Multimodal Robustness

TTM (Talking to Me) task is a pivotal component in understanding human social interactions, aiming to determine who is engaged in conversation with the camera-wearer. Traditional models often face challenges in real-world scenarios due to missing visual data, neglecting the role of head orientation, and background noise. This study addresses these limitations by introducing EgoAdapt, an adaptive framework designed for robust egocentric "Talking to Me" speaker detection under missing modalities. Specifically, EgoAdapt incorporates three key modules: (1) a Visual Speaker Target Recognition (VSTR) module that captures head orientation as a non-verbal cue and lip movement as a verbal cue, allowing a comprehensive interpretation of both verbal and non-verbal signals to address TTM, setting it apart from tasks focused solely on detecting speaking status; (2) a Parallel Shared-weight Audio (PSA) encoder for enhanced audio feature extraction in noisy environments; and (3) a Visual Modality Missing Awareness (VMMA) module that estimates the presence or absence of each modality at each frame to adjust the system response dynamically.Comprehensive evaluations on the TTM benchmark of the Ego4D dataset demonstrate that EgoAdapt achieves a mean Average Precision (mAP) of 67.39% and an Accuracy (Acc) of 62.01%, significantly outperforming the state-of-the-art method by 4.96% in Accuracy and 1.56% in mAP.

Estimating Emotional Mimicry Intensity (EMI) in naturalistic environments is a critical yet challenging task in affective computing. The primary difficulty lies in effectively modeling the complex, nonlinear temporal dynamics across highly heterogeneous modalities, especially when physical signals are corrupted or missing. To tackle this, we propose TAEMI (Text-Anchored Emotional Mimicry Intensity estimation), a novel multimodal framework designed for the 10th ABAW Competition. Motivated by the observation that continuous visual and acoustic signals are highly susceptible to transient environmental noise, we break the traditional symmetric fusion paradigm. Instead, we leverage textual transcript--which inherently encode a stable, time-independent semantic prior--as central anchors. Specifically, we introduce a Text-Anchored Dual Cross-Attention mechanism that utilizes these robust textual queries to actively filter out frame-level redundancies and align the noisy physical streams. Furthermore, to prevent catastrophic performance degradation caused by inevitably missing data in unconstrained real-world scenarios, we integrate Learnable Missing-Modality Tokens and a Modality Dropout strategy during training. Extensive experiments on the Hume-Vidmimic2 dataset demonstrate that TAEMI effectively captures fine-grained emotional variations and maintains robust predictive resilience under imperfect conditions. Our framework achieves a state-of-the-art mean Pearson correlation coefficient across six continuous emotional dimensions, significantly outperforming existing baseline methods.

Robust depth completion is fundamental to real-world 3D scene understanding, yet existing RGB--LiDAR fusion methods degrade significantly under adverse weather, where both camera images and LiDAR measurements suffer from weather-induced corruption. In this paper, we introduce AURORA-KITTI, the first large-scale, physics-consistent synthetic RGB-LiDAR benchmark for any-weather depth completion and denoising. We further formulate Depth Completion and Denoising~(DCD) as a unified task that reconstructs clean dense metric depth from weather-corrupted RGB images and sparse LiDAR inputs while suppressing weather-induced noise. AURORA-KITTI contains over 82K weather-consistent RGB--LiDAR pairs with metric depth supervision, spanning diverse weather types, three severity levels, day/night scenes, paired clean references, lens occlusion conditions, and textual descriptions.Moreover, we provide DDCD as a strong and efficient distillation-based reference baseline. DDCD is trained on AURORA-KITTI using a frozen teacher applied to paired clean RGB images only during training; at inference, it uses weather-corrupted RGB and sparse LiDAR. Experiments on AURORA-KITTI and the real-world DENSE benchmark show that existing depth completion and foundation depth models remain vulnerable under severe weather, while DDCD improves over weather-aware retraining. Controlled experiments further show that physics-consistent adverse-weather data is the dominant driver of robustness, with clean-prior distillation providing complementary gains. Comparisons with off-the-shelf zero-shot methods use different training protocols; same-data retraining and component ablations separately quantify the contributions of weather-aware data and DDCD. Github: https://github.com/Warwick-Jocelyn/AURORA-KITTI.

