Neural Network I / O

Question

Can someone explain to me how to make more complex datasets like team statistics, weather, bones, complex number types.

I understand all the math and how it works, I just don’t know how to enter more complex data, and then how to read the data that it spits out.

if someone can provide examples in python that would be a big help

Answer 1

You must encode your input and output to what can be represented by units of a neural network. (for example, 1 for "x has a specific property p" -1 for "x does not have a property p" if the range of your units is in [-1, 1])

, .

, " " ( , boltzman, ).

Answer 2

, . - , ( )

---

, , .

• : [ , , , ...] ​​
• : [ave height, ave weight, max height, top score,...]
• Dice: , , , ? *
• : [a, ai, b, bi,...]

* , (0.0,1.0). , , .

---

, , , . , .

• : [ , temp,...] **
• : [ , 20,...]
• : [x, xi,...]

** : 1.0, , 0.0,

---

, , , - .

Answer 3

.

"" , ( ..) , , , , 1 0 ... , , , .

Answer 4

, . n, n . . n .

, , , m m . . , , , , .

Another aspect of coding is that you have to estimate the number of input and hidden units depending on the amount of data. Too many units associated with the data can give a poor approximation due to a dimensional problem. In some cases, you can combine some of the input in some way to avoid this problem or use some kind of reduction mechanism like PCA.