jacobian; and (iii) the projection of the gradient noise onto the operator-norm Hessian of parameter matrices. Furthermore, our theoretical framework sheds light on such as how noisy supervision, parameter initialization, batch size, and mini-batch sampling trajectories, among other factors, shape the evolution of the Lipschitz continuity of neural networks. Our experimental results demonstrate strong agreement between the theoretical implications and the observed behaviors.

arXiv:2506.19891v2 Announce Type: replace-cross Abstract: Efficient and controllable data unlearning in federated learning remains challenging, due to the trade-off between forgetting and retention performance. Especially under non-independent and identically distributed (non-IID) settings, where deep feature entanglement exacerbates this dilemma. To address this challenge, we propose FedOrtho, a federated unlearning framework that combines orthogonalized deep convolutional kernels with an activation-driven controllable one-shot soft pruning (OSP) mechanism. FedOrtho enforces kernel orthogonality and local-global alignment to decouple feature representations and mitigate client drift. This structural independence enables precise one-shot pruning of forgetting-related kernels while preserving retained knowledge. FedOrtho achieves SOTA performance on CIFAR-10, CIFAR100 and TinyImageNet with ResNet and VGG frameworks, verifying that FedOrtho supports class-, client-, and sample-level unlearning with over 98% forgetting quality. It reduces computational and communication costs by 2-3 orders of magnitude in federated settings and achieves subsecond-level erasure in centralized scenarios while maintaining over 97% retention accuracy and mitigating membership inference risks.

arXiv:2511.13309v1 Announce Type: new Abstract: The generation of realistic LiDAR point clouds plays a crucial role in the development and evaluation of autonomous driving systems. Although recent methods for 3D LiDAR point cloud generation have shown significant improvements, they still face notable limitations, including the lack of sequential generation capabilities and the inability to produce accurately positioned foreground objects and realistic backgrounds. These shortcomings hinder their practical applicability. In this paper, we introduce DriveLiDAR4D, a novel LiDAR generation pipeline consisting of multimodal conditions and a novel sequential noise prediction model LiDAR4DNet, capable of producing temporally consistent LiDAR scenes with highly controllable foreground objects and realistic backgrounds. To the best of our knowledge, this is the first work to address the sequential generation of LiDAR scenes with full scene manipulation capability in an end-to-end manner. We evaluated DriveLiDAR4D on the nuScenes and KITTI datasets, where we achieved an FRD score of 743.13 and an FVD score of 16.96 on the nuScenes dataset, surpassing the current state-of-the-art (SOTA) method, UniScene, with an performance boost of 37.2% in FRD and 24.1% in FVD, respectively.

arXiv:2507.01463v3 Announce Type: replace-cross Abstract: Instance segmentation of novel objects instances in RGB images, given some example images for each object, is a well known problem in computer vision. Designing a model general enough to be employed for all kinds of novel objects without (re-) training has proven to be a difficult task. To handle this, we present a new training-free framework, called: Novel Object Cyclic Threshold based Instance Segmentation (NOCTIS). NOCTIS integrates two pre-trained models: Grounded-SAM 2 for object proposals with precise bounding boxes and corresponding segmentation masks; and DINOv2 for robust class and patch embeddings, due to its zero-shot capabilities. Internally, the proposal-object matching is realized by determining an object matching score based on the similarity of the class embeddings and the average maximum similarity of the patch embeddings with a new cyclic thresholding (CT) mechanism that mitigates unstable matches caused by repetitive textures or visually similar patterns. Beyond CT, NOCTIS introduces: (i) an appearance score that is unaffected by object selection bias; (ii) the usage of the average confidence of the proposals' bounding box and mask as a scoring component; and (iii) an RGB-only pipeline that performs even better than RGB-D ones. We empirically show that NOCTIS, without further training/fine tuning, outperforms the best RGB and RGB-D methods regarding the mean AP score on the seven core datasets of the BOP 2023 challenge for the "Model-based 2D segmentation of unseen objects" task.

