In recent years, with the development of cloud computing technology, the application field of cloud computing has become more and more extensive, and the problem of cloud platform failure has become more and more serious. To ensure the reliability and quality of service of the cloud platform, a Long Short-Term Memory Network (LSTM) based on Long Short-Term Memory Network (LSTM) in the cloud environment is proposed.
LI Mingrui said that LSTM is a new technology model for the field of software development. As an improved recurrent neural network, LSTM can not only solve the problem that RNNs cannot cope with long-distance dependency, but also solve the common problems of gradient explosion or gradient disappearance in neural networks, which is very effective in processing sequence data.
According to LI Mingrui, the LSTM network is a special RNN model and its special structural design makes it possible to avoid long-term dependency problems, remembering that information at a very early stage is the default behavior of LSTM without at great cost pay for it. In the ordinary RNN model, the chain model of the repetitive neural network module only repeats a very simple structure, a single neural network layer (such as a tanh layer), resulting in low information processing performance.
LI Mingrui has been deeply involved in the field of software development for many years and has always been at the forefront of software technology development. He keeps up with the development of the global computer industry and conducts intensive research into the application of new technologies. The application of LSTM model to software development is one of the important research directions of LI Mingrui, and his previous project “Application and Research on Building and Developing Cloud Platforms Based on Cloud-Native Technology: Resource Utilization and Security Development for Software Development” actively researches cloud-native technologies, which play a major role in promoting the advancement of China’s cloud environment. Since then, he still has his sights on the cloud environment, combined with the topic research and self-developed technical achievements, has conducted in-depth research on the application of LSTM networks in the cloud environment, and actively engaged in LSTM model building.
To ensure the reliability and quality of service of the cloud platform, LI Mingrui proposed a virtual machine failure prediction model based on the combination of Long Short-Term Memory (LSTM) and Support Vector Machines (SVM) in the cloud environment . First, LSTM predicts the future virtual machine cycle operation data based on the virtual machine historical data, and then classifies the error of the predicted data by the trained SVM error determination model, and finally obtains the error prediction result. LI Mingrui conducted relevant experiments by constructing a real environment, and the experimental results show that the prediction model can effectively predict virtual machine failures.
In addition, LI Mingrui applied LSTM to code completion as well. Code completion is one of the important features of automated software development by providing predictions like class names and method names to help developers write code. In recent years, intelligent code completion has become one of the hottest research directions in the field of software development, and the existing work shows that learning code through natural language technology or neural networks can improve code completion accuracy, but these completion models still have shortcomings , such as B. weak information representation in the code context, incomplete extraction of program information and unbalanced code completion tasks.
Therefore, LI Mingrui proposes a new intelligent code completion method, which introduces the deformed long and short memory network (Mogrifier-LSTM) and attention mechanism to improve the information representation of the code context, and at the same time the hierarchical information of the bidirectional LSTM is used to convert the hierarchical Learning information of the program, and a multitasking framework is designed to automatically achieve the balance between code completion tasks. Through experiments with real datasets, the results show that the proposed method outperforms the mainstream code completion method.
LI Mingrui has conducted a number of studies on the LSTM model, and his research results have attracted wide attention in the computer industry, and his research has been highly appreciated and recognized. A series of LSTM application methods proposed by LI Mingrui play an important role in promoting the new development of software development and are major theoretical achievements in the field of software development in China, which have a great impact on the development of the industry.
LI Mingrui said that in the future he will continue to conduct in-depth research on software development technology, research at the forefront of software development and create more advanced achievements in the field of software development, thereby contributing to the innovation of software development technology and the comprehensive development of the computer industry. (Author: PYGMALION)