Towards Distillation-Resistant Large Language Models: An Information-Theoretic Perspective

Proprietary large language models (LLMs) embody substantial economic value and are generally exposed only as black-box APIs, yet adversaries can still exploit their outputs to extract knowledge via distillation. Existing defenses focus exclusively on text-based distillation, leaving the important logit-based distillation largely unexplored. In this work, we analyze this problem and present an effective solution from an information-theoretic perspective. We characterize distillation-relevant information in teacher outputs using the conditional mutual information (CMI) between teacher logits and input queries conditioned on ground-truth labels. This quantity captures contextual information beneficial for model extraction, motivating us to defend distillation via CMI minimization. Guided by our theoretical analysis, we propose learning a transformation matrix that purifies the original outputs to enhance distillation resistance. We further derive a CMI-inspired anti-distillation objective to optimize this transformation, which effectively removes distillation-relevant information while preserving output utility. Extensive experiments across multiple LLMs and strong distillation algorithms demonstrate that the proposed method significantly degrades distillation performance while preserving task accuracy, effectively protecting models' intellectual property.

Seeing Through the Chain: Mitigate Hallucination in Multimodal Reasoning Models via CoT Compression and Contrastive Preference Optimization

While multimodal reasoning models (MLRMs) have exhibited impressive capabilities, they remain prone to hallucinations, and effective solutions are still underexplored. In this paper, we experimentally analyze the hallucination cause and propose C3PO, a training-based mitigation framework comprising \textbf{C}hain-of-Thought \textbf{C}ompression and \textbf{C}ontrastive \textbf{P}reference \textbf{O}ptimization. Firstly, we identify that introducing reasoning mechanisms exacerbates models' reliance on language priors while overlooking visual inputs, which can produce CoTs with reduced visual cues but redundant text tokens. To this end, we propose to selectively filter redundant thinking tokens for a more compact and signal-efficient CoT representation that preserves task-relevant information while suppressing noise. In addition, we observe that the quality of the reasoning trace largely determines whether hallucination emerges in subsequent responses. To leverage this insight, we introduce a reasoning-enhanced preference tuning scheme that constructs training pairs using high-quality AI feedback. We further design a multimodal hallucination-inducing mechanism that elicits models' inherent hallucination patterns via carefully crafted inducers, yielding informative negative signals for contrastive correction. We provide theoretical justification for the effectiveness and demonstrate consistent hallucination reduction across diverse MLRMs and benchmarks.

ViTMAlis: Towards Latency-Critical Mobile Video Analytics with Vision Transformers

Edge-assisted mobile video analytics (MVA) applications are increasingly shifting from using vision models based on convolutional neural networks (CNNs) to those built on vision transformers (ViTs) to leverage their superior global context modeling and generalization capabilities. However, deploying these advanced models in latency-critical MVA scenarios presents significant challenges. Unlike traditional CNN-based offloading paradigms where network transmission is the primary bottleneck, ViT-based systems are constrained by substantial inference delays, particularly for dense prediction tasks where the need for high-resolution inputs exacerbates the inherent quadratic computational complexity of ViTs. To address these challenges, we propose a dynamic mixed-resolution inference strategy tailored for ViT-backboned dense prediction models, enabling flexible runtime trade-offs between speed and accuracy. Building on this, we introduce ViTMAlis, a ViT-native device-to-edge offloading framework that dynamically adapts to network conditions and video content to jointly reduce transmission and inference delays. We implement a fully functional prototype of ViTMAlis on commodity mobile and edge devices. Extensive experiments demonstrate that, compared to state-of-the-art accuracy-centric, content-aware, and latency-adaptive baselines, ViTMAlis significantly reduces end-to-end offloading latency while improving user-perceived rendering accuracy, providing a practical foundation for next-generation mobile intelligence.

