Logo des Repositoriums

Auflistung nach Schlagwort "active learning"

1 - 2 von 2
  • Konferenzbeitrag
    Active-learning-driven deep interactive segmentation for cost-effective labeling of crop-weed image data
    (43. GIL-Jahrestagung, Resiliente Agri-Food-Systeme, 2023) Sikouonmeu, Freddy; Atzmueller, Martin
    Active learning has shown its reliability in (semi-)supervised machine learning tasks to reduce the labeling cost for large datasets in various areas. However, in the agricultural field, despite past attempts to reduce the labeling cost and the burden on the labeler in acquiring image labels, the load during the acquisition of pixel-level labels for semantic image segmentation tasks remains high. Typically, the respective pixel-level masks are acquired manually by drawing polygons over irregular and complex-shaped object boundaries. In contrast, this paper proposes a method leveraging the power of a click-based deep interactive segmentation model (DISEG) in an active learning approach to harvest high-quality image segmentation labels at a low cost for training a real-time task model by only clicking on the objects’ fore- and background surfaces. Our first experimental results indicate that with an average of 3 clicks per image object and using only 3% of the unlabeled dataset, we can acquire pixel-level labels with good quality at low cost.
  • Textdokument
    A Crowdsourcing-based Learning Approach to activate Active Learning
    (Bildungsräume 2017, 2017) Koschmider, Agnes; Schaarschmidt, Mario
    Usually students consume learning material and write an exam at the end of the lecture. Such a process follows a summative learning pattern, which can be considered a standard approach at universities. Studies in educational theory indicate, however, that active involvement – instead of passive consumption – should be fostered in learning since active learning proved to be superior to passive learning. To benefit from active learning arrangements, we implemented an active involvement of students into the exam preparation for an introduction to Information Systems course at the University of Cologne. Students were asked to design exercises and provide solutions to selected topics. Subsequently, they received feedback to their submissions, which supports the self-assessment on the subject. An empirical evaluation shows general agreement for such active involvement of students and also indicates that students participating in the task creation are more likely to pass an exam than students denying the participation. This paper presents our crowdsourcing-based learning approach and discusses challenges for its implementation.