(Wednesday, 8/1/2018) 10:15 AM - 12:15 PM
Room: Davidson Ballroom B
The use of machine learning algorithms in medical research has become increasingly popular over the past decade. As these approaches are closely derived from statistical learning principles, the application of these techniques require additional understanding and preparation of the data prior to training a model. In the first part of this session we will cover data curation/preparation, provide insights on splitting datasets into training, validation, and test sets for parameter optimization, as well as cover additional principles including underfitting/overfitting. In addition, we will introduce and provide some mathematical details to popular machine learning algorithms used in radiomics research.
In the second part of this session, we will be transitioning from traditional machine learning approaches towards deep learning. We will start by introducing some intuition behind deep learning and its perceived ability to approximate very complex non-linear relationships in data. We will cover the building blocks of neural networks, discuss how “learning� works, introduce key concepts for hyper-parameter optimization, and outline some regularization approaches for improving prediction performance.
Learning Objectives:
1. To provide some insight on data understanding and preparation prior to using machine learning algorithms
2. To introduce some commonly used machine learning approaches used in radiomics research
3. To provide some intuition as to how deep learning works
4. To introduce key concepts used in deep learning
Handouts
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