arXiv:2507.03745v3 Announce Type: replace-cross Abstract: Recently, great progress has been achieved in text-to-video (T2V) generation by scaling transformer-based diffusion models to billions of parameters, which can generate high-quality videos. However, existing models typically produce only short clips offline, restricting their use cases in interactive and real-time applications. This paper addresses these challenges by proposing StreamDiT, a streaming video generation model. StreamDiT training is based on flow matching by adding a moving buffer. We design mixed training with different partitioning schemes of buffered frames to boost both content consistency and visual quality. StreamDiT modeling is based on adaLN DiT with varying time embedding and window attention. To practice the proposed method, we train a StreamDiT model with 4B parameters. In addition, we propose a multistep distillation method tailored for StreamDiT. Sampling distillation is performed in each segment of a chosen partitioning scheme. After distillation, the total number of function evaluations (NFEs) is reduced to the number of chunks in a buffer. Finally, our distilled model reaches real-time performance at 16 FPS on one GPU, which can generate video streams at 512p resolution. We evaluate our method through both quantitative metrics and human evaluation. Our model enables real-time applications, e.g. streaming generation, interactive generation, and video-to-video. We provide video results and more examples in our project website: https://cumulo-autumn.github.io/StreamDiT/

arXiv:2507.11017v2 Announce Type: replace-cross Abstract: Post-training quantization (PTQ) offers an efficient approach to compressing large language models (LLMs), significantly reducing memory access and computational costs. Existing compensation-based weight calibration methods often rely on a second-order Taylor expansion to model quantization error, under the assumption that the first-order term is negligible in well-trained full-precision models. However, we reveal that the progressive compensation process introduces accumulated first-order deviations between latent weights and their full-precision counterparts, making this assumption fundamentally flawed. To address this, we propose FOEM, a novel PTQ method that explicitly incorporates first-order gradient terms to improve quantization error compensation. FOEM approximates gradients by performing a first-order Taylor expansion around the pre-quantization weights. This yields an approximation based on the difference between latent and full-precision weights as well as the Hessian matrix. When substituted into the theoretical solution, the formulation eliminates the need to explicitly compute the Hessian, thereby avoiding the high computational cost and limited generalization of backpropagation-based gradient methods. This design introduces only minimal additional computational overhead. Extensive experiments across a wide range of models and benchmarks demonstrate that FOEM consistently outperforms the classical GPTQ method. In 3-bit weight-only quantization, FOEM reduces the perplexity of Llama3-8B by 17.3% and increases the 5-shot MMLU accuracy from 53.8% achieved by GPTAQ to 56.1%. Moreover, FOEM can be seamlessly combined with advanced techniques such as SpinQuant, delivering additional gains under the challenging W4A4KV4 setting and further narrowing the performance gap with full-precision baselines, surpassing existing state-of-the-art methods.

arXiv:2511.10222v2 Announce Type: replace-cross Abstract: Recent progress in large language models (LLMs) has enabled understanding of both speech and non-speech audio, but exposing new safety risks emerging from complex audio inputs that are inadequately handled by current safeguards. We introduce SACRED-Bench (Speech-Audio Composition for RED-teaming) to evaluate the robustness of LLMs under complex audio-based attacks. Unlike existing perturbation-based methods that rely on noise optimization or white-box access, SACRED-Bench exploits speech-audio composition mechanisms. SACRED-Bench adopts three mechanisms: (a) speech overlap and multi-speaker dialogue, which embeds harmful prompts beneath or alongside benign speech; (b) speech-audio mixture, which imply unsafe intent via non-speech audio alongside benign speech or audio; and (c) diverse spoken instruction formats (open-ended QA, yes/no) that evade text-only filters. Experiments show that, even Gemini 2.5 Pro, the state-of-the-art proprietary LLM, still exhibits 66% attack success rate in SACRED-Bench test set, exposing vulnerabilities under cross-modal, speech-audio composition attacks. To bridge this gap, we propose SALMONN-Guard, a safeguard LLM that jointly inspects speech, audio, and text for safety judgments, reducing attack success down to 20%. Our results highlight the need for audio-aware defenses for the safety of multimodal LLMs. The benchmark and SALMONN-Guard checkpoints can be found at https://huggingface.co/datasets/tsinghua-ee/SACRED-Bench. Warning: this paper includes examples that may be offensive or harmful.

