Cross-Modal Representation Learning

Overview

The field studies how models can represent different kinds of data so that related observations can be matched, combined, or used to transfer knowledge. Common approaches include CLIP-style contrastive alignment, masked cross-modal reconstruction and predictive learning, teacher–student distillation, and methods that separate shared from modality-specific features; anchors and optimal transport can help when paired examples are scarce. These representations support image-text retrieval, cross-sensor recognition, and transfer across domains or missing-modality settings. A central tension is that shared spaces can improve transfer but discard useful modality-specific structure, and apparent semantic alignment does not always imply fine-grained correspondence; the best choice between alignment and prediction depends on data quality and cross-modal relationships.

Latest papers

224 papers17 papers in the last four weeks, down 15% on the four weeks before. 0.1% of all new papers.

Feb 23, 2026

CLCR: Cross-Level Semantic Collaborative Representation for Multimodal Learning

Multimodal learning aims to capture both shared and private information from multiple modalities. However, existing methods that project all modalities into a single latent space for fusion often overlook the asynchronous, multi-level semantic structure of multimodal data. This oversight induces semantic misalignment and error propagation, thereby degrading representation quality. To address this issue, we propose Cross-Level Co-Representation (CLCR), which explicitly organizes each modality's features into a three-level semantic hierarchy and specifies level-wise constraints for cross-modal interactions. First, a semantic hierarchy encoder aligns shallow, mid, and deep features across modalities, establishing a common basis for interaction. And then, at each level, an Intra-Level Co-Exchange Domain (IntraCED) factorizes features into shared and private subspaces and restricts cross-modal attention to the shared subspace via a learnable token budget. This design ensures that only shared semantics are exchanged and prevents leakage from private channels. To integrate information across levels, the Inter-Level Co-Aggregation Domain (InterCAD) synchronizes semantic scales using learned anchors, selectively fuses the shared representations, and gates private cues to form a compact task representation. We further introduce regularization terms to enforce separation of shared and private features and to minimize cross-level interference. Experiments on six benchmarks spanning emotion recognition, event localization, sentiment analysis, and action recognition show that CLCR achieves strong performance and generalizes well across tasks.
Feb 22, 2026

CREM: Compression-Driven Representation Enhancement for Multimodal Retrieval and Comprehension

Multimodal Large Language Models (MLLMs) have shown remarkable success in comprehension tasks such as visual description and visual question answering. However, their direct application to embedding-based tasks like retrieval remains challenging due to the discrepancy between output formats and optimization objectives. Previous approaches often employ contrastive fine-tuning to adapt MLLMs for retrieval, but at the cost of losing their generative capabilities. We argue that both generative and embedding tasks fundamentally rely on shared cognitive mechanisms, specifically cross-modal representation alignment and contextual comprehension. To this end, we propose CREM (Compression-driven Representation Enhanced Model), with a unified framework that enhances multimodal representations for retrieval while preserving generative ability. Specifically, we introduce a compression-based prompt design with learnable chorus tokens to aggregate multimodal semantics and a compression-driven training strategy that integrates contrastive and generative objectives through compression-aware attention. Extensive experiments demonstrate that CREM achieves state-of-the-art retrieval performance on MMEB while maintaining strong generative performance on multiple comprehension benchmarks. Our findings highlight that generative supervision can further improve the representational quality of MLLMs under the proposed compression-driven paradigm.
Feb 22, 2026

MoBind: Motion Binding for Fine-Grained IMU-Video Pose Alignment

We aim to learn a joint representation between inertial measurement unit (IMU) signals and 2D pose sequences extracted from video, enabling accurate cross-modal retrieval, temporal synchronization, subject and body-part localization, and action recognition. To this end, we introduce MoBind, a hierarchical contrastive learning framework designed to address three challenges: (1) filtering out irrelevant visual background, (2) modeling structured multi-sensor IMU configurations, and (3) achieving fine-grained, sub-second temporal alignment. To isolate motion-relevant cues, MoBind aligns IMU signals with skeletal motion sequences rather than raw pixels. We further decompose full-body motion into local body-part trajectories, pairing each with its corresponding IMU to enable semantically grounded multi-sensor alignment. To capture detailed temporal correspondence, MoBind employs a hierarchical contrastive strategy that first aligns token-level temporal segments, then fuses local (body-part) alignment with global (body-wide) motion aggregation. Evaluated on mRi, TotalCapture, and EgoHumans, MoBind consistently outperforms strong baselines across all four tasks, demonstrating robust fine-grained temporal alignment while preserving coarse semantic consistency across modalities. Code is available at https://github.com/bbvisual/ MoBind.
Feb 19, 2026

