Overview

Cross-modal alignment studies how to connect information from different data types, such as images and text, audio and video, or brain signals and visual features, so models can retrieve, interpret, or generate across them. Common approaches use contrastive learning and shared embedding spaces, alongside optimal transport, prototypes, geometric mappings, and local or temporal correspondence; some methods rely on weakly paired or unpaired data. Work is evaluated on tasks such as image-text retrieval, grounding, zero-shot transfer, and multimodal generation, including in medical and scientific settings. A central tension is how to align shared meaning without erasing useful modality-specific information, since stronger alignment scores do not always translate into better downstream performance or consistent behavior across tasks.

Latest papers

474 papers30 papers in the last four weeks, down 47% on the four weeks before. 0.2% of all new papers.

Nov 18, 2026

MS-Point: Multi-Modal Hierarchical Spectral Learning for Point Cloud Understanding

Point-image alignment learns useful object representations, but different local feature patterns can produce the same pooled descriptor and receive the same alignment signal. We introduce MS-Point, a cross-modal pretraining framework that directly supervises intermediate point features and their neighborhood responses. At selected encoder stages, we compute neighborhood averages and residuals before pooling and align their branch descriptors with image features. Branch-level objectives align each representation with its visual target, while cross-level consistency and adaptive fusion coordinate information across branches. A common interface of intermediate features and spatial neighborhoods supports both graph encoders and point Transformers. Evaluations across object classification, few-shot adaptation, dense prediction, and low-label scene learning demonstrate the transfer value of this design. Same-backbone comparisons and capacity-matched and computation-matched controls connect these improvements to the structure of supervision. Together, our findings support directly supervising intermediate features and neighborhood variation to learn transferable, geometry-aware representations across modalities.
Nov 18, 2026

From Points to Trajectories: Rethinking Multimodal Alignment for Physical Events

Advancing physical artificial intelligence (Physical AI) requires moving beyond static object recognition to ground how real-world mechanics (e.g., impact, friction, and force transfer) unfold over time. While multimodal aligners seek to capture these interactions by matching multi-sensor streams, prevailing architectures compress each temporal sequence into a single pooled embedding. We show that this paradigm fails on twin events: distinct physical executions recorded under identical task conditions. Collapsing the temporal dimension averages out transient cross-modal dynamics, effectively reducing physical event alignment to static scene recognition. Formally, we prove that as sequence length increases, any temporally pooled representation converges to the stationary recording condition. Consequently, as windows grow, no choice of loss function or embedding geometry can disambiguate twin events from pooled features alone. Instead, we show that physical events are uniquely identified by the event gap, namely the instantaneous, synchronous co-fluctuations across modalities that are erased by pooling and accumulate only along unpooled trajectories. To recover this signal, we introduce CounterPoint, a lightweight plug-in module that strips out each modality's temporal mean to score frame-level cross-modal agreement. Integrated across four state-of-the-art (SOTA) multimodal aligners, CounterPoint shifts representations from coarse condition matching to fine-grained physical event alignment, consistently outperforming existing temporal alignment methods.
Nov 18, 2026

Stabilizing Neural-to-Image Retrieval with Frequency-Domain Priors and Adaptive Supervision

Visual neural decoding aims to infer perceived visual stimuli from brain activity, offering important insights into human visual cognition while enabling transformative applications in brain computer interfaces and artificial intelligence. Recent advances in pretrained vision-language models have provided a semantically rich representation space for this task. However, reliable alignment between EEG/MEG signals and visual embeddings remains challenging because neural responses are noisy, highly variable across trials, and subject dependent. To address this issue, we propose a unified neural-to-image retrieval framework that improves retrieval performance by aligning neural and image representations in a shared semantic space. The framework integrates dual-stream adaptation, frequency-domain calibration, and phase-guided temperature modulation to promote more robust cross-modal semantic alignment. Specifically, dual-stream adaptation models a low-dimensional shared subspace with correlation regularization to suppress redundant noise and enhance cross-modal consistency. Frequency-domain calibration conditionally modulates neural representations using amplitude and phase statistics. Phase-guided temperature modulation further adjusts the supervision strength in contrastive learning according to the stability of neural responses. Experiments on THINGS-EEG and THINGS-MEG under both intra-subject and inter-subject settings demonstrate that the proposed framework consistently outperforms strong baselines. These results suggest that robust neural-to-image retrieval under heterogeneous neural conditions benefits from a stable shared semantic space, frequency-informed representation calibration, and adaptive supervision. Code will be released at github.com/wanfkuabc/Geometry-Regularized-Brain-Embeddings.
Nov 18, 2026

