Awarded 4 million SEK for developing new Deep Learning methods
Johanna Björklund, researcher at the Department of Computing Science at Umeå University, has been granted 4 million SEK for her research on semantic parsing of multimodal data by the Swedish Research Council.
Text: Victoria Skeidsvoll
The project title “Semantic parsing of multimodal data” means that Johanna Björklund and her team develops methods for translating data into formats that are is easy to process with computers. The term 'semantic' means that the focus is on the substance of the data rather than its surface form, and 'multimodal' that the data consists of a combination of image, sound, text and video.
“Put together, we aim to develop algorithms to represent the meaning in composite media objects in a format that is suitable for automatic processing” Johanna Björklund explains.
She is delighted to have been awarded 4 million SEK from the Swedish Research Council.
“The project contains both practical experiments and theoretical analysis, so we will not have a dull moment”, Johanna Björklund says. “The conditions are perfect – we already have strong international collaborations underway and there is a broad interested in the topic.”
New deep learning methods
In research, the project will open up for deeper forms of machine learning in a wide range of areas related to media, including indexing, media search and automatic summation of different types of content.
“For the industry, traditional tasks can be automatized, and in the long run it may lead to completely new products and services” says Johanna Björklund.