Content deleted Content added
Line 27:
=== Decision-level Fusion ===
Decision-level fusion (sometimes known as late fusion), feeds data from each modality (text, audio, or visual) independently into its own classification algorithm, and obtains the final [[sentiment]] classification results by fusing each result into a single decision vector.<ref name="s3"></ref> One of the advantages of this fusion technique is that it eliminates the need to fuse heterogeneous data, and each [[modality (human–computer interaction)|modality]] can utilize its most appropriate classification algorithm.<ref name="s1"></ref>
=== Hybrid Fusion ===
| |||