SIGNTOK: A Text-Aligned Continuous Pose Tokenizer for Sign Language Generation

Text-conditioned sign language generation (SLG) depends not only on the generator but also on the space in which signing is generated. Existing approaches generate directly in pose space, in quantized code spaces, or in learned latent spaces trained for reconstruction. We find that these spaces are organized largely by signer identity rather than linguistic content: on CSL-Daily, a test clip's nearest neighbor in raw pose space is the same sentence signed by another person in only 2% of cases. To address this, we propose SIGNTOK, a continuous pose tokenizer whose latent space is shaped by three training-only heads: a kinematics head that predicts motion dynamics, a sentence-level head that aligns signing with the overall text, and a word-level head that aligns short motion windows with content words. Together, these objectives raise the same-sentence cross-signer nearest-neighbor rate to 82%. On three sign language benchmarks, the same generator trained in the SIGNTOK latent instead of pose space raises back-translation BLEU-4 in five of six benchmark-scorer settings at comparable joint accuracy, and word-level alignment gives higher back-translation scores than sentence-level alignment on every benchmark.
Nov 17, 2026

TextSphere: Efficient Text-to-Image Generation via Shared Hyperspherical Alignment and Uniformity

Recent advances in text-to-image generation have substantially improved image quality, particularly through diffusion-based methods that require multiple inference steps. In contrast, Sphere Encoder offers an efficient generative framework capable of producing images with fewer forward steps. However, extending it to free-form text is nontrivial, because its class-conditioned design lacks cross-modal alignment. To address this challenge, we propose TextSphere, an efficient text-to-image generative framework that extends Sphere Encoder with language conditioning and cross-modal alignment. We incorporate pretrained CLIP text embeddings through adaptive layer normalization (AdaLN) to guide image generation while retaining the original spherical sampling mechanism. To bridge the cross-modal representation gap, we introduce a Spherical Semantic Alignment and Uniformity (SSAU) objective that jointly promotes semantic alignment and representation uniformity on the hypersphere. Experiments on CUB-200-2011, MS-COCO-2014, and Flickr30k demonstrate that TextSphere consistently reduces generation FID (gFID) over the text-conditioned baseline while achieving reconstruction FID (rFID) below 1.0. With only 4 and 6 sampling steps, TextSphere achieves lower gFID than the compared diffusion- and GAN-based models trained from scratch on CUB-200-2011 and MS-COCO-2014, respectively. We further observe that improved reconstruction fidelity does not necessarily yield better generation quality. In addition, we use Partial Least Squares Singular Value Decomposition (PLS-SVD) to analyze cross-modal alignment and modality gap, offering insight into the effectiveness of our approach.
Nov 17, 2026

Understanding Cross-Modal Inconsistency through Representation Subspaces in Omni Models

Omni Models process audio, vision, and text within a unified language backbone, yet often produce different answers for semantically equivalent inputs across modalities. Existing representation-level analyses do not distinguish between information that predicts whether disagreement will occur and the representation whose perturbation affects the model's answer. In this paper, we estimate a low-dimensional Answer Subspace from answer-position representations and use it to separate answer-related components from the remaining representation. Across three Omni Models and two multimodal reasoning tasks, information predictive of cross-modal inconsistency lies mainly outside the Answer Subspace, concentrating in a reproducible two-dimensional representation that is not explained by a fixed modality offset. Matched interventions reveal distinct behavioral roles: perturbing the Answer Subspace substantially changes the model's predictions, whereas perturbing the disagreement representation, like a random subspace of the same dimensionality, leaves them almost unchanged. Our key finding is that cross-modal disagreement can be induced by misaligning modality-specific answer representations: perturbing the Answer Subspace for only one modality induces disagreement, whereas applying the same perturbation to both largely preserves agreement. Thus, the representation that predicts disagreement is distinct from the answer-level representation through which disagreement can be induced, showing that linear decodability does not imply behavioral efficacy.
Nov 16, 2026

Learning What Pairs Miss for Higher-Order Multimodal Alignment

Contrastive learning beyond two modalities now scores the whole tuple with multilinear products, volumes, and fusion heads, critics that can represent the joint distribution. In practice these critics still behave like bags of pairs. Their negatives break at least two of the three pairs, a single pairwise check rejects them, and when most samples can be solved by pairs the higher-order term receives almost no gradient. To solve this problem, we propose CHORD (Contrastive Higher-Order Residuals with Decoys), which trains a residual against decoys drawn from a calibrated pairwise model, negatives that the pairwise model finds plausible, so that rejecting them often needs all three modalities. Drawing decoys costs one inner-product query over a memory bank, and the optimal residual is the log-ratio between the data and the pairwise model, so the decoy loss also bounds what that model misses. On M3-Mix, where one in ten training samples needs all three modalities, our CHORD improves accuracy on those samples by 21.2 percentage points over the strongest of eight baselines. CHORD also gives the best two-to-one results on every real dataset, and its residual improves other retrievers without retraining.
Nov 15, 2026

