Action value Actor-Critic型Policy Gradientによる連続値動作の強化学習 - Qiita# -*- coding: utf-8 -*- import wx import wx.lib import wx.lib.plot as plot import math import random as rnd import numpy as np import copy # import pickle # Steps looking back STATE_NUM = 2 # State STATE_NUM = 2 NUM_EYES = 9 STATE_DIM = NUM_EYES * 3 * 2 class SState(object): def __init__(self): self.seq = np.ones((STATE_NUM, NUM_EYES*3), dtype=np.float32) def push_s(self, state): self.seq[1:STATE_
