ml.py 文件源码

def neg_log_likelihood(self, alphabetas):
        df = self.df

        alphas = alphabetas[0::2]
        betas = alphabetas[1::2]
        df = self.df[self.df['cue'].isin(self.cues)]
        actions, rewards = df['action'].values, df['reward'].values
        cues = df['cue'].values
        prob_log = 0
        Q = dict([[cue, np.zeros(self.n_actions)] for cue in self.cues])
        for action, reward, cue in zip(actions, rewards, cues):
            alpha = alphas[self.cues.index(cue)]
            beta = betas[self.cues.index(cue)]
            Q[cue][action] += alpha * (reward - Q[cue][action])
            prob_log += np.log(softmax(Q[cue], beta)[action])

        return -prob_log