Abstract
Deep learning (DL) is a family of machine learning methods capable of detecting multiple levels of latent representations from the data. This is achieved by combining consecutive layers of simple nonlinear transformations that allow the extraction of increasingly abstract features. DL has become one of the most popular and promising approaches in machine learning. In this chapter, we introduce the fundamentals of DL by focusing on the architecture and training process of a typical model: the deep neural network (DNN). We follow this by discussing the main challenges in its application to brain disorders. Finally, we review some exemplary implementations of DNNs to diagnostic and prognostic studies in clinical neuroimaging. Overall, although there are critical challenges to overcome, the use of DL has shown encouraging results. With the increasing availability of larger sample sizes, its application to brain disorders is likely to continue to expand in the next years.
Original language | English |
---|---|
Title of host publication | Machine Learning |
Subtitle of host publication | Methods and Applications to Brain Disorders |
Publisher | Elsevier |
Pages | 157-172 |
Number of pages | 16 |
ISBN (Electronic) | 9780128157398 |
ISBN (Print) | 9780128157398 |
DOIs | |
Publication status | Published - 1 Jan 2019 |
Keywords
- Alzheimer’s disease
- Autism spectrum disorder
- Brain disorders
- Deep learning
- Deep neural network
- Machine learning
- Neuroimaging
- Representation learning
- Schizophrenia