NOSE: Neural Olfactory-Semantic Embedding with Tri-Modal Orthogonal Contrastive Learning

2026-04-12

AI Summary

Existing olfactory representation methods model only isolated segments of the olfactory pathway, lacking a holistic framework that integrates molecular structure, receptor response, and linguistic description, which results in embeddings devoid of biological grounding and semantic interpretability. This work proposes NOSE, a novel framework that, for the first time, jointly models the full olfactory pathway through three modalities—molecular graphs, receptor sequences, and natural language descriptions. By employing orthogonal contrastive learning, NOSE aligns these modalities while disentangling their modality-specific characteristics. Additionally, it introduces a weak positive sampling strategy to mitigate data sparsity in olfactory language annotations. The method achieves state-of-the-art performance across multiple tasks, demonstrates exceptional zero-shot generalization, and yields a representation space that closely aligns with human olfactory perception.

Technology Category

• Machine Learning: Representation Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Lexical Semantics and Morphology

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Olfaction lies at the intersection of chemical structure, neural encoding, and linguistic perception, yet existing representation methods fail to fully capture this pathway. Current approaches typically model only isolated segments of the olfactory pathway, overlooking the complete chain from molecule to receptors to linguistic descriptions. Such fragmentation yields learned embeddings that lack both biological grounding and semantic interpretability. We propose NOSE (Neural Olfactory-Semantic Embedding), a representation learning framework that aligns three modalities along the olfactory pathway: molecular structure, receptor sequence, and natural language description. Rather than simply fusing these signals, we decouple their contributions via orthogonal constraints, preserving the unique encoded information of each modality. To address the sparsity of olfactory language, we introduce a weak positive sample strategy to calibrate semantic similarity, preventing erroneous repulsion of similar odors in the feature space. Extensive experiments demonstrate that NOSE achieves state-of-the-art (SOTA) performance and excellent zero-shot generalization, confirming the strong alignment between its representation space and human olfactory intuition.

Problem

• olfaction
• representation learning
• multi-modal alignment
• semantic interpretability
• biological grounding

Innovation

• tri-modal alignment
• orthogonal contrastive learning
• olfactory representation
• weak positive sampling
• zero-shot generalization