Multi-task classification model training method, classification method, model and device
The invention discloses a training method of a multi-task classification model, a classification method, a model, a device, computing equipment and a computer readable storage medium, and aims to solve the problem of how to improve the convergence rate of the model. The training method comprises the...
Saved in:
Main Authors | , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
24.10.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses a training method of a multi-task classification model, a classification method, a model, a device, computing equipment and a computer readable storage medium, and aims to solve the problem of how to improve the convergence rate of the model. The training method comprises the following steps: training each to-be-trained unit to convergence by using a training sample set in sequence to obtain a trained feature extraction module and a classifier module of each intermediate state; and keeping the parameters of the trained feature extraction module unchanged, using the sample features extracted by the trained feature extraction module to train each classifier module in the intermediate state in sequence, and only updating the parameters of the currently trained classifier module in the intermediate state to obtain the trained classifier module. According to the method, the parameter adjustment of the feature extraction module of the next to-be-trained unit is only fine adjustment training, |
---|---|
Bibliography: | Application Number: CN202210368242 |