Abstract:Fine-grained representation is fundamental to species classification based on deep learning, and in this context, cross-modal contrastive learning is an effective method. The diversity of species coupled with the inherent contextual ambiguity of natural language poses a primary challenge in the cross-modal representation alignment of conservation area image data. Integrating cross-modal retrieval tasks with generation tasks contributes to cross-modal representation alignment based on contextual understanding. However, during the contrastive learning process, apart from learning the differences in the data itself, a pair of encoders inevitably learns the differences caused by encoder fluctuations. The latter leads to convergence shortcuts, resulting in poor representation quality and an inaccurate reflection of the similarity relationships between samples in the original dataset within the shared space of features. To achieve fine-grained cross-modal representation alignment, we first propose a residual attention network to enhance consistency during momentum updates in cross-modal encoders. Building upon this, we propose momentum encoding from a multi-task perspective as a bridge for cross-modal information, effectively improving cross-modal mutual information, representation quality, and optimizing the distribution of feature points within the cross-modal shared semantic space. By acquiring momentum encoding queues for cross-modal semantic understanding through multi-tasking, we align ambiguous natural language representations around the invariant image features of factual information, alleviating contextual ambiguity and enhancing model robustness. Experimental validation shows that our proposed multi-task perspective of cross-modal momentum encoders outperforms similar models on standardized image classification tasks and image–text cross-modal retrieval tasks on public datasets by up to 8% on the leaderboard, demonstrating the effectiveness of the proposed method. Qualitative experiments on our self-built conservation area image–text paired dataset show that our proposed method accurately performs cross-modal retrieval and generation tasks among 8142 species, proving its effectiveness on fine-grained cross-modal image–text conservation area image datasets.

On the diversity of multi-head attention

Multi-Head Attention with Disagreement Regularization

Select & Re-Rank: Effectively and Efficiently Matching Multimodal Data with Dynamically Evolving Attention

Disagreement Matters: Exploring Internal Diversification for Redundant Attention in Generic Facial Action Analysis

Self-attention Multi-view Representation Learning with Diversity-promoting Complementarity

Pay Better Attention to Attention: Head Selection in Multilingual and Multi-Domain Sequence Modeling

Generating Diverse Translation by Manipulating Multi-Head Attention

Finding the Pillars of Strength for Multi-Head Attention

On the Optimization and Generalization of Multi-head Attention

Multi-agent Reinforcement Learning with Multi-head Attention

Attention Diversification for Domain Generalization

Celebrating Diversity in Shared Multi-Agent Reinforcement Learning

Repulsive Attention: Rethinking Multi-head Attention as Bayesian Inference

A Diversified Attention Model for Interpretable Multiple Clusterings

Multi-Head Attention: Collaborate Instead of Concatenate

MoH: Multi-Head Attention as Mixture-of-Head Attention

Get The Point of My Utterance! Learning Towards Effective Responses with Multi-Head Attention Mechanism

Diversity matters: Cross-head mutual mean-teaching for semi-supervised medical image segmentation

Fine-Grained Cross-Modal Semantic Consistency in Natural Conservation Image Data from a Multi-Task Perspective

AttentionX: Exploiting Consensus Discrepancy In Attention from A Distributed Optimization Perspective

Multimodal sentiment analysis based on multi-head attention mechanism