Composed image retrieval (CIR) requires multi-modal models to jointly reason over visual content and semantic modifications presented in text-image input pairs. While current CIR models achieve strong performance on common benchmark cases, their accuracies often degrades in more challenging scenarios where negative candidates are semantically aligned with the query image or text. In this paper, we attribute this degradation to focus imbalances, where models disproportionately attend to one modality while neglecting the other. To validate this claim, we propose FBCIR, a multi-modal focus interpretation method that identifies the most crucial visual and textual input components to a model's retrieval decisions. Using FBCIR, we report that focus imbalances are prevalent in existing CIR models, especially under hard negative settings. Building on the analyses, we further propose a CIR data augmentation workflow that facilitates existing CIR datasets with curated hard negatives designed to encourage balanced cross-modal reasoning. Extensive experiments across multiple CIR models demonstrate that the proposed augmentation consistently improves performance in challenging cases, while maintaining their capabilities on standard benchmarks. Together, our interpretation method and data augmentation workflow provide a new perspective on CIR model diagnosis and robustness improvements.

Facial Action Unit (AU) detection in in-the-wild environments remains a formidable challenge due to severe spatial-temporal heterogeneity, unconstrained poses, and complex audio-visual dependencies. While recent multimodal approaches have made progress, they often rely on capacity-limited encoders and shallow fusion mechanisms that fail to capture fine-grained semantic shifts and ultra-long temporal contexts. To bridge this gap, we propose a novel multimodal framework driven by Hierarchical Granularity Alignment and State Space Models.Specifically, we leverage powerful foundation models, namely DINOv2 and WavLM, to extract robust and high-fidelity visual and audio representations, effectively replacing traditional feature extractors. To handle extreme facial variations, our Hierarchical Granularity Alignment module dynamically aligns global facial semantics with fine-grained local active patches. Furthermore, we overcome the receptive field limitations of conventional temporal convolutional networks by introducing a Vision-Mamba architecture. This approach enables temporal modeling with O(N) linear complexity, effectively capturing ultra-long-range dynamics without performance degradation. A novel asymmetric cross-attention mechanism is also introduced to deeply synchronize paralinguistic audio cues with subtle visual movements.Extensive experiments on the challenging Aff-Wild2 dataset demonstrate that our approach significantly outperforms existing baselines, achieving state-of-the-art performance. Notably, this framework secured top rankings in the AU Detection track of the 10th Affective Behavior Analysis in-the-wild Competition.

Multimodal medical image segmentation often faces missing modalities at inference, which induces disagreement among modality experts and makes fusion unstable, particularly on small foreground structures. We propose Consistency Learning of Experts (CLoE), a consistency-driven framework for missing-modality segmentation that preserves strong performance when all modalities are available. CLoE formulates robustness as decision-level expert consistency control and introduces a dual-branch Expert Consistency Learning objective. Modality Expert Consistency enforces global agreement among expert predictions to reduce case-wise drift under partial inputs, while Region Expert Consistency emphasizes agreement on clinically critical foreground regions to avoid background-dominated regularization. We further map consistency scores to modality reliability weights using a lightweight gating network, enabling reliability-aware feature recalibration before fusion. Extensive experiments on BraTS 2020 and MSD Prostate demonstrate that CLoE outperforms state-of-the-art methods in incomplete multimodal segmentation, while exhibiting strong cross-dataset generalization and improving robustness on clinically critical structures.

RGB-Thermal (RGB-T) semantic segmentation is essential for robotic systems operating in low-light or dark environments. However, traditional approaches often overemphasize modality balance, resulting in limited robustness and severe performance degradation when sensor signals are partially missing. Recent advances such as cross-modal knowledge distillation and modality-adaptive fine-tuning attempt to enhance cross-modal interaction, but they typically decouple modality fusion and modality adaptation, requiring multi-stage training with frozen models or teacher-student frameworks. We present RTFDNet, a three-branch encoder-decoder that unifies fusion and decoupling for robust RGB-T segmentation. Synergistic Feature Fusion (SFF) performs channel-wise gated exchange and lightweight spatial attention to inject complementary cues. Cross-Modal Decouple Regularization (CMDR) isolates modality-specific components from the fused representation and supervises unimodal decoders via stop-gradient targets. Region Decouple Regularization (RDR) enforces class-selective prediction consistency in confident regions while blocking gradients to the fusion branch. This feedback loop strengthens unimodal paths without degrading the fused stream, enabling efficient standalone inference at test time. Extensive experiments demonstrate the effectiveness of RTFDNet, showing consistent performance across varying modality conditions. Our implementation will be released to facilitate further research. Our source code are publicly available at https://github.com/curapima/RTFDNet.

