Skin lesion classification with crowdsourced similarity ratings
Machine learning algorithms for skin lesion classification typically learn from images which have been labeled as malignant (for example, melanoma) or not. Such tasks can still suffer from overfitting due to limited dataset size. In other computer vision tasks, crowdsourcing labels has been effective, but the average person typically does not have the background to classify skin lesions. However, assessing the similarity of images might still be possible, even without such background. The similarities can then be used together with ground truth labels to train a classification algorithm.
Multiple student projects are possible, focusing on:
- Researching how to best collect similarity ratings (some budget is available for Amazon Mechanical Turk)
- Researchung how to integrate similarity ratings in existing skin lesion classification algorithms