dnn_model_2 = DNN(
input_size=hidden_size,
output_size=user_list[0].n_labels,
architecture={'h1': 128, 'h2': 128},
learning_rate=0.001
)
train_dnn(dnn_model_2, user_list, ae_list, user_id=1, batch_size=128, epochs=1000, display_step=10)
'''
dnn_model = DNN(input_size=hidden_size,
output_size=user_list[0].n_labels,
architecture={
'h1': 128,
'h2': 128
},
learning_rate=0.001)
train_dnn(dnn_model,
user_list,
ae_list,
batch_size=128,
epochs=10000,
display_step=100)
batch_xs, batch_ys = user_list[0].next_batch(user_list[0].n_samples_test,
is_train=False)
for i in range(n_users):
acc_test = cal_acc(dnn_model.predict(ae_list[i].transform(batch_xs)),
batch_ys)
print(i, acc_test)
