# Cost-sensitive portfolio selection via deep reinforcement learning ###### tags: `paper` `Reinforcement Learning` `Finance` :::info Published in: IEEE Transactions on Knowledge and Data Engineering Yifan Zhang School of Software Engineering, South China University of Technology, Guangzhou, Guangdong China (e-mail: sezyifan@mail.scut.edu.cn) Peilin Zhao AI Lab, Tencent, 508929 Shenzhen, Guangdong China (e-mail: peilinzhao@hotmail.com) Bin Li School of Software Engineering, South China University of Technology, Guangzhou, Guangdong China (e-mail: qyw@scut.edu.cn) Qingyao Wu Department of Finance, Wuhan University, Wuhan, Hubei China (e-mail: binli.whu@whu.edu.cn) Junzhou Huang AI Lab, Tencent, Shenzhen, Guangdong China (e-mail: joehhuang@tencent.com) Mingkui Tan School of Software Engineering, South China University of Technology, 26467 Guangzhou, Guangdong China (e-mail: mingkuitan@scut.edu.cn) Date of Publication: 10 March 2020 dataset source : https://poloniex.com/ dataset api : https：//poloniex.com/support/api/ ::: :::warning 整份備忘: $r^c_t$ : 第t個period考慮cost 的reward $\hat{r}^c_t$ : 第t個period考慮cost 的reward 取log 預測時長==test data時長 ::: 7-2 Reinforcement Learning Algorithm Selection 為何使用DPG不使用AC(Actor-Critic)?