Collaborative perception integrates multi-agent perspectives to enhance the sensing range and overcome occlusion issues. While existing multimodal approaches leverage complementary sensors to improve performance, they are highly prone to failure--especially when a key sensor like LiDAR is unavailable. The root cause is that feature fusion leads to semantic mismatches between single-modality features and the downstream modules. This paper addresses this challenge for the first time in the field of collaborative perception, introducing Single-Modality-Operable Multimodal Collaborative Perception (SiMO). By adopting the proposed Length-Adaptive Multi-Modal Fusion (LAMMA), SiMO can adaptively handle remaining modal features during modal failures while maintaining consistency of the semantic space. Additionally, leveraging the innovative "Pretrain-Align-Fuse-RD" training strategy, SiMO addresses the issue of modality competition--generally overlooked by existing methods--ensuring the independence of each individual modality branch. Experiments demonstrate that SiMO effectively aligns multimodal features while simultaneously preserving modality-specific features, enabling it to maintain optimal performance across all individual modalities. The implementation details can be found in https://github.com/dempsey-wen/SiMO.

Multi-modal 3D object detection with bird's eye view (BEV) has achieved desired advances on benchmarks. Nonetheless, the accuracy may drop significantly in the real world due to data corruption such as sensor configurations for LiDAR and scene conditions for camera. One design bottleneck of previous models resides in the tightly coupling of multi-modal BEV features during fusion, which may degrade the overall system performance if one modality or both is corrupted. To mitigate, we propose a Multi-Modal Decouple and Recouple Network for robust 3D object detection under data corruption. Different modalities commonly share some high-level invariant features. We observe that these invariant features across modalities do not always fail simultaneously, because different types of data corruption affect each modality in distinct ways.These invariant features can be recovered across modalities for robust fusion under data corruption.To this end, we explicitly decouple Camera/LiDAR BEV features into modality-invariant and modality-specific parts. It allows invariant features to compensate each other while mitigates the negative impact of a corrupted modality on the other.We then recouple these features into three experts to handle different types of data corruption, respectively, i.e., LiDAR, camera, and both.For each expert, we use modality-invariant features as robust information, while modality-specific features serve as a complement.Finally, we adaptively fuse the three experts to exact robust features for 3D object detection. For validation, we collect a benchmark with a large quantity of data corruption for LiDAR, camera, and both based on nuScenes. Our model is trained on clean nuScenes and tested on all types of data corruption. Our model consistently achieves the best accuracy on both corrupted and clean data compared to recent models.

While diffusion models have revolutionized visual content generation, their rapid adoption has underscored the critical need to investigate vulnerabilities, e.g., to backdoor attacks. In multimodal diffusion models, it is natural to expect that attacking multiple modalities simultaneously (e.g., text and image) would yield complementary effects and strengthen the overall backdoor. In this paper, we challenge this assumption by investigating the phenomenon of Backdoor Modality Collapse, a scenario where the backdoor mechanism degenerates to rely predominantly on a subset of modalities, rendering others redundant. To rigorously quantify this behavior, we introduce two novel metrics: Trigger Modality Attribution (TMA) and Cross-Trigger Interaction (CTI). Through extensive experiments across diverse training configurations in multimodal conditional diffusion, we consistently observe a ``winner-takes-all'' dynamic in backdoor behavior. Our results reveal that (1) attacks often collapse into subset-modality dominance, and (2) cross-modal interaction is negligible or even negative, contradicting the intuition of synergistic vulnerability. These findings highlight a critical blind spot in current assessments, suggesting that high attack success rates often mask a fundamental reliance on a subset of modalities. This establishes a principled foundation for mechanistic analysis and future defense development.

