Image big data-oriented class increment classification method, system and device and medium

Abstract

The invention discloses an image big data-oriented class increment classification method, system and device and a medium. The method comprises an initialization training stage and an increment learning stage. The initialization training stage comprises the following steps: constructing an initial data set of an image; and training an initial classification model according to the initial data set. The incremental learning stage comprises the following steps: constructing an incremental learning data set according to the initial data set and new data of the image; obtaining a new incremental learning model according to the initial classification model, and training the new incremental learning model according to an incremental learning data set and a distillation algorithm to obtain a model capable of identifying new and old categories, wherein the distillation algorithm enables the inter-class distance of the model to be enlarged and the intra-class distance to be reduced. The incremental learning model is updated through the distillation algorithm, the inter-class distance of the model is enlarged, the intra-class distance of the model is reduced, the new and old data recognition performance of the model can be improved under limited storage space and computing resources, and the method, system and device can be widely applied to the field of big data application.

Description

technical field [0001] The present invention relates to the field of big data application, in particular to a method, system, device and medium for class increment classification oriented to image big data. Background technique [0002] With the rapid development of emerging information technologies and application models such as Internet cloud computing, Internet of Things, and social networks, the types and scale of data in human society are growing at an unprecedented rate, pushing human society into the era of big data with information explosion. Accompanying it is the continuous improvement of computer computing power, which has promoted the research and development of machine learning, especially in the field of deep learning, which has set off a research boom and is widely used in various fields such as finance, intelligent manufacturing, and medical health. . "Big data + machine learning" technology has become a familiar artificial intelligence technology in today's...

AI Technical Summary

Problems solved by technology

[0004] However, based on batch learning, training data needs to be obtained at one time. For new data that is continuously generated, it is necessary to mix new and old data to retrain the model. This method has problems of space storage and computing resource consumption and time cost. With the continuous arrival of new data , the amount of data that needs to be saved continues to expand, and the huge training set will lead to a decrease in the training speed of the model, a sharp increase in computing costs, and even the data cannot be loaded into the memory at one time, so that complete batch learning cannot be performed.

If the model is trained directly on new data, the model will have the problem of catastrophic forgetting of old data

Images