Development
Author: Nicolás Agustín Delmagro | Email: agustin.delamgro@gmail.com
Nicolás Agustín Delmagro1°, Bruno Bianchi1°2°, Juan Kamienkowski1°2°3°
1° Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Departamento de Computación. Buenos Aires, Argentina.
2° CONICET-Universidad de Buenos Aires. Instituto de Ciencias de la Computación (ICC). Buenos Aires, Argentina.
3° Facultad de Ciencias Exactas y Naturales, Maestría en Explotación de Datos y Descubrimiento del Conocimiento, Universidad de Buenos Aires, Buenos Aires, Argentina
In recent years, many studies have emerged in which fMRI signals are analyzed while certain cognitive or emotional tasks are performed. Several works have been based on capturing these brain responses while the study subjects listened to narrative stories, which were used for the semantic mapping of different areas of the brain and even the generation of decoders that are able to semantically reconstruct these stories from brain signals. Following a previous work we will train a decoder that has been developed to rebuild perceived speech from those fMRI signals. Since we consider that those stories with similar themes should have certain similarities in the vocabulary used, we will seek to group these stories automatically by the main theme they talk about. Using these subsets as training, we will analyze the effects they have on the semantic reconstructions generated and on the brain areas activated during the training. We expect getting better scores rebuilding a story that matches the category of the training set. It will also be interesting to examine if the rebuilded stories preserve the category of the original story or if they match the categories used in the training set.