{"url":"/sota/multimodal-sentiment-analysis-on-cmu-mosei-1","task":{"name":"Multimodal Sentiment Analysis","url":"/task/multimodal-sentiment-analysis","note":null},"dataset":{"name":"CMU-MOSEI","url":"/dataset/cmu-mosei"},"category":"Computer Vision","categories":["Computer Vision","Natural Language Processing"],"category_note":null,"description":"Multimodal sentiment analysis is the task of performing sentiment analysis with multiple data sources - e.g. a camera feed of someone's face and their recorded speech.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [ICON: Interactive Conversational Memory Network\r\nfor Multimodal Emotion Detection](https://www.aclweb.org/anthology/D18-1280.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Accuracy","MAE","F1","Acc-5","Acc-7","Corr"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Accuracy":"higher","MAE":"lower","F1":"higher","Acc-5":"higher","Acc-7":"higher","Corr":null}},"counts":{"rows":15,"rows_with_code":11,"rows_with_paper_page":15,"rows_dated":15,"rows_using_additional_data":8},"rows":[{"rank_in_archive_order":1,"model":"SeMUL-PCD","metrics":{"Accuracy":"88.62","F1":"89.04"},"uses_additional_data":false,"paper_date":"2023-10-27","paper":"/paper/multi-label-emotion-analysis-in-conversation","paper_url":"https://dl.acm.org/doi/10.1145/3581783.3612517","paper_title":"Multi-label Emotion Analysis in Conversation via Multimodal Knowledge Distillation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":2,"model":"MMML","metrics":{"Acc-5":"57.45","Acc-7":"54.77","Accuracy":"88.22","Corr":"81.42","F1":"88.04","MAE":"0.5072"},"uses_additional_data":false,"paper_date":"2023-08-01","paper":"/paper/multi-modality-multi-loss-fusion-network","paper_url":"https://arxiv.org/abs/2308.00264v4","paper_title":"Multimodal Multi-loss Fusion Network for Sentiment Analysis","code":"https://github.com/zehuiwu/MMML","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"UniMSE","metrics":{"Accuracy":"87.50","F1":"87.46","MAE":"0.523"},"uses_additional_data":true,"paper_date":"2022-11-21","paper":"/paper/unimse-towards-unified-multimodal-sentiment","paper_url":"https://arxiv.org/abs/2211.11256v1","paper_title":"UniMSE: Towards Unified Multimodal Sentiment Analysis and Emotion Recognition","code":"https://github.com/lemei/unimse","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"SPECTRA","metrics":{"Accuracy":"87.34"},"uses_additional_data":false,"paper_date":"2023-05-19","paper":"/paper/speech-text-dialog-pre-training-for-spoken","paper_url":"https://arxiv.org/abs/2305.11579v2","paper_title":"Speech-Text Dialog Pre-training for Spoken Dialog Understanding with Explicit Cross-Modal Alignment","code":"https://github.com/alibabaresearch/damo-convai","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"Transformer-based joint-encoding","metrics":{"Accuracy":"82.48"},"uses_additional_data":true,"paper_date":"2020-06-29","paper":"/paper/a-transformer-based-joint-encoding-for-1","paper_url":"https://arxiv.org/abs/2006.15955v1","paper_title":"A Transformer-based joint-encoding for Emotion Recognition and Sentiment Analysis","code":"https://github.com/jbdel/MOSEI_UMONS","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":3,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"Modulated-fusion transformer","metrics":{"Accuracy":"82.45"},"uses_additional_data":true,"paper_date":"2020-10-05","paper":"/paper/modulated-fusion-using-transformer-for","paper_url":"https://arxiv.org/abs/2010.02057v1","paper_title":"Modulated Fusion using Transformer for Linguistic-Acoustic Emotion Recognition","code":"https://github.com/jbdel/modulated_fusion_transformer","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"MMLatch","metrics":{"Accuracy":"82.4","MAE":"0.7"},"uses_additional_data":true,"paper_date":"2022-01-24","paper":"/paper/mmlatch-bottom-up-top-down-fusion-for","paper_url":"https://arxiv.org/abs/2201.09828v1","paper_title":"MMLatch: