Model training method and device based on multiple sample sets, equipment and medium

Abstract

The embodiment of the invention provides a model training method and device based on multiple sample sets, equipment and a medium, and relates to the technical field of computers. The method comprises the following steps: respectively combining general data in a general sample set and specific data in a specific sample set according to a preset first sample proportion and a preset second sample proportion to obtain a first training sample set and a second training sample set; performing multi-round general training on the first neural network based on the first training sample set to obtain a second neural network; in at least one round of universal training, the relative proportion of the universal data and the specific data in the first training sample set is adjusted; performing at least one round of specific scene training on the second neural network based on the specific sample set to obtain a third neural network; and performing at least one round of convergence training on the third neural network based on the second training sample set and the specific sample set to obtain a target neural network classification model. The method is applied to the device, the equipment and the storage medium, and the precision of the neural network classification model can be improved.

Description

technical field [0001] The present invention relates to the field of computer technology, and in particular, to a method, apparatus, device and medium for model training based on multiple sample sets. Background technique [0002] As artificial intelligence algorithms are more and more widely used in various production systems, the precision requirements for artificial intelligence are also getting higher and higher. Among them, neural network training, as the most representative algorithm in the concept of artificial intelligence, is widely used in neural network classification models related to face recognition and sorting. Taking face recognition as an example, in face recognition, with the With the continuous increase of face data sets, the model accuracy is also continuously improved. However, a face recognition model with sufficient generalization is difficult to achieve, so the accuracy in specific scenarios often shows insufficient recognition accuracy. At the same ...

AI Technical Summary

Problems solved by technology

However, it is difficult to achieve a sufficiently generalized face recognition model, so the accuracy in specific scenarios often shows insufficient recognition accuracy. At the same time, with the popularization of privacy protection awareness, the collection of face data is becoming more and more difficult. Existing transfer learning (fintune) is difficult to obtain a more accurate model

However, with a small number of sample sets for training, it is difficult to guarantee high precision in specified scenarios.

Images