Camera-LiDAR fusion is widely used in autonomous driving to enable accurate 3D object detection. However, bird's-eye view (BEV) fusion detectors can degrade significantly under domain shift and sensor failures, limiting reliability in real-world deployment. Existing robustness approaches often require modifying the fusion architecture or retraining specialized models, making them difficult to integrate into already deployed systems. We propose a Post Fusion Stabilizer (PFS), a lightweight module that operates on intermediate BEV representations of existing detectors and produces a refined feature map for the original detection head. The design stabilizes feature statistics under domain shift, suppresses spatial regions affected by sensor degradation, and adaptively restores weakened cues through residual correction. Designed as a near-identity transformation, PFS preserves performance while improving robustness under diverse camera and LiDAR corruptions. Evaluations on the nuScenes benchmark demonstrate that PFS achieves state-of-the-art results in several failure modes, notably improving camera dropout robustness by +1.2% and low-light performance by +4.4% mAP while maintaining a lightweight footprint of only 3.3 M parameters.

Multimodal sarcasm detection requires resolving pragmatic incongruity across textual, acoustic, and visual cues through cross-modal reasoning. To enable robust sarcasm reasoning with foundation models, we propose SarcasmMiner, a reinforcement learning based post-training framework that resists hallucination in multimodal reasoning. We reformulate sarcasm detection as structured reasoning and adopt a dual-track distillation strategy: high-quality teacher trajectories initialize the student model, while the full set of trajectories trains a generative reward model (GenRM) to evaluate reasoning quality. The student is optimized with group relative policy optimization (GRPO) using decoupled rewards for accuracy and reasoning quality. On MUStARD++, SarcasmMiner increases F1 from 59.83% (zero-shot), 68.23% (supervised finetuning) to 70.22%. These findings suggest that reasoning-aware reward modeling enhances both performance and multimodal grounding.

Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet their dual-stream architecture opens an underexplored attack surface: an adversary who injects content into the webpage DOM simultaneously corrupts both observation channels with a consistent deceptive narrative. Our vulnerability analysis on MiniWob++ reveals that attacks including a visual component far outperform text-only injections, exposing critical gaps in text-centric VLM safety training. Motivated by this finding, we propose Dual-Modality Multi-Stage Adversarial Safety Training (DMAST), a framework that formalizes the agent-attacker interaction as a two-player general-sum Markov game and co-trains both players through a three-stage pipeline: (1) imitation learning from a strong teacher model, (2) oracle-guided supervised fine-tuning that uses a novel zero-acknowledgment strategy to instill task-focused reasoning under adversarial noise, and (3) adversarial reinforcement learning via Group Relative Policy Optimization (GRPO) self-play. On out-of-distribution tasks, DMAST nearly halves the attack success rate (41.2%$\rightarrow$21.4%) while raising task completion by over 60% relative (6.2%$\rightarrow$10.2%). Our approach outperforms established training-based defenses and complements prompt-based defenses, demonstrating genuine co-evolutionary progress and robust generalization to complex, unseen environments. Code is available at https://github.com/huajianduzhuo-code/DMAST_official.

Supervised and unsupervised homography estimation methods depend on image pairs tailored to specific modalities to achieve high accuracy. However, their performance deteriorates substantially when applied to unseen modalities. To address this issue, we propose a training data synthesis method that generates unaligned image pairs with ground-truth offsets from a single input image. Our approach renders the image pairs with diverse textures and colors while preserving their structural information. These synthetic data empower the trained model to achieve greater robustness and improved generalization across various domains. Additionally, we design a network to fully leverage cross-scale information and decouple color information from feature representations, thus improving estimation accuracy. Extensive experiments show that our training data synthesis method improves generalization performance. The results also confirm the effectiveness of the proposed network.

Accurate brain tumor typing requires integrating heterogeneous clinical evidence, including magnetic resonance imaging (MRI), histopathology, and pathology reports, which are often incomplete at the time of diagnosis. We introduce CoRe-BT, a cross-modal radiology-pathology-text benchmark for brain tumor typing, designed to study robust multimodal learning under missing modality conditions. The dataset comprises 310 patients with multi-sequence brain MRI (T1, T1c, T2, FLAIR), including 95 cases with paired H&E-stained whole-slide pathology images and pathology reports. All cases are annotated with tumor type and grade, and MRI volumes include expert-annotated tumor masks, enabling both region-aware modeling and auxiliary learning tasks. Tumors are categorized into six clinically relevant classes capturing the heterogeneity of common and rare glioma subtypes. We evaluate tumor typing under variable modality availability by comparing MRI-only models with multimodal approaches that incorporate pathology information when present. Baseline experiments demonstrate the feasibility of multimodal fusion and highlight complementary modality contributions across clinically relevant typing tasks. CoRe-BT provides a grounded testbed for advancing multimodal glioma typing and representation learning in realistic scenarios with incomplete clinical data.

