One of big bottlenecks for Deep Learning algorithms is acquiring labelled data for training. Often data is acquired in large amounts and it takes a lot of time and money to get all the data labelled.
Snorkel is a weak supervision management system that allow for weak labelling of large datasets using “labelling functions”. The idea is that if using these weak labelling functions and treating the labels they produce as probabilistic we can use our data resources without having to do the expensive work of labelling.
So far Snorkel has been used in an industrial setting by Google for Natural Language Processing, but usage within Image processing has been limited. The topic of this thesis would be to investigate the usage of these weak supervision techniques for topics like Image Classification, Object Detection and Image Segmentation.
For more information, contact R&D manager Ola Friman (firstname.lastname@example.org) and apply below at "ANSÖKAN" before the 19th of October. Please notice: Application must include CV, Cover letter and LADOK (study results including courses)