eduzhai > Applied Sciences > Engineering >

Text-based classification of interviews for mental health -- juxtaposing the state of the art

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 33 pages

Abstract: Currently, the state of the art for classification of psychiatric illness isbased on audio-based classification. This thesis aims to design and evaluate astate of the art text classification network on this challenge. The hypothesisis that a well designed text-based approach poses a strong competition againstthe state-of-the-art audio based approaches. Dutch natural language models arebeing limited by the scarcity of pre-trained monolingual NLP models, as aresult Dutch natural language models have a low capture of long range semanticdependencies over sentences. For this issue, this thesis presents belabBERT, anew Dutch language model extending the RoBERTa[15] architecture. belabBERT istrained on a large Dutch corpus (+32GB) of web crawled texts. After this thesisevaluates the strength of text-based classification, a brief exploration isdone, extending the framework to a hybrid text- and audio-based classification.The goal of this hybrid framework is to show the principle of hybridisationwith a very basic audio-classification network. The overall goal is to createthe foundations for a hybrid psychiatric illness classification, by provingthat the new text-based classification is already a strong stand-alonesolution.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×