Bottom-up Top-down Fusion for Multimodal Sentiment Analysis","code":"https://github.com/georgepar/mmlatch","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"Multilogue-Net","metrics":{"Accuracy":"82.10","MAE":"0.59"},"uses_additional_data":true,"paper_date":"2020-02-19","paper":"/paper/multilogue-net-a-context-aware-rnn-for-multi","paper_url":"https://arxiv.org/abs/2002.08267v3","paper_title":"Multilogue-Net: A Context Aware RNN for Multi-modal Emotion Detection and Sentiment Analysis in Conversation","code":"https://github.com/amanshenoy/multilogue-net","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"Proposed: B2 + B4 w/ multimodal fusion","metrics":{"Accuracy":"81.14"},"uses_additional_data":true,"paper_date":"2020-02-21","paper":"/paper/gated-mechanism-for-attention-based","paper_url":"https://arxiv.org/abs/2003.01043v1","paper_title":"Gated Mechanism for Attention Based Multimodal Sentiment Analysis","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"CAE-LR","metrics":{"Accuracy":"78"},"uses_additional_data":true,"paper_date":"2021-10-06","paper":"/paper/unsupervised-multimodal-language","paper_url":"https://arxiv.org/abs/2110.03007v2","paper_title":"Unsupervised Multimodal Language Representations using Convolutional Autoencoders","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":11,"model":"Graph-MFN","metrics":{"Accuracy":"76.9","MAE":"0.71"},"uses_additional_data":true,"paper_date":"2018-07-01","paper":"/paper/multimodal-language-analysis-in-the-wild-cmu","paper_url":"https://aclanthology.org/P18-1208","paper_title":"Multimodal Language Analysis in the Wild: CMU-MOSEI Dataset and Interpretable Dynamic Fusion Graph","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"MARLIN (ViT-L)","metrics":{"Accuracy":"74.83"},"uses_additional_data":false,"paper_date":"2022-11-12","paper":"/paper/marlin-masked-autoencoder-for-facial-video","paper_url":"https://arxiv.org/abs/2211.06627v3","paper_title":"MARLIN: Masked Autoencoder for facial video Representation LearnINg","code":"https://github.com/ControlNet/MARLIN","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"MARLIN (ViT-B)","metrics":{"Accuracy":"73.7"},"uses_additional_data":false,"paper_date":"2022-11-12","paper":"/paper/marlin-masked-autoencoder-for-facial-video","paper_url":"https://arxiv.org/abs/2211.06627v3","paper_title":"MARLIN: Masked Autoencoder for facial video Representation LearnINg","code":"https://github.com/ControlNet/MARLIN","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"MARLIN (ViT-S)","metrics":{"Accuracy":"72.69"},"uses_additional_data":false,"paper_date":"2022-11-12","paper":"/paper/marlin-masked-autoencoder-for-facial-video","paper_url":"https://arxiv.org/abs/2211.06627v3","paper_title":"MARLIN: Masked Autoencoder for facial video Representation LearnINg","code":"https://github.com/ControlNet/MARLIN","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"ALMT","metrics":{"Acc-5":"55.96","Acc-7":"54.28","Corr":"0.779","MAE":"0.526"},"uses_additional_data":false,"paper_date":"2023-10-09","paper":"/paper/learning-language-guided-adaptive-hyper","paper_url":"https://arxiv.org/abs/2310.05804v2","paper_title":"Learning Language-guided Adaptive Hyper-modality Representation for Multimodal Sentiment Analysis","code":"https://github.com/Haoyu-ha/ALMT","n_code_links":1,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,795 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6795,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2785},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":2,"rows_with_any_sample_ran":2,"distinct_papers_with_graph_line":2,"distinct_papers_with_any_sample_ran":2,"samples_over_distinct_papers":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":0,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":0,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}