What Matters for Cross-Modal Alignment

The Platonic Representation Hypothesis posits that models trained on different modalities converge toward a shared representation. Related work aligns frozen image and text encoders on natural images without paired data. We test zero-pair alignment on five datasets across four modalities: natural images, environmental audio, chest radiographs, and 3D computed tomography (CT). It succeeds only on COCO and fails on the other four. We analyze the failures and find that alignment must first locate the subspace where the modalities co-vary and then orient the mapping within it. Consistent with previous studies, medical imaging is the hardest case: its shared subspace carries the smallest fraction of the variance. When this fraction is small, alignment needs image-text pairs, and random-matrix theory predicts how many pairs are needed. These findings motivate LightAlign: a few pairs let a ridge regression between two frozen encoders locate and orient the mapping simultaneously, and a classifier fitted only on report text then predicts findings from the mapped images. On 3D CT, LightAlign surpasses models trained end to end on image-report pairs without finding-level labels: with only 200 pairs, LightAlign outperforms CT-CLIP on CT-RATE (0.754 vs. 0.733 AUROC), and with only 100 CT-RATE pairs, it does so on RAD-ChestCT, an out-of-distribution dataset from a different hospital. On CheXpert, it reaches 0.862 mean AUROC with only about 5% as many pairs as models trained on MIMIC-CXR.
Nov 14, 2026

Attribution-Guided Shared-Private Decoupling for Noise-Reduced Audio-Visual Representation Learning

Audio-visual representation learning commonly aligns paired audio and visual signals with sample-level contrastive objectives. However, real-world audio-visual pairs usually contain heterogeneous patch tokens: only a subset carries reliable cross-modal shared semantics, while many others mainly encode modality-private residuals. Aggregating all tokens into a single global representation can therefore introduce semantic noise into cross-modal alignment. Motivated by the minimal-sufficiency principle, we propose an attribution-guided shared-private decoupling framework to mitigate this issue. Our key idea is to identify shared-cover subviews that preserve dominant cross-modal information while suppressing modality-private residuals, and use them to supervise decoupled shared and private representations. To identify such subviews efficiently, we draw on transformer attribution method and introduce Layer-Truncated Attribution (LTA), which provides a lightweight estimate of token-level relevance to cross-modal shared information. We further adopt a teacher-student framework, where the teacher provides attribution-induced supervision and the student internalizes shared-private decoupling into its representations, without introducing extra inference cost. Experiments on retrieval, classification, and sound-prompted semantic segmentation show consistent gains over strong baselines. Compared with baseline, our method improves the average zero-shot R@1 by 4.0 across AudioSet and VGGSound retrieval, and improves AS20K classification mAP by 3.1, demonstrating the effectiveness of fine grained attribution guidance for reducing semantic noise.reducing semantic noise. Code will be released at https://github.com/wanglg20/LTA
Nov 13, 2026

STaG: Human-Annotation-Free Unsupervised Domain Adaptation for Text-Based Person Retrieval

Real-world deployment of Text-Based Person Retrieval (TBPR) is bottlenecked by the absence of textual annotations in new domains. Existing Unsupervised Domain Adaptation (UDA-TBPR) methods still assume access to target-domain human-written annotations (albeit unpaired with images), limiting their practicality. We consider a more practical setting: human-annotation-free UDA-TBPR, with labeled image-text data in the source domain and annotation-free, unlabeled images in the target domain. This setting introduces two core challenges: the absence of cross-modal alignment in the target domain, and visual domain shifts from source to target. To tackle these, we propose STaG, a dual-path framework comprising Style-Agnostic Text Alignment (SATA) and Heterogeneous Graph Memory Alignment (HGMA). Specifically, SATA leverages source-domain data to decouple identity-discriminative semantics from machine-generated stylistic artifacts, establishing robust cross-modal alignment for the target domain. Meanwhile, HGMA exploits cross-domain semantic neighborhoods via a memory-based graph to learn domain-invariant visual features. Ultimately, STaG achieves state-of-the-art performance across three benchmarks under this realistic setting.
Nov 12, 2026

Asymmetric Multimodal Learning with Decomposed JEPAs

Cross-modal alignment is the standard objective for building multimodal representations. We show that alignment as a sole objective is lossy wherever modalities are informationally asymmetric---where neither can express what the other carries. Visual question answering is an obvious case: a question is a selective relational operator probing a dense visual field whose fine-grained features no query recovers, so what alignment preserves is not what the task requires. We prove an Alignment Redundancy Bound: alignment-only Dirichlet energy minimization retains at most the redundant information atom \Red(Z;X1,X2)\Red(Z; X_1, X_2), annihilating private modality-specific structure \Unq(Z;Xm)\Unq(Z; X_m) and cross-modal synergy \Syn(Z;X1,X2)\Syn(Z; X_1, X_2). We introduce Decomposed JEPA (D-JEPA), which factorizes latents into an aligned shared core and private modality-specific fibers. The shared core retains most of the structure alignment targets; the private fibers recover what alignment discards, via contextual masked prediction and asymmetric cross-subspace independence. Under information asymmetry, alignment-only training undergoes catastrophic semantic collapse on compositional reasoning benchmarks, surviving only on binary queries. D-JEPA averts this collapse, substantially improving downstream accuracy over both alignment-only and alignment-with-reconstruction baselines, and restoring a positive multimodal synergy gap in place of the baseline's inverted gap. Information-theoretic audits confirm that D-JEPA recovers most of the mutual information available from a frozen backbone, and causal fiber-swap interventions demonstrate that the recovered information is causally localized in the private fibers.
Nov 11, 2026

