How can I change the single precision to double precision for Deep learning .

I'm using the deep learning toolbox offering Matlab function. The target variable I want to learn has very precise accuracy. So, I want to change the deep learning toolbox that is currently available in single precision to double precision, but I don't know how. In Python, if you use a function such as float64, it changes immediately, but how can I change it in MATLAB?

Answer 1

[KSSV]

Read about double.

[Mark Loui]

I have read and dont really understand it is there any other methods?

[KSSV]

x = single(pi) 

x = single 3.1416

class(x)

ans = 'single'

y = double(x)

y = 3.1416

class(y)

ans = 'double'

[DAYO YUN]

The deep learning toolbox provided by MATLAB basically performs training with single precision.

Converting a number to double precision in this way is not expected to work with double precision in the built-in function.

[KSSV]

Of course yes... There is no double precision support according to my knowledge..