Any suggestions would be appreciated.
Eby
The goal is to predict whether a 28x28 picture is A, B, C, D, E, F, G,
H, I or J.
You can use the dataset to practice:
1) Look at learning curses. See how the number of examples improves
accuracy.
2) Recognize high bias or high variance. And lower them.
3) Get the best possible accuracy on a validation set. Use anything
you want to get there.
Can you get above 80%? I can help you, if you get stuck.
Note that the labels are 0 to 9 here. Add 1 to them, if you want to
have y in range 1 to 10.
Looks interesting. Thank you Ivo.