Sub-actions in Action: Text-Guided Hand-Role Alignment for Sub-Action Recognition

Our objective is to enhance VLMs ability to distinguish the roles of the left and right hands in egocentric actions. We introduce sub-action detection, a task that requires models to explicitly recognise the left- and right-hand roles and their manipulated objects within an ongoing high-level action. To address this challenge, we propose BiScopeNet, a framework that augments existing video-LLMs without modifying their vision encoders or language models. BiScopeNet incorporates two components: FocusAlign, a gated text-guided attention-based module that selectively aligns video features with textual context to emphasise action-relevant cues, and MetaLoss, a masked, object-weighted loss that prioritises visually grounded terms. To supervise sub-action detection, we introduce BiScopeCap-100K, a large-scale video-level dataset curated from existing egocentric datasets, providing dense captions that jointly describe high-level actions, left/right-hand roles, and manipulated objects, enriched with spatial metadata for grounding. Across the evaluated Video-LLM backbones, BiScopeNet improves action, hand-role, and object recognition over the corresponding supervised baselines, achieving the strongest aggregate performance among the evaluated configurations on BiScopeCap.
Nov 11, 2026

DuoJEPA: Cross-Modal Translation Between Unaligned Self-Supervised Embedding Spaces

Learning from paired modalities is commonly equated with aligning them: two encoders are trained until the embeddings of a pair coincide in a shared space. We study whether cross-modal retrieval and classification can instead be obtained while the two encoders' own embedding spaces remain directly unaligned. DuoJEPA trains one encoder per modality with the self-supervised LeJEPA objective and appends to each encoder a single transformer layer that translates its embedding into the other encoder's space, supervised by a contrastive objective computed entirely within that target space. No loss term compares the two own spaces directly, although the encoders are optimised jointly through the translation objective. In our experiments, the translator's residual connections determine where the cross-modal pressure is absorbed: with a residual translator the two encoders become directly aligned, whereas with a skipless (residual-free) translator a paired-cosine test detects no alignment in the final models, and at most a weak one in the ablations, at comparable cross-modal performance. The learned class-token translation is well approximated by a single scaled rotation of the source space. We evaluate DuoJEPA on two modality pairs that differ in nearly every respect. On chest radiographs and reports from CheXpert Plus, it attains a zero-shot AUC of 86.6 and a cross-modal retrieval score of 78.5, compared with 60.0 and 64.2 for our reimplementation of LeVLJEPA with the same encoders and data. On Sentinel-2 optical and Sentinel-1 radar imagery, the frozen encoders fall within the range of published Sentinel foundation models under our reproduction of the Copernicus-Bench linear-probe protocol, the translator retrieves the paired image of the other sensor among 4,000 candidates in 77-79% of cases, and a patch-level extension transfers the within-image similarity structure of the scene.
Nov 10, 2026

Relaxing Cross-Modal Correspondence for Multimodal Remote Sensing Pretraining

Multimodal remote sensing pretraining commonly relies on exact instance-level correspondence across modalities, implicitly treating strictly paired observations as a prerequisite for learning transferable representations. In this work, we revisit this assumption by relaxing cross-modal correspondence from the instance level to the semantic level. We introduce Semantic Pairing (SemPair), a pairing-relaxed pretraining strategy that replaces exact instance pairs with class-consistent cross-instance pairs. To characterize the effect of correspondence relaxation, we formulate a latent-factor model that separates cross-modal dependence into shared semantic structure and instance-specific correlation, showing that SemPair preserves the former while marginalizing instance-specific cross-modal correlations in expectation. We further introduce Mixed Pairing (MixPair) as a controlled interpolation between semantic and exact pairing to study correspondence granularity along a continuous spectrum. Extensive experiments consistently show that SemPair achieves transfer performance comparable to exact pairing (ExaPair), while MixPair reveals a non-monotonic relationship between downstream transfer and the instance-pairing ratio. These results demonstrate that exact instance correspondence is not a prerequisite for effective multimodal remote sensing pretraining and establish cross-modal correspondence granularity as an important design dimension for transferable multimodal representation learning. More broadly, this finding suggests a practical route toward scaling multimodal pretraining with reduced dependence on costly instance-aligned data.
Nov 10, 2026

Integrating Action Prefix Prediction and Alignment Across Modalities for Multimodal Reinforcement Learning

