On Using Fuzzy Reinforcement Learning to Control the Cancer Cells

Today to prevent the growth of cancer cells, the use of synthetic drugs and radiation therapyprocedures have been used increasingly as a significant barrier for accelerating the growth ofthese cells. Recently, extensive research has been done on various cancer control methods.In these studies, classical and modern controllers are applied on mathematical models thatshow the growth of cancer cells. In the present simulations on these models, they can also beseen successful results. However, modeling the growth of cancer cells is a complex issueand therefore certainly a complete and perfect model cannot be introduced to simulate the allof cancer cell aspects. In this paper we introduce reinforcement learning, as a method whichis not based on any particular model to examine the treatment of cancers. Afterwards, Bloodcancer control by adjusting the amount of chemical drugs and also breast cancer treatmentby controlling the flux of laser device is proposed. Reinforcement learning is an approach inwhich the agent or agents with respect to the series of positive or negative rewards do asuboptimal action. This technique proves its usefulness as a system which its modelnormally is not available. In this way it can be considered as a bypass method for othercontrolled methods. One disadvantage of this approach is to use discrete actions. This isdespite the fact that many continuous systems will not be successful using this proceeding.To solve this problem, different approaches, such as fuzzy logic has been proposed forcontinuous operations. In this case, the rules set in fuzzy controller are created byreinforcement learning as an expert machine and therefore a continuous operation can beproduced.

کلیدواژه ها:

Reinforcement Learning ، Fuzzy Logic ، Cancer Cells Control ، Hyperthermia ، Cancer Cell Model