{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/learn-to-combine-modalities-in-multimodal","title":"Learn to Combine Modalities in Multimodal Deep Learning","arxiv_id":"1805.11730","date":"2018-05-29","proceeding":null,"authors":["Kuan Liu","Yanen Li","Ning Xu","Prem Natarajan"],"abstract":"Combining complementary information from multiple modalities is intuitively\nappealing for improving the performance of learning-based approaches. However,\nit is challenging to fully leverage different modalities due to practical\nchallenges such as varying levels of noise and conflicts between modalities.\nExisting methods do not adopt a joint approach to capturing synergies between\nthe modalities while simultaneously filtering noise and resolving conflicts on\na per sample basis. In this work we propose a novel deep neural network based\ntechnique that multiplicatively combines information from different source\nmodalities. Thus the model training process automatically focuses on\ninformation from more reliable modalities while reducing emphasis on the less\nreliable modalities. Furthermore, we propose an extension that multiplicatively\ncombines not only the single-source modalities, but a set of mixtured source\nmodalities to better capture cross-modal signal correlations. We demonstrate\nthe effectiveness of our proposed technique by presenting empirical results on\nthree multimodal classification tasks from different domains. The results show\nconsistent accuracy improvements on all three tasks.","url_abs":"http://arxiv.org/abs/1805.11730v1","url_pdf":"http://arxiv.org/pdf/1805.11730v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"learn-to-combine-modalities-in-multimodal","repo_url":"https://github.com/skywaLKer518/MultiplicativeMultimodal","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"multimodal-deep-learning","task_name":"Multimodal Deep Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.11730","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}