Learning representations from multimodal observations is difficult in reinforcement learning with limited interaction data. Multimodal Joint Predictive Representation combines multimodal latent prediction with contrastive learning across modalities, but its recursive rollout depends on earlier predictions and its pairwise alignment focuses on correspondence between modalities. We propose Improved Predictive Representation, which combines prediction from action prefixes with alignment at the observed anchor based on Contrastive Multimodal learning. At each horizon, it predicts future representations directly from the observed anchor and the corresponding sequence of actions. We evaluate two crop configurations: IPR-S with a minimum retained crop area of 10%, and IPR-H with 50%. Across five DeepMind Control Suite tasks, IPR-H achieves an average return across tasks that is 10.0% higher than the baseline with visual and depth images and 3.4% higher with visual images and position vectors. The relative performance of the two configurations varies across tasks. In Webots, both IPR configurations achieve higher mean final returns than the baseline without LiDAR, but not when LiDAR is available.
Nov 9, 2026

Where Does Alignment Emerge and What Does It Mean? A Closer Look at the Platonic Representation Hypothesis

The Platonic Representation Hypothesis (PRH) suggests that neural networks are converging toward similar representation spaces. As language models become more capable, their representations are aligning with those of vision models. However, recent work has questioned whether this trend continues to hold for newer language models. Moreover, it has been shown that representation similarity metrics can be confounded by model depth and width, requiring appropriate baselines for interpretation. In this work, we first investigate where alignment emerges between unimodal models across model layers and representation dimensions. Next, we find that the PRH capability-alignment trend persists for newer models as long as the captions are sufficiently clean and informative. Finally, we contextualize cross-modal alignment, introducing unimodal and multimodal baselines on various datasets. Our results show that the cross-modal representational alignment of unimodal models can surpass that of explicitly aligned vision-language models, but remains substantially behind the alignment observed for models of the same modality. Code is available at https://github.com/ExplainableML/prh-closer-look.
Nov 9, 2026

Complementarity-Driven Dual-Teacher Hierarchical Alignment for Zero-Shot EEG-to-Image Retrieval

Decoding visual information from EEG is central to brain--computer interfaces and brain-inspired intelligence, and contrastive alignment with pretrained vision foundation models (VFMs) has become the dominant paradigm. However, existing methods align EEG solely with CLIP-family VFMs, leaving hierarchical visual information in neural responses under-exploited. To address this gap, we formulate a mechanism-paradigm correspondence view linking hierarchical visual processing (from appearance to semantics) to the complementary biases of cross-paradigm VFMs, with image-only models (e.g., DINO) and image--text models (e.g., CLIP) capturing distinct facets of this hierarchy. Guided by this view, we propose the Dual-Teacher Hierarchical Alignment (DTHA) framework. Concretely, a shared EEG encoder is supervised by frozen multi-blur CLIP and shallow DINOv2 teachers via dual-branch contrastive alignment, so each branch fully supervises the visual cues it emphasizes. A fused-retrieval objective learns adaptive weights to combine their complementary visual representations for zero-shot EEG-to-image retrieval. For the shared EEG encoder, we introduce the Attentive Modulated Scalp-Pattern Bank Encoder (AMSPBE), combining temporo-spatial attention with a learnable scalp-pattern bank for accurate, interpretable visual feature extraction, while modeling regional temporal dynamics and filter-bank topographies. On THINGS-EEG zero-shot retrieval, DTHA achieves 92.4% Top-1 accuracy, exceeding prior state of the art by 12.4 percentage points. Spatial, spectral, and representational-similarity analyses, together with THINGS-MEG evaluation, validate DTHA's effectiveness, biological plausibility, and cross-modality generalization. This complementarity-driven perspective shows that supervision from cross-paradigm VFMs enables more complete extraction of hierarchical visual information from brain activity.
Nov 9, 2026

UniOMA: Unified Optimal-Transport Multi-Modal Structural Alignment for Robot Perception

Achieving generalizable and well-aligned multimodal representations remains a core challenge in robot learning. While recent approaches have attempted to align modalities by modeling conditional or higher-order statistical dependencies, they do not explicitly consider the structural coherence across modalities. In this work, we propose a novel multimodal alignment method that augments existing contrastive losses with a geometry-aware Gromov--Wasserstein (GW) distance-based regularization. To this end, we encode intra-modal geometry for each modality and reduce cross-modal structural discrepancies through a dynamically learned GW barycenter, thereby encouraging structural consistency in addition to instance-level contrastive alignment. We apply this optimal-transport-based alignment strategy to robot perception tasks involving underexplored modalities such as force and tactile signals, where modality data often exhibit varying sample densities. Extensive experiments on five robotic benchmarks show that our method improves structural alignment and downstream performance over state-of-the-art multimodal contrastive baselines across a wide range of downstream tasks, including robot and environment state prediction. Moreover, our GW-based regularization can be seamlessly integrated into most InfoNCE-like objectives.
Nov 9, 2026