Multimodal data encountered in real-world scenarios are typically of low quality, with noisy modalities and missing modalities being typical forms that severely hinder model performance and robustness. However, prior works often handle noisy and missing modalities separately. In contrast, we jointly address missing and noisy modalities to enhance model robustness in low-quality data scenarios. We regard both noisy and missing modalities as a unified low-quality modality problem, and propose a unified modality-quality (UMQ) framework to enhance low-quality representations for multimodal affective computing. Firstly, we train a quality estimator with explicit supervised signals via a rank-guided training strategy that compares the relative quality of different representations by adding a ranking constraint, avoiding training noise caused by inaccurate absolute quality labels. Then, a quality enhancer for each modality is constructed, which uses the sample-specific information provided by other modalities and the modality-specific information provided by the defined modality baseline representation to enhance the quality of unimodal representations. Finally, we propose a quality-aware mixture-of-experts module with particular routing mechanism to enable multiple modality-quality problems to be addressed more specifically. UMQ consistently outperforms state-of-the-art baselines on multiple datasets under the settings of complete, missing, and noisy modalities.

The deployment of multimodal models in high-stakes domains, such as self-driving vehicles and medical diagnostics, demands not only strong predictive performance but also reliable mechanisms for detecting failures. In this work, we address the largely unexplored problem of failure detection in multimodal contexts. We propose Adaptive Confidence Regularization (ACR), a novel framework specifically designed to detect multimodal failures. Our approach is driven by a key observation: in most failure cases, the confidence of the multimodal prediction is significantly lower than that of at least one unimodal branch, a phenomenon we term confidence degradation. To mitigate this, we introduce an Adaptive Confidence Loss that penalizes such degradations during training. In addition, we propose Multimodal Feature Swapping, a novel outlier synthesis technique that generates challenging, failure-aware training examples. By training with these synthetic failures, ACR learns to more effectively recognize and reject uncertain predictions, thereby improving overall reliability. Extensive experiments across four datasets, three modalities, and multiple evaluation settings demonstrate that ACR achieves consistent and robust gains. The source code will be available at https://github.com/mona4399/ACR.

Vision-guided robot grasping based on Deep Neural Networks (DNNs) generalizes well but poses safety risks in the Human-Robot Interaction (HRI). Recent works solved it by designing benign adversarial attacks and patches with RGB modality, yet depth-independent characteristics limit their effectiveness on RGBD modality. In this work, we propose the Multimodal Adversarial Quality Policy (MAQP) to realize multimodal safe grasping. Our framework introduces two key components. First, the Heterogeneous Dual-Patch Optimization Scheme (HDPOS) mitigates the distribution discrepancy between RGB and depth modalities in patch generation by adopting modality-specific initialization strategies, employing a Gaussian distribution for depth patches and a uniform distribution for RGB patches, while jointly optimizing both modalities under a unified objective function. Second, the Gradient-Level Modality Balancing Strategy (GLMBS) is designed to resolve the optimization imbalance from RGB and Depth patches in patch shape adaptation by reweighting gradient contributions based on per-channel sensitivity analysis and applying distance-adaptive perturbation bounds. We conduct extensive experiments on the benchmark datasets and a cobot, showing the effectiveness of MAQP.

