Model optimization method

Abstract

The invention provides a model optimization method capable of realizing cross-platform intelligent model deployment. The model optimization method executed on the first device can comprise the steps that a first training data set is used for training to obtain a first model; transforming the first training data set by using a transformation function to obtain a transformed data set, and generating a second training data set based on the first training data set and the transformed data set; training the first model by using the second training data set to obtain an adjusted first model; performing performance test on the first model and the adjusted first model by using the test data set to obtain a first performance index and a second performance index respectively, and calculating a performance index difference between the first performance index and the second performance index; generating an adjusted second training data set based on the performance index difference, wherein the adjusted second training data set is used for training the adjusted first model to obtain a second model; and dynamically quantifying the second model to obtain an optimization model which can be deployed on the second equipment.

Images