Good Views Are Not Enough: Global Counterexamples for Gaussian-Regularized JEPAs

Good views do not suffice to guarantee scene invariance or uniqueness up to rotation at the global optima of Gaussian-regularized JEPAs. We give globally certified counterexamples showing that the uniqueness premise of the Gaussian extension in Good Views fails for specified soft objectives. Gaussian marginal matching can reward retaining independently resampled observation variation, and this benefit can exceed its prediction cost. In a four-state model, comparison against the entire scene-only class proves that every global minimizer splits a scene and admits an equal-value partner outside its rotation orbit. In a 512-dimensional model with a complete scene basis, a spectral gap, and exact population whitening, pooled Epps-Pulley regularization still forces every global minimizer to retain nuisance and yields positive cross-modal alignment error even after the best rotation when modality nuisances are independent. Better scene alignment can also accompany worse prediction. With matched 512-label readout budgets and 1% observation noise, whitened prediction JEPA aligns better to the scene than next-token prediction but incurs 0.2872 nats/token higher linear-readout test NLL. A Bayes-risk decomposition distinguishes information lost through the observation channel from the fitted readout's difficulty using retained information. These results connect three determinants of learned representations: what views share, what regularization favors, and what a prediction task can use.
Nov 6, 2026

Flow-Guided Target-Space Alignment via Path Consistency

Many multimodal alignment problems require mapping a source modality into a fixed space defined by a target modality, where the target space is not merely an intermediate embedding but is used directly for downstream analysis. In this setting, successful alignment should preserve both the global structure of the target distribution and the instance-level correspondence observed in paired source--target training data. Existing objectives typically address only part of this goal: contrastive losses encourage relative similarity, distribution-matching losses improve marginal agreement, and pointwise regression fits each source sample to its paired target without explicitly modeling target-space path geometry. We propose FlowAlign, a flow-guided target-space alignment method that repurposes conditional flow matching as a training-time alignment signal rather than an inference-time transport model. FlowAlign uses the learned flow field to guide the alignment of source-side representations with the fixed target space, encouraging them to preserve paired correspondence while remaining compatible with the global target-space structure. We further provide theoretical analysis showing how FlowAlign helps improve paired correspondence by controlling endpoint accuracy and capturing local target-space geometry. Experiments on synthetic data, paired single-cell multiome benchmarks, and image--text alignment demonstrate that FlowAlign improves instance level correspondence while preserving target-space structure across different modality settings.
Nov 3, 2026

BrainBind: Bridging Synchronized EEG and fMRI through Long-Form Naturalistic Video Representations

Real-world vision is dynamic and scene-rich, while EEG and fMRI resolve it at complementary temporal and spatial scales. Most neural decoding systems, however, align one modality in isolation and mostly against static images or short clips. We study the synchronized case: EEG and fMRI are recorded during the same long-form naturalistic viewing, giving a shared stimulus but also recurring narrative context and many semantically similar candidates. We present BrainBind, a framework that maps synchronized EEG and fMRI into frozen long-form video embedding spaces and treats those spaces as common visual-semantic references, not literal neural codes. On CineBrain, a five-seed Recall@K 2 x 4 matrix across two video anchors and four system configurations shows distinct retrieval profiles. With Omni, fMRI organizes the full candidate list most strongly (MRR = 23.82, nDCG@10 = 25.31), while Joint-Dual gives the highest mean label-aware R@1 (13.52%) with a distinct rank-wide profile. The complementary episode-disjoint evaluation quantifies the contribution of repeated stimulus identity and narrative context to the random split. Across these experiments, proxy choice and system configuration produce distinct top-rank and rank-wide retrieval profiles within the present synchronized dataset. The code is available at https://github.com/Forainest789/Brainbind.
Nov 3, 2026

The Platonic Brain: Representational Convergence in EEG Foundation Models

The Platonic Representation Hypothesis (PRH) predicts that artificial neural networks converge on shared representations as scale and performance increase. We test whether these predictions hold for representations of the human brain with EEG foundation models. Using EEG recordings from six participants watching a television series, together with aligned visual and language content, we tested whether EEG foundation-model representations converge within and across modalities as their embeddings become better at decoding an informative set of EEG markers. We found that EEG models with greater decodability showed stronger cross-family alignment in both local neighbourhood similarity and global structure, not consistently increasing with model parameter count. Cross-modal convergence was modality and metric dependent: greater EEG decodability was associated with stronger global alignment to language, but not consistently to vision models, while local alignment showed no consistent association. This dissociation may reflect local sensitivity to low-level sensory features, whereas global geometry may capture weaker shared semantic structure.
Nov 3, 2026

ProAlign: Progressive Positional and Prototype-Guided Alignment for Aerial-Ground Person Re-Identification

