Machine learning model selection method

Abstract

The invention discloses a machine learning model selection method, mainly comprising the steps of model setting, training testing, model evaluation, model selection, prediction reasoning and model monitoring. The adopted algorithm model is selected through the model selection strategy and the model evaluation result, and the model selection process is decomposed into multiple aspects such as resource consumption, performance and business risks, and necessary and key processes involved in machine learning model selection are covered more widely, and the machine learning model selection method can be suitable for selection of various types of machine learning algorithm models and is high in universality. Meanwhile, multiple dimensions such as resource consumption, performance and business risk are adopted as the basis of model selection, and the engineering cost and the business risk are introduced into the model selection process besides the conventional model performance, and the highengineering availability and the low application risk of the algorithm model are effectively ensured, and the practical value is high.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method for selecting a machine learning model. Background technique [0002] Machine learning is a widely influential technology. It uses intelligent algorithms and computer systems to automatically extract valuable information from data, significantly improving the system's decision-making efficiency, accuracy, and real-time performance. At present, machine learning technology is widely used in many aspects of social production and life, such as monitoring security, voice assistants, medical diagnosis, product quality inspection, etc., bringing changes in production and lifestyle. [0003] Machine learning includes three elements: data, algorithm and computing power. Machine learning has a wealth of algorithm models, such as logistic regression model (Logistics Regression), support vector machine model (SVM), neural network model (NeuralNetwork), decision tree (Decision Tree...

AI Technical Summary

Problems solved by technology

In practical applications, it is necessary to select an appropriate algorithm model. However, because different machine learning models may have similar functions or performances, it brings certain difficulties to the selection and application of machine learning algorithm models.

[0004] Model selection is usually based on the performance of the algorithm itself, such as recall rate (Recall), accuracy rate (Precision), etc. However, this method may ignore engineering and business constraints on the model, such as computing and storage resource costs, Business risk costs, etc., may lead to application defects in the selected model

Images