Research On Cell Migration Process Based On Deep Reinforcement Learning And The Algorithm Improvement

In recent years,machine learning has become more and more integrated with other fields,such as travel and entertainment.At the same time,the application of machine learning to the biological field has gradually attracted the attention of researchers.Previously,articles have combined the research of cells,tissues and other organisms with machine learning to identify the characteristics of cells,but these models are more comparable to other fields.Backward and rough,and these studies are based on supervised learning to identify cell characteristics,and few scholars use machine learning to study cell migration patterns.In the early stages of embryonic development,cells are affected by a highly complex multi-scale regulatory mechanism,which is a long-standing biological problem for the study of regulatory mechanisms.In this project,we use the trajectory data of cells to study the cell migration process based on the deep reinforcement learning algorithm.First of all,this topic starts with the study of individual cell migration methods.Based on the deep reinforcement learning algorithm,it is verified that individual cells migrate in an active way,and also provides a way to find the leader cells.Secondly,this topic is based on deep reinforcement learning to study the game mode in the interaction process of group cells.At the same time,in order to solve the problem that the traditional multi-agent can't learn the optimal strategy under certain circumstances,this paper proposes a deep deterministic strategy gradient algorithm(MAD3PG)for multi-agent competition network,which adds competitive network structure to multi-intelligence.In the depth deepening learning,the agent can learn the global optimal strategy.Finally,based on the MAD3 PG algorithm,different reward functions are set up for different games to train,so as to verify the relationship between competition and cooperation between cells.With the help of reverse engineering thinking,this topic verifies the intrinsic regulation mechanism of cell migration process based on the motion data of cells recorded in real-time imaging technology,which provides a new perspective for the study of biological cell problems based on machine learning.