TCAP: Tri-Component Attention Profiling for Unsupervised Backdoor Detection in MLLM Fine-Tuning

Fine-Tuning-as-a-Service (FTaaS) facilitates the customization of Multimodal Large Language Models (MLLMs) but introduces critical backdoor risks via poisoned data. Existing defenses either rely on supervised signals or fail to generalize across diverse trigger types and modalities. In this work, we uncover a universal backdoor fingerprint-attention allocation divergence-where poisoned samples disrupt the balanced attention distribution across three functional components: system instructions, vision inputs, and user textual queries, regardless of trigger morphology. Motivated by this insight, we propose Tri-Component Attention Profiling (TCAP), an unsupervised defense framework to filter backdoor samples. TCAP decomposes cross-modal attention maps into the three components, identifies trigger-responsive attention heads via Gaussian Mixture Model (GMM) statistical profiling, and isolates poisoned samples through EM-based vote aggregation. Extensive experiments across diverse MLLM architectures and attack methods demonstrate that TCAP achieves consistently strong performance, establishing it as a robust and practical backdoor defense in MLLMs.

AoI-Driven Queue Management and Power Control in V2V Networks: A GNN-Enhanced MARL Approach

Queue management and resource allocation play a critical role in enabling cooperative status awareness in vehicular networks. This paper investigates the problem of age of information (AoI)-aware status updates in vehicle-to-vehicle (V2V) communication, where each vehicle's status is represented by multiple interdependent packets. To enable fine-grained queue management at the packet level under resource constraints, we formulate a joint optimization problem that simultaneously learns active packet dropping and transmit power control strategies. A hybrid action space is designed to support both discrete dropping decisions and continuous power control. To exploit the graph-structured interference inherent in V2V topology, a graph neural network (GNN) is introduced to aggregate slowly varying large-scale fading, allowing agents to capture topological dependencies implicitly without frequent message exchange. The overall framework is built upon multi-agent proximal policy optimization (MAPPO), with centralized training and decentralized execution (CTDE). Simulations demonstrate that the proposed method significantly reduces average AoI across a wide range of network densities, channel conditions, and traffic loads, consistently outperforming several baselines.

RPNT: Robust Pre-trained Neural Transformer -- A Pathway for Generalized Motor Decoding

Brain decoding aims to interpret and translate neural activity into behaviors. As such, it is imperative that decoding models are able to generalize across variations, such as recordings from different brain sites, distinct sessions, different types of behavior, and a variety of subjects. Current models can only partially address these challenges and warrant the development of pretrained neural transformer models capable to adapt and generalize. In this work, we propose RPNT - Robust Pretrained Neural Transformer, designed to achieve robust generalization through pretraining, which in turn enables effective finetuning given a downstream task. In particular, RPNT unique components include 1) Multidimensional rotary positional embedding (MRoPE) to aggregate experimental metadata such as site coordinates, session name and behavior types; 2) Context-based attention mechanism via convolution kernels operating on global attention to learn local temporal structures for handling non-stationarity of neural population activity; 3) Robust self-supervised learning (SSL) objective with uniform causal masking strategies and contrastive representations. We pretrained two separate versions of RPNT on distinct datasets a) Multi-session, multi-task, and multi-subject microelectrode benchmark; b) Multi-site recordings using high-density Neuropixel 1.0 probes. The datasets include recordings from the dorsal premotor cortex (PMd) and from the primary motor cortex (M1) regions of nonhuman primates (NHPs) as they performed reaching tasks. After pretraining, we evaluated the generalization of RPNT in cross-session, cross-type, cross-subject, and cross-site downstream behavior decoding tasks. Our results show that RPNT consistently achieves and surpasses the decoding performance of existing decoding models in all tasks.

HOLO: Homography-Guided Pose Estimator Network for Fine-Grained Visual Localization on SD Maps

Visual localization on standard-definition (SD) maps has emerged as a promising low-cost and scalable solution for autonomous driving. However, existing regression-based approaches often overlook inherent geometric priors, resulting in suboptimal training efficiency and limited localization accuracy. In this paper, we propose a novel homography-guided pose estimator network for fine-grained visual localization between multi-view images and standard-definition (SD) maps. We construct input pairs that satisfy a homography constraint by projecting ground-view features into the BEV domain and enforcing semantic alignment with map features. Then we leverage homography relationships to guide feature fusion and restrict the pose outputs to a valid feasible region, which significantly improves training efficiency and localization accuracy compared to prior methods relying on attention-based fusion and direct 3-DoF pose regression. To the best of our knowledge, this is the first work to unify BEV semantic reasoning with homography learning for image-to-map localization. Furthermore, by explicitly modeling homography transformations, the proposed framework naturally supports cross-resolution inputs, enhancing model flexibility. Extensive experiments on the nuScenes dataset demonstrate that our approach significantly outperforms existing state-of-the-art visual localization methods. Code and pretrained models will be publicly released to foster future research.