Explainability and transparent decision-making are essential for the safe deployment of autonomous driving systems. Scene captioning summarizes environmental conditions and risk factors in natural language, improving transparency, safety, and human--robot interaction. However, most existing approaches target structured urban scenarios; in off-road environments, they are vulnerable to single-modality degradations caused by rain, fog, snow, and darkness, and they lack a unified framework that jointly models structured scene captioning and path planning. To bridge this gap, we propose Wild-Drive, an efficient framework for off-road scene captioning and path planning. Wild-Drive adopts modern multimodal encoders and introduces a task-conditioned modality-routing bridge, MoRo-Former, to adaptively aggregate reliable information under degraded sensing. It then integrates an efficient large language model (LLM), together with a planning token and a gate recurrent unit (GRU) decoder, to generate structured captions and predict future trajectories. We also build the OR-C2P Benchmark, which covers structured off-road scene captioning and path planning under diverse sensor corruption conditions. Experiments on OR-C2P dataset and a self-collected dataset show that Wild-Drive outperforms prior LLM-based methods and remains more stable under degraded sensing. The code and benchmark will be publicly available at https://github.com/wangzihanggg/Wild-Drive.

Multimodal adversarial attacks for dense prediction remain largely underexplored. In particular, visual-infrared (VI) perception systems introduce unique challenges due to heterogeneous spectral characteristics and modality-specific intensity distributions. Existing adversarial patch methods are primarily designed for single-modal inputs and fail to account for crossspectral inconsistencies, leading to reduced attack effectiveness and poor stealthiness when applied to VI dense prediction models. To address these challenges, we propose a joint position-color optimization framework (AP-PCO) for generating adversarial patches in visual-infrared settings. The proposed method optimizes patch placement and color composition simultaneously using a fitness function derived from model outputs, enabling a single patch to perturb both visible and infrared modalities. To further bridge spectral discrepancies, we introduce a crossmodal color adaptation strategy that constrains patch appearance according to infrared grayscale characteristics while maintaining strong perturbations in the visible domain, thereby reducing cross-spectral saliency. The optimization procedure operates without requiring internal model information, supporting flexible black-box attacks. Extensive experiments on visual-infrared dense prediction tasks demonstrate that the proposed AP-PCO achieves consistently strong attack performance across multiple architectures, providing a practical benchmark for robustness evaluation in VI perception systems.

The use of multimodal large language models has become widespread, and as such the study of these models and their failure points has become of utmost importance. We study a novel mode of failure that causes degradation in performance indirectly by optimizing a loss term that seeks to maximize numerical instability in the inference stage of these models. We apply this loss term as the optimization target to construct images that, when used on multimodal large language models, cause significant degradation in the output. We validate our hypothesis on state of the art models large vision language models (LLaVa-v1.5-7B, Idefics3-8B, SmolVLM-2B-Instruct) against standard datasets (Flickr30k, MMVet, TextVQA, VQAv2, POPE, COCO) and show that performance degrades significantly, even with a very small change to the input image, compared to baselines. Our results uncover a fundamentally different vector of performance degradation, highlighting a failure mode not captured by adversarial perturbations.

Multimodal models ideally should generalize to unseen domains while remaining data-efficient to reduce annotation costs. To this end, we introduce and study a new problem, Semi-Supervised Multimodal Domain Generalization (SSMDG), which aims to learn robust multimodal models from multi-source data with few labeled samples. We observe that existing approaches fail to address this setting effectively: multimodal domain generalization methods cannot exploit unlabeled data, semi-supervised multimodal learning methods ignore domain shifts, and semi-supervised domain generalization methods are confined to single-modality inputs. To overcome these limitations, we propose a unified framework featuring three key components: Consensus-Driven Consistency Regularization, which obtains reliable pseudo-labels through confident fused-unimodal consensus; Disagreement-Aware Regularization, which effectively utilizes ambiguous non-consensus samples; and Cross-Modal Prototype Alignment, which enforces domain- and modality-invariant representations while promoting robustness under missing modalities via cross-modal translation. We further establish the first SSMDG benchmarks, on which our method consistently outperforms strong baselines in both standard and missing-modality scenarios. Our benchmarks and code are available at https://github.com/lihongzhao99/SSMDG.