SpectralGCD: Spectral Concept Selection and Cross-modal Representation Learning for Generalized Category Discovery

Generalized Category Discovery (GCD) aims to identify novel categories in unlabeled data while leveraging a small labeled subset of known classes. Training a parametric classifier solely on image features often leads to overfitting to old classes, and recent multimodal approaches improve performance by incorporating textual information. However, they treat modalities independently and incur high computational cost. We propose SpectralGCD, an efficient and effective multimodal approach to GCD that uses CLIP cross-modal image-concept similarities as a unified cross-modal representation. Each image is expressed as a mixture over semantic concepts from a large task-agnostic dictionary, which anchors learning to explicit semantics and reduces reliance on spurious visual cues. To maintain the semantic quality of representations learned by an efficient student, we introduce Spectral Filtering which exploits a cross-modal covariance matrix over the softmaxed similarities measured by a strong teacher model to automatically retain only relevant concepts from the dictionary. Forward and reverse knowledge distillation from the same teacher ensures that the cross-modal representations of the student remain both semantically sufficient and well-aligned. Across six benchmarks, SpectralGCD delivers accuracy comparable to or significantly superior to state-of-the-art methods at a fraction of the computational cost. The code is publicly available at: https://github.com/miccunifi/SpectralGCD.
Feb 5, 2026

Semantic Purification for Conditional Representation Learning

Conditional representation learning aims to extract criterion-specific features for customized tasks. Recent methods construct conditional subspaces spanned by criterion-specific text bases in the embedding space of vision-language models (VLMs). Image embeddings are then projected onto these subspaces to obtain conditional representations. However, since VLMs are not explicitly trained to disentangle semantics associated with different criteria, the corresponding conditional subspaces remain coupled. This coupling induces semantic leakage during projection, thereby degrading the semantic purity of conditional representations. To suppress semantic leakage, we propose Semantic Purification for Conditional Representation Learning (SP-CRL). Specifically, SP-CRL first decomposes the original text basis and performs curvature-based adaptive truncation on the resulting basis vectors to construct a purer conditional subspace. It then identifies an appropriate noise subspace and projects image embeddings onto its null space to remove irrelevant semantic components. Extensive experiments across customized clustering, customized few-shot classification, and customized retrieval tasks demonstrate that SP-CRL achieves state-of-the-art performance with superior generalization.
Feb 2, 2026

Superman: Unifying Skeleton and Vision for Human Motion Perception and Generation

Human motion analysis tasks, such as temporal 3D pose estimation, motion prediction, and motion in-betweening, play an essential role in computer vision. However, current paradigms suffer from severe fragmentation. First, the field is split between perception'' models that understand motion from video but only output text, and generation'' models that cannot perceive from raw visual input. Second, generative MLLMs are often limited to single-frame, static poses using dense, parametric SMPL models, failing to handle temporal motion. Third, existing motion vocabularies are built from skeleton data alone, severing the link to the visual domain. To address these challenges, we introduce Superman, a unified framework that bridges visual perception with temporal, skeleton-based motion generation. Our solution is twofold. First, to overcome the modality disconnect, we propose a Vision-Guided Motion Tokenizer. Leveraging the natural geometric alignment between 3D skeletons and visual data, this module pioneers robust joint learning from both modalities, creating a unified, cross-modal motion vocabulary. Second, grounded in this motion language, a single, unified MLLM architecture is trained to handle all tasks. This module flexibly processes diverse, temporal inputs, unifying 3D skeleton pose estimation from video (perception) with skeleton-based motion prediction and in-betweening (generation). Extensive experiments on standard benchmarks, including Human3.6M, demonstrate that our unified method achieves state-of-the-art or competitive performance across all motion tasks. This showcases a more efficient and scalable path for generative motion analysis using skeletons.
Jan 31, 2026

Cross-Modal Visuo-Tactile Representation Learning with Action Chunking Transformers for Contact-Rich Manipulation

Tactile feedback is important for contact-rich robotic manipulation, yet effective use of tactile observations remains challenging when tactile signals are image-like, hardware-dependent, and only weakly aligned with external visual observations. This study addresses this representation-learning problem by proposing a visuo-tactile contrastive learning framework for imitation-based manipulation. The method aligns external RGB observations and calibrated tactile images in a shared embedding space using a CLIP-style objective, and integrates the resulting representation into an Action Chunking Transformer (ACT) policy. A low-cost visuo-tactile gripper (LVTG) is proposed to provide a modular and durable sensing platform for reproducible data collection, supplying tactile observations that can be used by downstream manipulation algorithms. Experiments on contact-rich manipulation tasks show that tactile feedback improves the average task completion rate from 30% for a vision-only ACT baseline to 42%, and that the proposed contrastive pretraining further increases the completion rate to 54%. These results indicate that explicitly aligning visual and tactile observations provides more useful contact-aware features for downstream policy learning than directly adding tactile images without pretraining.
Sep 24, 2025

