Forschungspraxis/Master’s thesis - Unsupervised Learning and Human-Assisted Labeling


Topic

Latest advancements in artificial intelligence show great promise for medical imaging. Automated screening and diagnosis tools based on machine learning techniques could help make medical tests faster, cheaper, and more readily available to patients. Large medical imaging datasets and successful machine learning architectures from the field of computer vision already exist. However, a major obstacle to actually training models on these datasets is the lack of available labels. Due to the domain expertise needed to assess medical images, obtaining labels is costly and time-intensive. Furthermore, labels may not be entirely reliable. Experts may disagree with each other, and even individual labelers’ judgments of the same image may vary at different points in time. The goal of this Forschungspraxis or Master’s thesis is to design, implement, and evaluate a prototype application that makes use of unsupervised learning to efficiently collect consistent labels from medical domain experts.

The student work will be carried out in conjunction with the CellFace project, which investigates the potential of computer vision and machine learning techniques for the task of automated blood cell diagnosis. You can read more about CellFace here.

Requirements

  • Python coding skills (Pytorch, Django)
  • Some prior knowledge of machine learning techniques
  • Experiences with Git
  • Knowledge of JavaScript (VueJS)
  • Capacity for teamwork

For a more in-depth introduction to the relevant research topics, see:

Kulesza, Todd, Saleema Amershi, Rich Caruana, Danyel Fisher, and Denis Charles. 2014. “Structured Labeling for Facilitating Concept Evolution in Machine Learning.” In Proceedings of the 32nd Annual ACM Conference on Human Factors in Computing Systems - CHI ’14, 3075–84. Toronto, Ontario, Canada: ACM Press. https://doi.org/10.1145/2556288.2557238.

Wigness, Maggie, Bruce A. Draper, and J. Ross Beveridge. 2015. “Efficient Label Collection for Unlabeled Image Datasets.” In 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 4594–4602. Boston, MA, USA: IEEE. https://doi.org/10.1109/CVPR.2015.7299090.

Supervisors

Alice Hein, M.Sc. and Stefan Röhrl, M.Sc. 

Chair for Data Processing

Contact Information

Alice Hein, M.Sc.

Chair for Data Processing

TUM Department of Electrical and Computer Engineering

Technical University of Munich

 

Arcisstr. 21, 80333 Munich

Room Z942

alice.hein@tum.de