FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3198850411> ?p ?o ?g. }

• W3198850411 abstract "Multimodal sentiment analysis is a trending area of research, and multimodal fusion is one of its most active topic. Acknowledging humans communicate through a variety of channels (i.e visual, acoustic, linguistic), multimodal systems aim at integrating different unimodal representations into a synthetic one. So far, a consequent effort has been made on developing complex architectures allowing the fusion of these modalities. However, such systems are mainly trained by minimising simple losses such as L1 or cross-entropy. In this work, we investigate unexplored penalties and propose a set of new objectives that measure the dependency between modalities. We demonstrate that our new penalties lead to a consistent improvement (up to 4.3 on accuracy) across a large variety of state-of-the-art models on two well-known sentiment analysis datasets: CMU-MOSI and CMU-MOSEI. Our method not only achieves a new SOTA on both datasets but also produces representations that are more robust to modality drops. Finally, a by-product of our methods includes a statistical network which can be used to interpret the high dimensional representations learnt by the model." @default.
• W3198850411 created "2021-09-13" @default.
• W3198850411 creator A5027506291 @default.
• W3198850411 creator A5044425827 @default.
• W3198850411 creator A5055866824 @default.
• W3198850411 creator A5064876801 @default.
• W3198850411 date "2021-01-01" @default.
• W3198850411 modified "2023-10-09" @default.
• W3198850411 title "Improving Multimodal fusion via Mutual Dependency Maximisation" @default.
• W3198850411 cites W1520494951 @default.
• W3198850411 cites W1522301498 @default.
• W3198850411 cites W1670132599 @default.
• W3198850411 cites W1921523184 @default.
• W3198850411 cites W1940872118 @default.
• W3198850411 cites W1988846555 @default.
• W3198850411 cites W2029996593 @default.
• W3198850411 cites W2030695833 @default.
• W3198850411 cites W2066477856 @default.
• W3198850411 cites W2075398069 @default.
• W3198850411 cites W2079725295 @default.
• W3198850411 cites W2095176743 @default.
• W3198850411 cites W2095439994 @default.
• W3198850411 cites W2095705004 @default.
• W3198850411 cites W2122925692 @default.
• W3198850411 cites W2146334809 @default.
• W3198850411 cites W2149350210 @default.
• W3198850411 cites W2165644552 @default.
• W3198850411 cites W2166944917 @default.
• W3198850411 cites W2250539671 @default.
• W3198850411 cites W2329925520 @default.
• W3198850411 cites W2518518020 @default.
• W3198850411 cites W2594683362 @default.
• W3198850411 cites W2619383789 @default.
• W3198850411 cites W2624871570 @default.
• W3198850411 cites W2738581557 @default.
• W3198850411 cites W2753840835 @default.
• W3198850411 cites W2783047733 @default.
• W3198850411 cites W2787581402 @default.
• W3198850411 cites W2792643794 @default.
• W3198850411 cites W2798965674 @default.
• W3198850411 cites W2842511635 @default.
• W3198850411 cites W2883409523 @default.
• W3198850411 cites W2900338016 @default.
• W3198850411 cites W2901272442 @default.
• W3198850411 cites W2917014163 @default.
• W3198850411 cites W2929134750 @default.
• W3198850411 cites W2949391930 @default.
• W3198850411 cites W2949517790 @default.
• W3198850411 cites W2950541952 @default.
• W3198850411 cites W2951873722 @default.
• W3198850411 cites W2962900302 @default.
• W3198850411 cites W2962931510 @default.
• W3198850411 cites W2963341956 @default.
• W3198850411 cites W2963710346 @default.
• W3198850411 cites W2964051877 @default.
• W3198850411 cites W2970518055 @default.
• W3198850411 cites W2970597249 @default.
• W3198850411 cites W2980313216 @default.
• W3198850411 cites W2980360762 @default.
• W3198850411 cites W2990629080 @default.
• W3198850411 cites W3034266838 @default.
• W3198850411 cites W3037428746 @default.
• W3198850411 cites W3037572520 @default.
• W3198850411 cites W3039256091 @default.
• W3198850411 cites W3044911731 @default.
• W3198850411 cites W3049723069 @default.
• W3198850411 cites W3100544011 @default.
• W3198850411 cites W3100848296 @default.
• W3198850411 cites W3103181422 @default.
• W3198850411 cites W3158935470 @default.
• W3198850411 cites W3174873147 @default.
• W3198850411 cites W3198254360 @default.
• W3198850411 cites W3198402388 @default.
• W3198850411 cites W1857789879 @default.
• W3198850411 cites W2970603031 @default.
• W3198850411 doi "https://doi.org/10.18653/v1/2021.emnlp-main.21" @default.
• W3198850411 hasPublicationYear "2021" @default.
• W3198850411 type Work @default.
• W3198850411 sameAs 3198850411 @default.
• W3198850411 citedByCount "10" @default.
• W3198850411 countsByYear W31988504112021 @default.
• W3198850411 countsByYear W31988504112022 @default.
• W3198850411 countsByYear W31988504112023 @default.
• W3198850411 crossrefType "proceedings-article" @default.
• W3198850411 hasAuthorship W3198850411A5027506291 @default.
• W3198850411 hasAuthorship W3198850411A5044425827 @default.
• W3198850411 hasAuthorship W3198850411A5055866824 @default.
• W3198850411 hasAuthorship W3198850411A5064876801 @default.
• W3198850411 hasBestOaLocation W31988504111 @default.
• W3198850411 hasConcept C106301342 @default.
• W3198850411 hasConcept C119857082 @default.
• W3198850411 hasConcept C121332964 @default.
• W3198850411 hasConcept C136197465 @default.
• W3198850411 hasConcept C144024400 @default.
• W3198850411 hasConcept C154945302 @default.
• W3198850411 hasConcept C19768560 @default.
• W3198850411 hasConcept C204321447 @default.
• W3198850411 hasConcept C2779903281 @default.
• W3198850411 hasConcept C2780226545 @default.
• W3198850411 hasConcept C36289849 @default.

• next