PPGFlowECG: Latent Rectified Flow with Cross-Modal Encoding for PPG-Guided ECG Generation and Cardiovascular Disease Detection

Electrocardiography (ECG) is the clinical gold standard for cardiovascular disease (CVD) assessment, yet continuous monitoring is constrained by the need for dedicated hardware and trained personnel. Photoplethysmography (PPG) is ubiquitous in wearable devices and readily scalable, but it lacks electrophysiological specificity, limiting diagnostic reliability. While generative methods aim to translate PPG into clinically useful ECG signals, existing approaches are limited by the misalignment of physiological semantics in generative models and the complexity of modeling in high-dimensional signals. To address these limitations, we propose PPGFlowECG, a two-stage framework that aligns PPG and ECG in a shared latent space using the CardioAlign Encoder and then synthesizes ECGs with latent rectified flow. We further provide a formal analysis of this coupling, showing that the CardioAlign Encoder is necessary to guarantee stable and semantically consistent ECG synthesis under our formulation. Extensive experiments on four datasets demonstrate improved synthesis fidelity and downstream diagnostic utility. These results indicate that PPGFlowECG supports scalable, wearable-first CVD screening when standard ECG acquisition is unavailable.
Aug 17, 2025

Inverse-LLaVA: Rethinking Multimodal Alignment via Text-to-Vision Mapping

Connecting pretrained vision and language models usually involves projecting image features into the language model's input space. Inverse-LLaVA reverses this mapping within decoder attention: language states are projected to the visual feature dimension, and modality-specific maps produce residual query, key, and value updates. Fusion and low-rank adaptation (LoRA) learn jointly from 665K visual instructions, with frozen backbones and no separate alignment stage. Across nine primary benchmark evaluations, the final 7B model approaches two-stage LLaVA-1.5 on several tasks. It scores 78.45% on VQAv2 versus 79.13% for official LLaVA-LoRA, and 50.96% versus 48.56% on VizWiz; TextVQA is lower at 56.96% versus 58.47%. Controlled studies examine fusion components, insertion depth, visual features, and language-model size. Representation analysis shows that the text maps preserve much of the pairwise similarity ordering while changing its geometric spread. Additional paired supervision improves celebrity recognition, while instruction replay repairs caption-induced answer-format failures. Analytical cost expressions and fixed-work profiles separate the additional fusion computation from the omitted alignment stage. These findings establish text-to-vision attention fusion as a practical alternative for instruction-only multimodal adaptation.
Aug 1, 2025

Platonic Representations for Poverty Mapping: Unified Vision-Language Codes or Agent-Induced Novelty?

We investigate whether socioeconomic indicators, like household wealth, leave recoverable informational imprints in both satellite imagery (capturing features like buildings and roads) and Internet-sourced text (reflecting historical, cultural, and narratives of neighborhoods). Using DHS data from African neighborhoods (clusters), we pair high-resolution Landsat images with textual descriptions generated by LLMs conditioned on location/year, plus text retrieved by an LLM-driven AI Search Agent from web sources. We develop a multimodal framework that predicts household wealth (International Wealth Index; IWI) via five pipelines: (i) a vision model on satellite images, (ii) an LLM using only location and year, (iii) an AI agent that searches and synthesizes web text, (iv) a joint image-text encoder, and (v) an ensemble of all signals. Our framework yields three contributions. First, evaluations show that fusing vision and agent/LLM-generated text improves on vision-only baselines in wealth prediction (e.g., R-squared of 0.77 vs. 0.63 on out-of-sample splits), with LLM-internal knowledge (artificial neural memory) proving surprisingly predictive in out-of-country/time generalization. Second, we find suggestive evidence of partial representational alignment: fused embeddings from vision and language modalities correlate moderately (median cosine similarity across modalities of about 0.60 after alignment). This pattern is broadly consistent with the Platonic Representation Hypothesis, but does not by itself establish convergence to a single shared latent representation. Because agent-retrieved data yields only marginal and unstable gains across splits, our evidence for the Agent-Induced Novelty Hypothesis is limited. Third, we release a large-scale multimodal dataset of about 60,000 DHS clusters, each linked to satellite images, LLM-generated descriptions, and AI-agent-retrieved texts.
Jun 19, 2025

