log_bit_rate = np.log(VIDEO_BIT_RATE[bit_rate] / float(VIDEO_BIT_RATE[-1])) # log_last_bit_rate = np.log(VIDEO_BIT_RATE[last_bit_rate] / float(VIDEO_BIT_RATE[-1])) # reward = log_bit_rate \ # - REBUF_PENALTY * rebuf \ # - SMOOTH_PENALTY * np.abs(log_bit_rate - log_last_bit_rate)帮我将这段奖励代码改写成相应对数尺度的QoE
时间: 2023-06-26 19:09:28 浏览: 199
魔改卡尔曼扩展滤波器 robot_pose_ekf,topic名称对上就可以使用,滤波器中不需要改任何代码
log_bit_rate_QoE = np.log2(VIDEO_BIT_RATE[bit_rate] / VIDEO_BIT_RATE[-1])
log_last_bit_rate_QoE = np.log2(VIDEO_BIT_RATE[last_bit_rate] / VIDEO_BIT_RATE[-1])
qoe_reward = log_bit_rate_QoE \
- REBUF_PENALTY * rebuf \
- SMOOTH_PENALTY * np.abs(log_bit_rate_QoE - log_last_bit_rate_QoE)
阅读全文