方法证据记录
Multimodal Word2Vec
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.
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 Word2Vec (Cross-Modal Distributional Semantics)
分类方法记录 · ml-model / deep-learning
- Bruni, E., Tran, N.-K., & Baroni, M. (2014). Multimodal Distributional Semantics. Journal of Artificial Intelligence Research, 49, 1–47. · DOI 10.1613/jair.4135
- Mikolov, T., Sutskever, I., Chen, K., Corrado, G., & Dean, J. (2013). Distributed Representations of Words and Phrases and their Compositionality. Advances in Neural Information Processing Systems (NIPS), 26. · 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。