手法を比較

	マルチモーダルDoc2Vec	マルチモーダル Word2Vec
分野	深層学習	深層学習
系統	Machine learning	Machine learning
提唱年	2014–2017	2014
提唱者	Le, Q. V. & Mikolov, T. (Doc2Vec core); multimodal extensions by various authors post-2014	Bruni, E., Tran, N.-K., & Baroni, M. (building on Mikolov et al.)
種類	Multimodal document embedding	Multimodal word embedding model
原典	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.	Bruni, E., Tran, N.-K., & Baroni, M. (2014). Multimodal Distributional Semantics. Journal of Artificial Intelligence Research, 49, 1–47.
別名	Multimodal Paragraph Vector, Cross-modal Doc2Vec, Multi-source PV-DM, Multimodal Document Embedding	multimodal word embeddings, visual-linguistic Word2Vec, cross-modal Word2Vec, MM-W2V
関連	6	5
概要	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.	Multimodal Word2Vec extends the classic Word2Vec framework by grounding word representations in perceptual signals — typically image features — alongside distributional text statistics. The result is word vectors that capture both linguistic co-occurrence patterns and visual meaning, enabling richer semantic similarity judgements and better performance on concept-level tasks where purely text-based embeddings fall short.