Agent_1
0101....01110
data_selected
0不选/1选择
n_samples * 1
State2
action_n_features
feature_selected
n_samples * n_feature_selected
Agent_2
action_list
Reward__n_features
Reward_1
State1
Reward_2
action_2
X_train
action_1
n_samples * n_features
Reward
Agent_n_features
GCN
ACC
下游模型
