{"url":"/method/multiplicative-attention","slug":"multiplicative-attention","name":"Multiplicative Attention","full_name":"Multiplicative Attention","full_name_withheld":false,"description_markdown":"**Multiplicative Attention** is an attention mechanism where the alignment score function is calculated as:\r\n\r\n$$f_{att}\\left(\\textbf{h}_{i}, \\textbf{s}\\_{j}\\right) = \\mathbf{h}\\_{i}^{T}\\textbf{W}\\_{a}\\mathbf{s}\\_{j}$$\r\n\r\nHere $\\mathbf{h}$ refers to the hidden states for the encoder/source, and $\\mathbf{s}$ is the hidden states for the decoder/target. The function above is thus a type of alignment score function. We can use a matrix of alignment scores to show the correlation between source and target words, as the Figure to the right shows. Within a neural network, once we have the alignment scores, we calculate the final scores using a [softmax](https://paperswithcode.com/method/softmax) function of these alignment scores (ensuring it sums to 1).\r\n\r\nAdditive and multiplicative attention are similar in complexity, although multiplicative attention is faster and more space-efficient in practice as it can be implemented more efficiently using matrix multiplication. Both variants perform similar for small dimensionality $d_{h}$ of the decoder states, but [additive attention](https://paperswithcode.com/method/additive-attention) performs better for larger dimensions. One way to mitigate this is to scale $f_{att}\\left(\\textbf{h}_{i}, \\textbf{s}\\_{j}\\right)$ by $1/\\sqrt{d\\_{h}}$ as with [scaled dot-product attention](https://paperswithcode.com/method/scaled).","description_state":"present","introduced_year":null,"introduced_by":{"title":null,"paper":null,"first_author":null,"n_authors":0,"url_abs":null,"archive_paper_url":null},"source":{"url":"https://ruder.io/deep-learning-nlp-best-practices/","title":"Deep Learning for NLP Best Practices by Sebastian Ruder","url_on_a_paper_host":false},"code_snippet_url":"https://github.com/LukasMut/ATNLP/blob/0aa097062dd5c520cfd163c0bfd0cb4a945c3430/models/Attention.py#L37","code_snippet_url_on_a_code_host":true,"categories":[{"area":"General","area_id":"general","collection":"Attention Mechanisms","url":"/methods/category/attention-mechanisms","pwc_aliases":["attention-mechanisms-1"]}],"n_papers_tagged":6,"archive_num_papers":null,"papers_newest_first":[{"paper":"/paper/hierarchical-attention-models-for-multi","title":"Hierarchical Attention Models for Multi-Relational Graphs","date":"2024-04-14","arxiv_id":"2404.09365","n_code_links":1,"syntology":null},{"paper":"/paper/arithmetic-feature-interaction-is-necessary","title":"Arithmetic Feature Interaction Is Necessary for Deep Tabular Learning","date":"2024-02-04","arxiv_id":"2402.02334","n_code_links":1,"syntology":null},{"paper":null,"title":"Global Pointer: Novel Efficient Span-based Approach for Named Entity Recognition","date":"2022-08-05","arxiv_id":"2208.03054","n_code_links":0,"syntology":null},{"paper":"/paper/deep-learning-based-non-intrusive-multi","title":"Deep Learning-based Non-Intrusive Multi-Objective Speech Assessment Model with Cross-Domain Features","date":"2021-11-03","arxiv_id":"2111.02363","n_code_links":1,"syntology":{"ran":0,"of":4,"unverified":4,"pointer_only":0}},{"paper":"/paper/stoi-net-a-deep-learning-based-non-intrusive","title":"STOI-Net: A Deep Learning based Non-Intrusive Speech Intelligibility Assessment Model","date":"2020-11-09","arxiv_id":"2011.04292","n_code_links":1,"syntology":null},{"paper":"/paper/effective-approaches-to-attention-based","title":"Effective Approaches to Attention-based Neural Machine Translation","date":"2015-08-17","arxiv_id":"1508.04025","n_code_links":44,"syntology":{"ran":2,"of":8,"unverified":6,"pointer_only":2}}],"papers_shown":6,"tasks":[{"task":"/task/graph-attention","name":"Graph Attention","papers":1},{"task":"/task/image-guided-story-ending-generation","name":"Image-guided Story Ending Generation","papers":1},{"task":"/task/inductive-bias","name":"Inductive Bias","papers":1},{"task":"/task/link-prediction","name":"Link Prediction","papers":1},{"task":"/task/machine-translation","name":"Machine Translation","papers":1},{"task":"/task/cg","name":"NER","papers":1},{"task":"/task/nmt","name":"NMT","papers":1},{"task":"/task/named-entity-recognition-1","name":"Named Entity Recognition","papers":1},{"task":"/task/named-entity-recognition-ner","name":"Named Entity Recognition (NER)","papers":1},{"task":"/task/node-classification","name":"Node Classification","papers":1},{"task":"/task/prediction","name":"Prediction","papers":1},{"task":null,"name":"Relation","papers":1},{"task":"/task/sentence","name":"Sentence","papers":1},{"task":"/task/speech-enhancement","name":"Speech Enhancement","papers":1},{"task":"/task/translation","name":"Translation","papers":1},{"task":"/task/named-entity-recognition","name":"named-entity-recognition","papers":1}],"tasks_shown":16,"n_tasks":16,"usage_by_year":[{"year":"2015","papers":1},{"year":"2020","papers":1},{"year":"2021","papers":1},{"year":"2022","papers":1},{"year":"2024","papers":2}],"row_source":"embedded","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/multiplicative-attention"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}