Aerial-Ground Person Re-Identification (AG-ReID) aims to match pedestrian images captured by unmanned aerial vehicles (UAVs) and ground-based surveillance cameras. The task remains highly challenging due to severe viewpoint discrepancies, frequent occlusions, and substantial domain gaps between aerial and ground imagery. Beyond these observable factors, existing methods often overlook two fundamental structural issues: cumulative positional drift during hierarchical feature extraction and semantic inconsistency across cross-view feature distributions. To address these challenges, we propose ProAlign, a Progressive Positional and Prototype-Guided Alignment Network for AG-ReID. ProAlign comprises two key components: a Layer-wise Progressive Positional Embedding (LPPE) module and a View-Aware Prototype Contrastive Learning (PCL) module. LPPE performs adaptive positional calibration throughout transformer layers by employing conditional positional encoding in shallow layers to mitigate local spatial distortions, while introducing learnable static positional embeddings in deeper layers to reinforce global semantic priors. Meanwhile, PCL maintains view-specific aerial and ground prototypes for each identity and enforces cross-view semantic consistency via a dual-prototype contrastive objective. Extensive experiments on CARGO and two real-world AG-ReID benchmarks demonstrate the effectiveness of ProAlign, achieving state-of-the-art overall retrieval performance on CARGO with improvements of 1.60% in Rank-1 accuracy and 5.01% in mAP over the strongest competing methods.
Nov 2, 2026

Unimodal Augmentation-based SSL Objectives Induce Cross-Modal Alignment

Recent works have observed implicit cross-modal alignment between independently trained unimodal models. However, the mechanisms underlying the emergence of this phenomenon remain unresolved. In this paper, we demonstrate that unimodal augmentation-based self-supervised objectives implicitly induce cross-modal alignment. We show that low unimodal losses force high cross-modal alignment, whereas a high unimodal loss prevents such alignment. We theoretically connect the widely used InfoNCE loss to local CKA, the popular measure of cross-modal alignment. We further provide extensive empirical evidence supporting this claim, both in controlled settings and across 53 unconstrained frozen text, vision, and audio unimodal encoders. In both cases, unimodal augmentation-based objectives control cross-modal alignment. Overall, our results provide a theoretical and empirical foundation for understanding the emergence of implicit cross-modal alignment by establishing a direct connection between unimodal training objectives and cross-modal structure.
Oct 30, 2026

IsoAlign: Isoform-Aware DNA-Protein Alignment via Joint-Embedding Predictive Architectures

Genomic foundation models (GFMs) have emerged as a powerful paradigm for learning DNA sequence representations, yet existing self-supervised pretraining strategies overlook alternative splicing and underperform compared to specialized models. To address this gap, we introduce IsoAlign, a multimodal framework that grounds genomic pretraining in isoform-aware DNA-protein alignment to learn biologically meaningful representations. IsoAlign aligns transcripts to their corresponding protein isoforms, capturing both coding sequences and broader regulatory contexts, using two Joint-Embedding Predictive Architectures (JEPA): DNA-Protein JEPA (DP-JEPA), which predicts protein representations from DNA using contrastive alignment, and Isoform-Masked JEPA (IM-JEPA), which models relationships among alternative isoforms of the same gene through masked protein prediction. We evaluate the learned DNA representations of these architectures across a multi-metric evaluation suite spanning tasks, including the prediction of pathogenicity, fitness, gene expression, and TF binding. Our results demonstrate that isoform-aware cross-modal alignment using JEPA significantly improves functional representations and downstream performance, establishing a promising direction for GFM pretraining.
Oct 30, 2026

From Fragmentation to Fusion: A Survey and Roadmap for Omnimodal Embeddings

Embedding models underpin virtually every modern AI system, from search engines to recommendation systems. Despite shared mathematical foundations, research on embedding models remains fragmented by modality: the communities develop largely in isolation, seldom exchanging insights. We argue that this fragmentation obscures a deeper convergence. Driven by contrastive learning, Transformer architectures, and a granularity shift, embedding research across modalities is converging toward a common set of principles, even as modality-specific challenges remain underexplored. To reason across this convergence without conflating it, we propose a framework of three progressively harder capabilities: coverage, comparability, and compositionality. Three paths toward unification satisfy these in turn: transformation broadens coverage, alignment enables comparability, and joint embedding, the terminus we advocate, aims at compositionality. This position paper traces the resulting evolution, providing the first unified, cross-cutting treatment of omnimodal embedding research.
Oct 29, 2026

CrossWeave: Cross-Modal Scene and Instance Retrieval from Sparse 2D-3D Alignment

Cross-modal scene and instance retrieval aims to retrieve a corresponding 3D environment or object given query 2D images, serving as a foundational mechanism for spatial reasoning in robotics and AR/VR. Existing methods achieve this by learning alignment only at the level of whole scenes or objects, which provides no supervision for the fine-grained local structure within them. In this work, we show that supervision on a set of sparse 2D-3D correspondences, without scene- or instance-level contrastive objectives, is sufficient to train a single model for both scene- and instance-level retrieval. Specifically, we propose CrossWeave\textbf{CrossWeave}, a novel framework for cross-modal retrieval that learns sparse patch-point alignment, which alone outperforms existing methods. An optimal transport-based aggregator over the aligned features gives further gains by clustering structurally redundant regions, freeing descriptor capacity for semantically distinct local features. CrossWeave achieves state-of-the-art performance on both scene- and instance-level retrieval on ScanNet and 3RScan with a single model\textit{single model}, whereas existing methods require separate models for each task. Our results suggest that whole-scene encoding is unnecessary for cross-modal retrieval and that sparse local correspondence supervision is a more effective and flexible alternative.
Oct 29, 2026