Self-Supervised Compression and Artifact Correction for Streaming Underwater Imaging Sonar

Real-time imaging sonar has become an important tool for underwater monitoring in environments where optical sensing is unreliable. Its broader use is constrained by two coupled challenges: highly limited uplink bandwidth and severe sonar-specific artifacts (speckle, motion blur, reverberation, acoustic shadows) that affect up to 98% of frames. We present SCOPE, a self-supervised framework that jointly performs compression and artifact correction without clean-noise pairs or synthetic assumptions. SCOPE combines (i) Adaptive Codebook Compression (ACC), which learns frequency-encoded latent representations tailored to sonar, with (ii) Frequency-Aware Multiscale Segmentation (FAMS), which decomposes frames into low-frequency structure and sparse high-frequency dynamics while suppressing rapidly fluctuating artifacts. A hedging training strategy further guides frequency-aware learning using low-pass proxy pairs generated without labels. Evaluated on months of in-situ ARIS sonar data, SCOPE achieves a structural similarity index (SSIM) of 0.77, representing a 40% improvement over prior self-supervised denoising baselines, at bitrates down to <= 0.0118 bpp. It reduces uplink bandwidth by more than 80% while improving downstream detection. The system runs in real time, with 3.1 ms encoding on an embedded GPU and 97 ms full multi-layer decoding on the server end. SCOPE has been deployed for months in three Pacific Northwest rivers to support real-time salmon enumeration and environmental monitoring in the wild. Results demonstrate that learning frequency-structured latents enables practical, low-bitrate sonar streaming with preserved signal details under real-world deployment conditions.

Empowering DINO Representations for Underwater Instance Segmentation via Aligner and Prompter

Underwater instance segmentation (UIS), integrating pixel-level understanding and instance-level discrimination, is a pivotal technology in marine resource exploration and ecological protection. In recent years, large-scale pretrained visual foundation models, exemplified by DINO, have advanced rapidly and demonstrated remarkable performance on complex downstream tasks. In this paper, we demonstrate that DINO can serve as an effective feature learner for UIS, and we introduce DiveSeg, a novel framework built upon two insightful components: (1) The AquaStyle Aligner, designed to embed underwater color style features into the DINO fine-tuning process, facilitating better adaptation to the underwater domain. (2) The ObjectPrior Prompter, which incorporates binary segmentation-based prompts to deliver object-level priors, provides essential guidance for instance segmentation task that requires both object- and instance-level reasoning. We conduct thorough experiments on the popular UIIS and USIS10K datasets, and the results show that DiveSeg achieves the state-of-the-art performance. Code: https://github.com/ettof/Diveseg.

TBStar-Edit: From Image Editing Pattern Shifting to Consistency Enhancement

Recent advances in image generation and editing technologies have enabled state-of-the-art models to achieve impressive results in general domains. However, when applied to e-commerce scenarios, these general models often encounter consistency limitations. To address this challenge, we introduce TBStar-Edit, an new image editing model tailored for the e-commerce domain. Through rigorous data engineering, model architecture design and training strategy, TBStar-Edit achieves precise and high-fidelity image editing while maintaining the integrity of product appearance and layout. Specifically, for data engineering, we establish a comprehensive data construction pipeline, encompassing data collection, construction, filtering, and augmentation, to acquire high-quality, instruction-following, and strongly consistent editing data to support model training. For model architecture design, we design a hierarchical model framework consisting of a base model, pattern shifting modules, and consistency enhancement modules. For model training, we adopt a two-stage training strategy to enhance the consistency preservation: first stage for editing pattern shifting, and second stage for consistency enhancement. Each stage involves training different modules with separate datasets. Finally, we conduct extensive evaluations of TBStar-Edit on a self-proposed e-commerce benchmark, and the results demonstrate that TBStar-Edit outperforms existing general-domain editing models in both objective metrics (VIE Score) and subjective user preference.