arXiv:2511.13315v1 Announce Type: new Abstract: Group activity detection in multi-person scenes is challenging due to complex human interactions, occlusions, and variations in appearance over time. This work presents a computer vision based framework for group activity recognition and action spotting using a combination of deep learning models and graph based relational reasoning. The system first applies Mask R-CNN to obtain accurate actor localization through bounding boxes and instance masks. Multiple backbone networks, including Inception V3, MobileNet, and VGG16, are used to extract feature maps, and RoIAlign is applied to preserve spatial alignment when generating actor specific features. The mask information is then fused with the feature maps to obtain refined masked feature representations for each actor. To model interactions between individuals, we construct Actor Relation Graphs that encode appearance similarity and positional relations using methods such as normalized cross correlation, sum of absolute differences, and dot product. Graph Convolutional Networks operate on these graphs to reason about relationships and predict both individual actions and group level activities. Experiments on the Collective Activity dataset demonstrate that the combination of mask based feature refinement, robust similarity search, and graph neural network reasoning leads to improved recognition performance across both crowded and non crowded scenarios. This approach highlights the potential of integrating segmentation, feature extraction, and relational graph reasoning for complex video understanding tasks.

arXiv:2511.13344v1 Announce Type: new Abstract: This paper presents a novel Mixture-of-Experts framework for object detection, incorporating adaptive routing among multiple YOLOv9-T experts to enable dynamic feature specialization and achieve higher mean Average Precision (mAP) and Average Recall (AR) compared to a single YOLOv9-T model.

arXiv:2511.13353v1 Announce Type: new Abstract: Retinal image quality assessment (RIQA) supports computer-aided diagnosis of eye diseases. However, most tools classify only overall image quality, without indicating acquisition defects to guide recapture. This gap is mainly due to the high cost of detailed annotations. In this paper, we aim to mitigate this limitation by introducing a hybrid semi-supervised learning approach that combines manual labels for overall quality with pseudo-labels of quality details within a multi-task framework. Our objective is to obtain more interpretable RIQA models without requiring extensive manual labeling. Pseudo-labels are generated by a Teacher model trained on a small dataset and then used to fine-tune a pre-trained model in a multi-task setting. Using a ResNet-18 backbone, we show that these weak annotations improve quality assessment over single-task baselines (F1: 0.875 vs. 0.863 on EyeQ, and 0.778 vs. 0.763 on DeepDRiD), matching or surpassing existing methods. The multi-task model achieved performance statistically comparable to the Teacher for most detail prediction tasks (p > 0.05). In a newly annotated EyeQ subset released with this paper, our model performed similarly to experts, suggesting that pseudo-label noise aligns with expert variability. Our main finding is that the proposed semi-supervised approach not only improves overall quality assessment but also provides interpretable feedback on capture conditions (illumination, clarity, contrast). This enhances interpretability at no extra manual labeling cost and offers clinically actionable outputs to guide image recapture.

