Multimodal Fusion: A Review, Taxonomy, Open Challenges, Research Roadmap and Future Directions

• Published in: Neutrosophic sets and systems Vol. 45; pp. 96 - 120
• Main Authors: Wajid, Mohd Anas, Zafar, Aasim
• Format: Journal Article
• Language: English
• Published: Neutrosophic Sets and Systems 01.12.2021; University of New Mexico
• Subjects: fuzzy logic; imperfections; Machine learning; Machine vision; multimodal data; multimodal fusion; neutrosophic logic; Smarandache, Florentin
• ISSN: 2331-6055; 2331-608X
• DOI: 10.5281/zenodo.5485570

The present work collects a plethora of previous research work in the field of multimodal fusion which despite a lot of research could not handle the imperfections. These imperfections could be at any stage initiating from the imperfections in data and its sources to imperfections in fusion strategies. Further, the work explores various applications of Neutrosophy in the field of handling imperfections along with description of previous work in this regard. These applications include the one which addresses the notion of imperfection and uncertainty among multimodal data which is being collected for fusion. In this way, the present work tries to incorporate neutrosophic logic and its applications in the field of computer vision including multimodal data fusion and information systems. It is assumed that if the notion of uncertainty is included in multimodal research, the development of newer algorithms for solving the problems of imperfections in multimodal systems will provide impetus to the existing research in this field. Keywords: Multimodal Data; Multimodal Fusion; Imperfections; Fuzzy Logic; Neutrosophic Logic; Machine Learning.