Ali Can Soylemezoglu
MIT EECS Mason Undergraduate Research and Innovation Scholar
Early Cancer Detection Through Deep Learning
2016–2017
Electrical Engineering and Computer Science
- Artificial Intelligence & Machine Learning
Caroline Uhler
Early Cancer Detection Through Deep Learning
Neural networks are the state of the art when it comes to image classification. We intend to exploit the success deep learning has had in image classification in an attempt to classify images of cells to detect the early onset of cancer. Currently features are extracted manually. We attempt to use deep learning to first automatize the feature extraction process and then classify cell images. Furthermore we will be investigating techniques such as Multiple-Instance Learning (MIL) in classifying these images. At a high level we will be interested in the following two end goals: a) binary classification of cells (i.e. cancerous or non-cancerous) and b) using manifold learning to learn how cancer progresses.
Having participated in several UROPs I feel prepared to commit to a year-long independent projectI am excited to be working on a project at the intersection of cancer research and ML. It will give me an opportunity to apply my knowledge of ML on a very important topic. I hope to learn more about cellular biology further my ML knowledge and gain experience in independent research.