We explore the use of the Focal Tversky Loss function
A data-driven method to reduce training sample size to mitigate careful model tuning required when using transfer learning for Alzheimer’s disease classification.
In IEEE Access, 2019.