arXiv:2508.14831v3 Announce Type: replace-cross Abstract: We prove a square-root space simulation for deterministic multitape Turing machines, showing $\mathrm{TIME}[t]\subseteq \mathrm{SPACE}[O(\sqrt{t})]$ \emph{measured in tape cells over a fixed finite alphabet}. The key step is a Height Compression Theorem that uniformly (and in logspace) reshapes the canonical left-deep succinct computation tree for a block-respecting run into a binary tree whose evaluation-stack depth along any DFS path is $O(\log T)$ for $T=\lceil t/b\rceil$, while preserving $O(b)$ workspace at leaves and $O(1)$ at internal nodes. Edges have \emph{addressing/topology} checkable in $O(\log t)$ space, and \emph{semantic} correctness across merges is witnessed by an exact $O(b)$ bounded-window replay at the unique interface. Algorithmically, an Algebraic Replay Engine with constant-degree maps over a constant-size field, together with pointerless DFS, index-free streaming, and a \emph{rolling boundary buffer that prevents accumulation of leaf summaries}, ensures constant-size per-level tokens and eliminates wide counters, yielding the additive tradeoff $S(b)=O(b+t/b)$. Choosing $b=\Theta(\sqrt{t})$ gives $O(\sqrt{t})$ space with no residual multiplicative polylog factors. The construction is uniform, relativizes, and is robust to standard model choices. Consequences include branching-program upper bounds $2^{O(\sqrt{s})}$ for size-$s$ bounded-fan-in circuits, tightened quadratic-time lower bounds for $\mathrm{SPACE}[n]$-complete problems via the standard hierarchy argument, and $O(\sqrt{t})$-space certifying interpreters; under explicit locality assumptions, the framework extends to geometric $d$-dimensional models. Conceptually, the work isolates path bookkeeping as the chief obstruction to $O(\sqrt{t})$ and removes it via structural height compression with per-path analysis rather than barrier-prone techniques.

arXiv:2508.19257v3 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models process visual inputs independently at each timestep, discarding valuable temporal information inherent in robotic manipulation tasks. This frame-by-frame processing makes models vulnerable to visual noise while ignoring the substantial coherence between consecutive frames in manipulation sequences. We propose Temporal Token Fusion (TTF), a training-free approach that intelligently integrates historical and current visual representations to enhance VLA inference quality. Our method employs dual-dimension detection combining efficient grayscale pixel difference analysis with attention-based semantic relevance assessment, enabling selective temporal token fusion through hard fusion strategies and keyframe anchoring to prevent error accumulation. Comprehensive experiments across LIBERO, SimplerEnv, and real robot tasks demonstrate consistent improvements: 4.0 percentage points average on LIBERO (72.4\% vs 68.4\% baseline), cross-environment validation on SimplerEnv (4.8\% relative improvement), and 8.7\% relative improvement on real robot tasks. Our approach proves model-agnostic, working across OpenVLA and VLA-Cache architectures. Notably, TTF reveals that selective Query matrix reuse in attention mechanisms enhances rather than compromises performance, suggesting promising directions for direct KQV matrix reuse strategies that achieve computational acceleration while improving task success rates.

arXiv:2509.09168v2 Announce Type: replace-cross Abstract: Large-scale transformer models have emerged as a powerful tool for semantic communication systems, enabling edge devices to extract rich representations for robust inference across noisy wireless channels. However, their substantial computational demands remain a major barrier to practical deployment in resource-constrained 6G networks. In this paper, we present a training-free framework for adaptive token merging in pretrained vision transformers to jointly reduce inference time and transmission resource usage. We formulate the selection of per-layer merging proportions as a multi-objective optimization problem to balance accuracy and computational cost. We employ Gaussian process-based Bayesian optimization to construct a Pareto frontier of optimal configurations, enabling flexible runtime adaptation to dynamic application requirements and channel conditions. Extensive experiments demonstrate that our method consistently outperforms other baselines and achieves significant reductions in floating-point operations while maintaining competitive accuracy across a wide range of signal-to-noise ratio (SNR) conditions. Additional results highlight the effectiveness of adaptive policies that adjust merging aggressiveness in response to channel quality, providing a practical mechanism to trade off latency and semantic fidelity on demand. These findings establish a scalable and efficient approach for deploying transformer-based semantic communication in future edge intelligence systems.

