Optimization Design of Concurrency Control Algorithm for Main Memory Database Based on Machine Learning

Effective real-time network measurement provides a basis for generating routing optimization strategies, so that the network can sense congestion. The existing out of band network telemetry technology will transmit additional probes to measure the network status, which will inevitably lead to the "observer" effect, leading to inaccurate measurement information. In addition, it is difficult to describe the relationship between complex network states and routing optimization strategies with accurate mathematical models. Let some neurons stop working temporarily, reduce the interaction between neurons, implicitly delete neurons in the network, and prevent some functional synergies to alleviate excessive matching. The algorithm selects temporarily discarded neurons with random probability, while the optimization algorithm uses Ising model in machine learning data to identify neurons with low link energy, and temporarily discards these neurons in training and reasoning. This algorithm makes the model more general and effectively alleviates the over fitting problem of network training. Its purpose is to select key features closely related to the learning purpose or eliminate irrelevant features to improve the performance of the model. Feature sparse methods can be divided into two categories. And in machine learning, the use of control algorithms can effectively improve the development of machine learning, and in machine learning, the proportion of optimization using control algorithms is very large. This paper puts forward many new ideas, hoping to improve the use of machine learning control algorithm.