Trait analysis based on multimodal prediction and optimization of the output parameters: A survey

This paper provides a survey of apparent personality trait analysis based on multimodal prediction and optimization output parameters. analyzes, synthesizes, compares the development different methods Personality computing within Big Five model (OCEAN model) audio, visual textual inputs. The results detection are for methods, from extracted handwriting with an accuracy 62-83%, 56-62%, audio 70%, element images videos 87-91%, aggregation function min max 52 - 81%, mean median 85%, robustness Huber 72 77 %. Trait is problem numerous real-life applications. Accurate recognition various human traits tasks that have been studied very long time in field psychology lately represent important area study computer science computing. best were achieved convolutional neural networks. Multimodal will give better than single modality, functions robust can achieve models.