Missing modalities present a fundamental challenge in multimodal models, often causing catastrophic performance degradation. Our observations suggest that this fragility stems from an imbalanced learning process, where the model develops an implicit preference for certain modalities, leading to the under-optimization of others. We propose a simple yet efficient method to address this challenge. The central insight of our work is that the dominance relationship between modalities can be effectively discerned and quantified in the frequency domain. To leverage this principle, we first introduce a Frequency Ratio Metric (FRM) to quantify modality preference by analyzing features in the frequency domain. Guided by FRM, we then propose a Multimodal Weight Allocation Module, a plug-and-play component that dynamically re-balances the contribution of each branch during training, promoting a more holistic learning paradigm. Extensive experiments demonstrate that MWAM can be seamlessly integrated into diverse architectural backbones, such as those based on CNNs and ViTs. Furthermore, MWAM delivers consistent performance gains across a wide range of tasks and modality combinations. This advancement extends beyond merely optimizing the performance of the base model; it also manifests as further performance improvements to state-of-the-art methods addressing the missing modality problem.

Medical image segmentation is a cornerstone of computer-assisted diagnosis and treatment planning. While recent multimodal vision-language models have shown promise in enhancing semantic understanding through textual descriptions, their resilience in "in-the-wild" clinical settings-characterized by scarce annotations and hardware-induced image degradations-remains under-explored. We introduce BiCLIP (Bidirectional and Consistent Language-Image Processing), a framework engineered to bolster robustness in medical segmentation. BiCLIP features a bidirectional multimodal fusion mechanism that enables visual features to iteratively refine textual representations, ensuring superior semantic alignment. To further stabilize learning, we implement an augmentation consistency objective that regularizes intermediate representations against perturbed input views. Evaluation on the QaTa-COV19 and MosMedData+ benchmarks demonstrates that BiCLIP consistently surpasses state-of-the-art image-only and multimodal baselines. Notably, BiCLIP maintains high performance when trained on as little as 1% of labeled data and exhibits significant resistance to clinical artifacts, including motion blur and low-dose CT noise.

High-definition (HD) maps are essential for autonomous driving, yet multi-modal fusion often suffers from inconsistency between camera and LiDAR modalities, leading to performance degradation under low-light conditions, occlusions, or sparse point clouds. To address this, we propose SEFMAP, a Subspace-Expert Fusion framework for robust multimodal HD map prediction. The key idea is to explicitly disentangle BEV features into four semantic subspaces: LiDAR-private, Image-private, Shared, and Interaction. Each subspace is assigned a dedicated expert, thereby preserving modality-specific cues while capturing cross-modal consensus. To adaptively combine expert outputs, we introduce an uncertainty-aware gating mechanism at the BEV-cell level, where unreliable experts are down-weighted based on predictive variance, complemented by a usage balance regularizer to prevent expert collapse. To enhance robustness in degraded conditions and promote role specialization, we further propose distribution-aware masking: during training, modality-drop scenarios are simulated using EMA-statistical surrogate features, and a specialization loss enforces distinct behaviors of private, shared, and interaction experts across complete and masked inputs. Experiments on nuScenes and Argoverse2 benchmarks demonstrate that SEFMAP achieves state-of-the-art performance, surpassing prior methods by +4.2% and +4.8% in mAP, respectively. SEF-MAPprovides a robust and effective solution for multi-modal HD map prediction under diverse and degraded conditions.

Action Quality Assessment (AQA) aims to score how well an action is performed and is widely used in sports analysis, rehabilitation assessment, and human skill evaluation. Multi-modal AQA has recently achieved strong progress by leveraging complementary visual and kinematic cues, yet real-world deployments often suffer from non-stationary modality imbalance, where certain modalities become missing or intermittently available due to sensor failures or annotation gaps. Existing continual AQA methods overlook this issue and assume that all modalities remain complete and stable throughout training, which restricts their practicality. To address this challenge, we introduce Bridged Modality Adaptation (BriMA), an innovative approach to multi-modal continual AQA under modality-missing conditions. BriMA consists of a memory-guided bridging imputation module that reconstructs missing modalities using both task-agnostic and task-specific representations, and a modality-aware replay mechanism that prioritizes informative samples based on modality distortion and distribution drift. Experiments on three representative multi-modal AQA datasets (RG, Fis-V, and FS1000) show that BriMA consistently improves performance under different modality-missing conditions, achieving 6--8% higher correlation and 12--15% lower error on average. These results demonstrate a step toward robust multi-modal AQA systems under real-world deployment constraints.

