- Dataset:
MNISTmini.mat
- Source: Prof. MACP's datasets on his faculty website
- These are grayscale images of 10x10 pixels of handwritten digits in 10 classes (digit-0 to digit-9). Note: use as training, validation and test sets images 1-1000, 1001-2000 and 2001-3000, respectively.
- Task: binary classification, for MNIST digits 4 and 7.
- Algorithms: logistic regression and random forests