Proving and Improving the Platonic Representation Hypothesis with JEPAs

Whether models trained on different modalities converge to the same representation of the world is central to AI. The Platonic Representation Hypothesis (PRH) conjectures that they do as they scale, but without a proof it offers neither a guarantee nor a recipe. We turn the hypothesis into a theorem. First, we prove a Platonic Representation Theorem: Joint-Embedding Predictive Architectures (JEPAs) trained separately on any number of modalities learn the same representation up to one rotation whenever their positive pairs are two observations of one scene--good views--and their misalignment is bounded otherwise. Second, we show that next-token prediction is a JEPA with bad views, parts of a single document, which explains why language models align with vision only through scale. Third, we validate the theory from 145k captions to the texts of 940M web images, and uncover a scaling law of alignment that predicts unseen runs within seed noise. Good views need 42 times fewer training samples than next-token prediction on the same corpus to reach the same alignment, and a 109M-parameter text JEPA aligns with DINOv2 and CLIP's image encoders better than every language model we evaluated, up to 72B parameters, and than CLIP's text encoder: next-token prediction is a remarkably inefficient route to platonic representations. As a showcase, one rotation turns separately trained encoders into an image captioner (CIDEr .838 on COCO). Even that rotation needs no pair: on the latest pair-free benchmark, our text JEPA beats every zero-shot accuracy reported with language models up to 8B parameters, e.g., 64% against at most 56% top-5 on ImageNet-100. Positive views, and not scale alone, are thus the lever to build platonic representations.
Oct 29, 2026

Uncertainty-Guided Cross-Modal Alignment for Diffusion-Based IoT Intrusion Detection

Diffusion models evolve noisy states toward a learned data distribution, so their sampling behavior depends on the geometry of the representation in which denoising occurs. This dependence is especially important for multimodal data: independently trained encoders can map synchronized views to incompatible latent regions, while uniform alignment can let an uncertain view distort the shared geometry. To address this gap, we introduce a novel reliability-guided alignment rule that uses normalized VAE posterior variance to control directional cross-modal influence before diffusion training. We study this issue in minority-attack generation for imbalanced Internet-of-Things intrusion detection. Tabular flow attributes, a DeepInsight image, and a packet sequence are encoded by modality-specific variational autoencoders. Per-modality normalized VAE posterior variances provide bounded, sample-specific weighting signals in an asymmetric stop-gradient alignment objective. We use these quantities as heuristic encoder-derived reliability proxies rather than calibrated probabilistic uncertainties. The full modality-specific latent means obtained after alignment fine-tuning are standardized and modeled jointly by a class-conditional EDM; a second-order sampler then generates paired attack-conditioned latent blocks for the three decoders. In controlled experiments on TON-IoT, uncertainty-aware alignment obtains the highest mean precision, recall, density, and coverage among three diffusion variants. Augmenting only the training split yields ROC-AUC 0.9515, PR-AUC 0.8954, balanced accuracy 0.8395, and attack-class F1 0.7940 on an untouched real test set. Among the two variants with matched alignment fine-tuning, posterior-variance-based weighting improves both tabular generative fidelity and downstream detection relative to uniform weighting.
Oct 28, 2026

BASILISC: Bayesian Alignment of Siamese-Integrated Latents for Inferring Single-cell Correspondence

Single-cell assays study the genome one cell at a time. Unimodal assays measure one molecular layer, the newer multimodal assays measure several. Most data are single-modality, so integrating them relies on models trained on multimodal assays, where the cell correspondence across modalities is known. These models return a matching, but rarely a confidence measure. We present BASILISC (Bayesian Alignment of Siamese-Integrated Latents for Inferring Single-cell Correspondence), a framework for matching single-cell representations with uncertainty on the correspondence itself. A BASILISC model is a probabilistic siamese network over the two modalities with a predictive distribution over the correspondence for a cross-modal pair of cells. We instantiate it at two levels: at the cell-type level the question is whether two cells share a cell type, at the cell level which cell is the partner in the other modality. To our knowledge, this is the first uncertainty score on a learned cross-modal correspondence evaluated against a known ground-truth pairing. BASILISC learns a similarity across modalities that reflects cell lineage: an unseen cell type is matched to its lineage relative, matching still works on such held-out types (balanced accuracy 0.86 on unseen against 0.93 on seen types), and the predictive uncertainty rises on these out-of-distribution cells. The uncertainty score also selects high-quality matches for downstream analysis: discarding the least certain fifth of the pairs raises accuracy from 0.93 to 0.98. At the cell level, a deep ensemble of the deterministic matcher reaches 0.062 matching accuracy on the benchmark test set, above the competition winner; on the Bayesian matcher's assignment, the posterior flags the wrong matches that cross a cell-type boundary better than a deterministic read of the same scores (AUROC 0.68 against 0.64).