arXiv:2511.13387v1 Announce Type: new Abstract: Recently, the Denoising Diffusion Codebook Models (DDCM) was proposed. DDCM leverages the Denoising Diffusion Probabilistic Model (DDPM) and replaces the random noise in the backward process with noise sampled from specific sets according to a predefined rule, thereby enabling image compression. However, DDCM cannot be applied to methods other than DDPM. In this paper, we propose the generalized Denoising Diffusion Compression Model (gDDCM), which extends DDCM to mainstream diffusion models and their variants, including DDPM, Score-Based Models, Consistency Models, and Rectified Flow. We evaluate our method on CIFAR-10 and LSUN Bedroom datasets. Experimental results demonstrate that our approach successfully generalizes DDCM to the aforementioned models and achieves improved performance.

arXiv:2511.13397v1 Announce Type: new Abstract: The remarkable progress of Vision-Language Models (VLMs) on a variety of tasks has raised interest in their application to automated driving. However, for these models to be trusted in such a safety-critical domain, they must first possess robust perception capabilities, i.e., they must be capable of understanding a traffic scene, which can often be highly complex, with many things happening simultaneously. Moreover, since critical objects and agents in traffic scenes are often at long distances, we require systems with not only strong perception capabilities at close distances (up to 20 meters), but also at long (30+ meters) range. Therefore, it is important to evaluate the perception capabilities of these models in isolation from other skills like reasoning or advanced world knowledge. Distance-Annotated Traffic Perception Question Answering (DTPQA) is a Visual Question Answering (VQA) benchmark designed specifically for this purpose: it can be used to evaluate the perception systems of VLMs in traffic scenarios using trivial yet crucial questions relevant to driving decisions. It consists of two parts: a synthetic benchmark (DTP-Synthetic) created using a simulator, and a real-world benchmark (DTP-Real) built on top of existing images of real traffic scenes. Additionally, DTPQA includes distance annotations, i.e., how far the object in question is from the camera. More specifically, each DTPQA sample consists of (at least): (a) an image, (b) a question, (c) the ground truth answer, and (d) the distance of the object in question, enabling analysis of how VLM performance degrades with increasing object distance. In this article, we provide the dataset itself along with the Python scripts used to create it, which can be used to generate additional data of the same kind.

arXiv:2511.13399v1 Announce Type: new Abstract: Scene Text Editing (STE) aims to naturally modify text in images while preserving visual consistency, the decisive factors of which can be divided into three parts, i.e., text style, text content, and background. Previous methods have struggled with incomplete disentanglement of editable attributes, typically addressing only one aspect - such as editing text content - thus limiting controllability and visual consistency. To overcome these limitations, we propose TripleFDS, a novel framework for STE with disentangled modular attributes, and an accompanying dataset called SCB Synthesis. SCB Synthesis provides robust training data for triple feature disentanglement by utilizing the "SCB Group", a novel construct that combines three attributes per image to generate diverse, disentangled training groups. Leveraging this construct as a basic training unit, TripleFDS first disentangles triple features, ensuring semantic accuracy through inter-group contrastive regularization and reducing redundancy through intra-sample multi-feature orthogonality. In the synthesis phase, TripleFDS performs feature remapping to prevent "shortcut" phenomena during reconstruction and mitigate potential feature leakage. Trained on 125,000 SCB Groups, TripleFDS achieves state-of-the-art image fidelity (SSIM of 44.54) and text accuracy (ACC of 93.58%) on the mainstream STE benchmarks. Besides superior performance, the more flexible editing of TripleFDS supports new operations such as style replacement and background transfer. Code: https://github.com/yusenbao01/TripleFDS