Heterogeneous-Modal Unsupervised Domain Adaptation via Latent Space Bridging

Unsupervised domain adaptation (UDA) effectively bridges the domain gap between a labeled source domain and an unlabeled target domain, but assumes that the two domains share the same modality. Heterogeneous domain adaptation (HDA) instead handles different feature spaces across domains, yet requires labeled target samples or paired data linking the source and target domains. Neither applies when a labeled source domain and a fully unlabeled target domain each hold an entirely distinct modality (e.g., 2D images and 3D point clouds). To address this limitation, we introduce a new setting termed Heterogeneous-Modal Unsupervised Domain Adaptation (HMUDA), which transfers knowledge across modalities via an unlabeled bridge domain containing paired observations from both modalities, whose distribution may deviate from those of the source and target domains. To learn under the HMUDA setting, we propose Latent Space Bridging (LSB), a dual-branch framework where a feature consistency loss on paired bridge samples closes the modality gap and a class-centroid alignment loss reduces the source-target discrepancy. Extensive experiments on eight benchmark settings covering both 2D-to-3D and 3D-to-2D transfer demonstrate that LSB achieves state-of-the-art performance.
Jun 10, 2025

Multimodal Representation Alignment for Cross-modal Information Retrieval

Different machine learning models can represent the same underlying concept in different ways. This variability is particularly valuable for in-the-wild multimodal retrieval, where the objective is to identify the corresponding representation in one modality given another modality as input. This challenge can be effectively framed as a representation alignment problem. For example, given a sentence encoded by a language model, retrieve the most semantically aligned image based on representations produced by an image encoder, or vice versa. To gain insights into the performance impact of different metrics, embedding spaces, and representation alignment for retrieval tasks, we first empirically investigate the geometric relationships between visual and textual embeddings derived from both vision-language models and combined unimodal models. We then align these representations using four standard similarity metrics as well as two learned ones, implemented via neural networks of different architectures with varying losses across multiple benchmarks. Our experimental findings indicate that cosine similarity consistently outperforms all the investigated metrics in representation alignment tasks, and that Wasserstein distance provides a complementary perspective on cross-modal distributional differences. We also observe that our proposed custom contrastive loss is advantageous over the MSE loss for aligning image and text representations, for both multilayer perceptrons and transformer-based models. Taken together, our findings offer novel insights and practical considerations for researchers working in multimodal information retrieval, particularly in real-world, cross-modal applications. Our code is publicly available.
May 21, 2025

A quantitative analysis of semantic information in deep representations of text and images

It was recently observed that the representations of different models that process identical or semantically related inputs tend to align. We analyze this phenomenon using the Information Imbalance, an asymmetric rank-based measure that quantifies the capability of a representation to predict another, providing a proxy of the cross-entropy which can be computed efficiently in high-dimensional spaces. By measuring the Information Imbalance between representations generated by DeepSeek-V3 processing translations, we find that semantic information is spread across many tokens, and that semantic predictability is strongest in a set of central layers of the network, robust across six language pairs. We measure clear information asymmetries: English representations are systematically more predictive than those of other languages, and DeepSeek-V3 representations are more predictive of those in a smaller model such as Llama3-8b than the opposite. In the visual domain, we observe that semantic information concentrates in middle layers for autoregressive models and in final layers for encoder models, and these same layers yield the strongest cross-modal predictability with textual representations of image captions. Our results support the hypothesis of semantic convergence across languages, modalities, and architectures, while showing that directed predictability between representations varies strongly with layer-depth, model scale, and language.
Apr 14, 2025

On the Value of Cross-Modal Misalignment in Multimodal Representation Learning

Multimodal representation learning, exemplified by multimodal contrastive learning (MMCL) using image-text pairs, aims to learn powerful representations by aligning cues across modalities. This approach relies on the core assumption that the exemplar image-text pairs constitute two representations of an identical concept. However, recent research has revealed that real-world datasets often exhibit cross-modal misalignment. There are two distinct viewpoints on how to address this issue: one suggests mitigating the misalignment, and the other leveraging it. We seek here to reconcile these seemingly opposing perspectives and to provide a practical guide for practitioners. Using latent variable models, we thus formalize cross-modal misalignment by introducing two specific mechanisms: selection bias, where some semantic variables are absent in the text, and perturbation bias, where semantic variables are altered---both leading to misalignment in data pairs. Our theoretical analysis shows that, under suitable assumptions, population global minimizers of an alignment--entropy surrogate yield representations that depend only on the semantic variables unaffected by selection and perturbation biases and locally identify this subset almost surely. Extensive empirical studies on synthetic data and real image-text datasets further examine how cross-modal misalignment shapes semantic predictability and downstream performance, shedding light on its nuanced impact on multimodal representation learning.