方法证据记录
Multimodal Doc2Vec
Multimodal Doc2Vec extends the Doc2Vec paragraph-vector framework to incorporate information from more than one modality — typically text alongside images, audio, or structured metadata — producing a shared document-level embedding that captures semantics from multiple sources simultaneously. It is used for cross-modal retrieval, multi-source classification, and document representation where text alone is insufficient.
源记录
引文逐字复制自方法源记录。这些引文不代表任何层级的验证。
Multimodal Doc2Vec (Paragraph Vector with Multi-Source Input)
分类方法记录 · ml-model / deep-learning
- Le, Q. V., & Mikolov, T. (2014). Distributed Representations of Sentences and Documents. Proceedings of the 31st International Conference on Machine Learning (ICML), PMLR 32(2), 1188–1196. · URL
- Ngiam, J., Khosla, A., Kim, M., Nam, J., Lee, H., & Ng, A. Y. (2011). Multimodal Deep Learning. Proceedings of the 28th International Conference on Machine Learning (ICML), 689–696. · URL
精选声明
声明已持久化到证据分类账中，每个声明都有自己的评估。
尚无精选声明
当分类账中没有声明时，此视图不会自行创建声明评估。
相关方法
从方法图中生成，显示为机器建议的关系 — 不推断任何证据声明。