arXiv:2509.15901v2 Announce Type: replace-cross Abstract: Meeting summarization with large language models (LLMs) remains error-prone, often producing outputs with hallucinations, omissions, and irrelevancies. We present FRAME, a modular pipeline that reframes summarization as a semantic enrichment task. FRAME extracts and scores salient facts, organizes them thematically, and uses these to enrich an outline into an abstractive summary. To personalize summaries, we introduce SCOPE, a reason-out-loud protocol that has the model build a reasoning trace by answering nine questions before content selection. For evaluation, we propose P-MESA, a multi-dimensional, reference-free evaluation framework to assess if a summary fits a target reader. P-MESA reliably identifies error instances, achieving >= 89% balanced accuracy against human annotations and strongly aligns with human severity ratings (r >= 0.70). On QMSum and FAME, FRAME reduces hallucination and omission by 2 out of 5 points (measured with MESA), while SCOPE improves knowledge fit and goal alignment over prompt-only baselines. Our findings advocate for rethinking summarization to improve control, faithfulness, and personalization.

arXiv:2509.24967v4 Announce Type: replace-cross Abstract: Prompt injection attacks pose a pervasive threat to the security of Large Language Models (LLMs). State-of-the-art prevention-based defenses typically rely on fine-tuning an LLM to enhance its security, but they achieve limited effectiveness against strong attacks. In this work, we propose \emph{SecInfer}, a novel defense against prompt injection attacks built on \emph{inference-time scaling}, an emerging paradigm that boosts LLM capability by allocating more compute resources for reasoning during inference. SecInfer consists of two key steps: \emph{system-prompt-guided sampling}, which generates multiple responses for a given input by exploring diverse reasoning paths through a varied set of system prompts, and \emph{target-task-guided aggregation}, which selects the response most likely to accomplish the intended task. Extensive experiments show that, by leveraging additional compute at inference, SecInfer effectively mitigates both existing and adaptive prompt injection attacks, outperforming state-of-the-art defenses as well as existing inference-time scaling approaches.

arXiv:2511.13400v1 Announce Type: new Abstract: With the rapid advancement of Large Models, numerous text-and-vision-fused Multimodal Large Models (MLMs) have emerged. However, these MLMs remain susceptible to informational interference in visual perception, particularly in color perception, which introduces an additional risk of hallucination. To validate this hypothesis, we introduce the "What Color Is It" dataset, a novel benchmark constructed using a simple method to trigger single-modality visual hallucination in MLMs. Based on this dataset, we further investigate the underlying causes of hallucination in the visual modality of MLMs and propose potential solutions to enhance their robustness.

arXiv:2511.13417v1 Announce Type: new Abstract: Accurate delineation of agricultural field boundaries from satellite imagery is essential for land management and crop monitoring, yet existing methods often produce incomplete boundaries, merge adjacent fields, and struggle to scale. We present the Delineate Anything Flow (DelAnyFlow) methodology, a resolution-agnostic approach for large-scale field boundary mapping. DelAnyFlow combines the DelAny instance segmentation model, based on a YOLOv11 backbone and trained on the large-scale Field Boundary Instance Segmentation-22M (FBIS 22M) dataset, with a structured post-processing, merging, and vectorization sequence to generate topologically consistent vector boundaries. FBIS 22M, the largest dataset of its kind, contains 672,909 multi-resolution image patches (0.25-10m) and 22.9million validated field instances. The DelAny model delivers state-of-the-art accuracy with over 100% higher mAP and 400x faster inference than SAM2. DelAny demonstrates strong zero-shot generalization and supports national-scale applications: using Sentinel 2 data for 2024, DelAnyFlow generated a complete field boundary layer for Ukraine (603,000km2) in under six hours on a single workstation. DelAnyFlow outputs significantly improve boundary completeness relative to operational products from Sinergise Solutions and NASA Harvest, particularly in smallholder and fragmented systems (0.25-1ha). For Ukraine, DelAnyFlow delineated 3.75M fields at 5m and 5.15M at 2.5m, compared to 2.66M detected by Sinergise Solutions and 1.69M by NASA Harvest. This work delivers a scalable, cost-effective methodology for field delineation in regions lacking digital cadastral data. A project landing page with links to model weights, code, national-scale vector outputs, and dataset is available at https://lavreniuk.github.io/Delineate-Anything/.