Training method, device, medium and computing device for predictive model

Abstract

Embodiments of the present invention provide a training method, apparatus, medium and computing device for a prediction model. The method includes: training the prediction model based on the acquired training data with the goal of making the network weight of the prediction model converge; outputting the prediction distribution corresponding to the training data through the prediction model; The loss function calculates the prediction distribution, so that the network weights of the prediction model are converged again, and the calibrated prediction distribution is obtained as the target to train the prediction model. The above technology of the present invention can train the prediction model based on the training data, output the prediction distribution corresponding to the training data through the prediction model, and train the prediction model based on the maximum mean difference loss function and the prediction distribution, so that the network weight of the prediction model is Convergence again to obtain the calibrated prediction distribution, which improves the accuracy of the prediction results output by the prediction model based on the time series data.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of deep learning, and more specifically, embodiments of the present invention relate to a training method, device, medium, and computing device for a prediction model. Background technique [0002] This section is intended to provide a background or context for implementations of the invention that are recited in the claims. The descriptions herein are not admitted to be prior art by inclusion in this section. [0003] Forecasting is a common task in machine learning and is widely used in various scenarios, such as sales forecasting, stock analysis, traffic flow forecasting, weather forecasting, etc. based on time series data. With the rapid development of deep learning technology in recent years, deep models have made remarkable progress in prediction tasks. Currently, neural networks are usually trained based on heteroscedastic neural networks. However, it has been found in practic...

AI Technical Summary

Problems solved by technology

However, it has been found in practice that neural networks based on heteroscedasticity may produce unreliable forecast distributions and uncertainty estimates based on existing time series data, and the resulting probability distributions are often uncalibrated

It can be seen that the accuracy of the prediction results obtained by the existing neural network based on time series data is low

Images