Cross-modal super-resolution (SR) on real-world misaligned data is challenging, as only unlabeled low-resolution (LR) source and high-resolution (HR) guide images with complex spatial misalignment are available. Previous methods either rely on simulated training data or adopt suboptimal alignment strategies that overlook cross-modal dependencies, limiting their practical performance. To address these issues, we propose RobSelf, a self-supervised model that jointly optimizes a misalignment-aware feature translator and a content-aware reference filter online. The translator resolves unsupervised cross-modal and cross-resolution alignment via weakly-supervised, misalignment-aware translation, yielding an aligned guide feature. Guided by this feature, the filter performs reference-based discriminative self-enhancement on the source, enabling SR prediction with high resolution and high fidelity. Experiments on synthesized data and collected real-world data demonstrate that RobSelf achieves state-of-the-art performance, outperforming existing self-supervised and supervised methods. Moreover, it achieves superior efficiency, being up to 15.3$\times$ faster than prior self-supervised methods.

Research on backdoor attacks against multimodal contrastive learning models faces two key challenges: stealthiness and persistence. Existing methods often fail under strong detection or continuous fine-tuning, largely due to (1) cross-modal inconsistency that exposes trigger patterns and (2) gradient dilution at low poisoning rates that accelerates backdoor forgetting. These coupled causes remain insufficiently modeled and addressed. We propose BadCLIP++, a unified framework that tackles both challenges. For stealthiness, we introduce a semantic-fusion QR micro-trigger that embeds imperceptible patterns near task-relevant regions, preserving clean-data statistics while producing compact trigger distributions. We further apply target-aligned subset selection to strengthen signals at low injection rates. For persistence, we stabilize trigger embeddings via radius shrinkage and centroid alignment, and stabilize model parameters through curvature control and elastic weight consolidation, maintaining solutions within a low-curvature wide basin resistant to fine-tuning. We also provide the first theoretical analysis showing that, within a trust region, gradients from clean fine-tuning and backdoor objectives are co-directional, yielding a non-increasing upper bound on attack success degradation. Experiments demonstrate that with only 0.3% poisoning, BadCLIP++ achieves 99.99% attack success rate (ASR) in digital settings, surpassing baselines by 11.4 points. Across nineteen defenses, ASR remains above 99.90% with less than 0.8% drop in clean accuracy. The method further attains 65.03% success in physical attacks and shows robustness against watermark removal defenses.

Multimodal systems are vulnerable to partial or complete loss of input channels at deployment, which undermines reliability in real-world settings. This paper presents ModalImmune, a training framework that enforces modality immunity by intentionally and controllably collapsing selected modality information during training so the model learns joint representations that are robust to destructive modality influence. The framework combines a spectrum-adaptive collapse regularizer, an information-gain guided controller for targeted interventions, curvature-aware gradient masking to stabilize destructive updates, and a certified Neumann-truncated hyper-gradient procedure for automatic meta-parameter adaptation. Empirical evaluation on standard multimodal benchmarks demonstrates that ModalImmune improves resilience to modality removal and corruption while retaining convergence stability and reconstruction capacity.

Multimodal Fusion Learning (MFL), leveraging disparate data from various imaging modalities (e.g., MRI, CT, SPECT), has shown great potential for addressing medical problems such as skin cancer and brain tumor prediction. However, existing MFL methods face three key limitations: a) they often specialize in specific modalities, and overlook effective shared complementary information across diverse modalities, hence limiting their generalizability for multi-disease analysis; b) they rely on computationally expensive models, restricting their applicability in resource-limited settings; and c) they lack robustness against adversarial attacks, compromising reliability in medical AI applications. To address these limitations, we propose a novel Multi-Attention Integration Learning (MAIL) network, incorporating two key components: a) an efficient residual learning attention block for capturing refined modality-specific multi-scale patterns and b) an efficient multimodal cross-attention module for learning enriched complementary shared representations across diverse modalities. Furthermore, to ensure adversarial robustness, we extend MAIL network to design Robust-MAIL by incorporating random projection filters and modulated attention noise. Extensive evaluations on 20 public datasets show that both MAIL and Robust-MAIL outperform existing methods, achieving performance gains of up to 9.34% while reducing computational costs by up to 78.3%. These results highlight the superiority of our approaches, ensuring more reliable predictions than top competitors. Code: https://github.com/misti1203/MAIL-Robust-MAIL.