Idae по определению временного действия в обучении с подкреплением [закрыто] ⇐ Python

[Гость]

There are list of features and operations. I would like the program to do the reinforcement learning to predict what to write for the next line by observing the code had been typed.

Код: Выделить всё

features = [a,b,c,d] operations = [+,-,*,/,mean,mode,median,...) # observation space is the following code blocks #####observation#### cache_1 = a+b   cache_2 = b+c/a   ####end observation### cache_3 = cache_1 + cache_2 # action/prediction 

Lets say there are 2 types of output, features_selection and operators_selection. In feature selection, there ll be 4 outputs at the final layers : [a,b,c,d]. My question is how to make the tem input (cache_1 and cache_2 in above example) also be included on the final layer or any ideas for those extra input.

if we just limit to use 2 cache on the program , it ll be [a,b,c,d,cache_1,cache_2] but the cache ll have different meaning in different task and obseravtion.

Are there any good way to do so?


Источник: https://stackoverflow.com/questions/781 ... t-learning